Search

A multi-sensor information fusion algorithm based on fault-tolerant Kalman filter is proposed in phase-sensitive optical time-domain reflectometer (Φ-OTDR) system, for achieving fading-free distributed vibration sensing. Firstly, a fault-tolerant dual-core complementary array model is designed. The Rayleigh scattering signal denoising, and vibration existence judgment of localization points are carried out to obtain the differentiated frequency demodulation results of the sensing points of the dual-core fiber array. Then a fault-tolerant control strategy is used to determine the sensor weight coefficients and vibration judgment coefficients during data fusion processing, and array data fusion is carried out based on time series data using Kalman filter to realize error value identification and filling. The advantage of this method is the combination of redundant data in a complementary way to improve the system stability. The frequency response ranges from 10 Hz to 2400 Hz and the localization accuracy is 98.33%. The influence of key parameters on the frequency demodulation performance of fault-tolerant Kalman filter is discussed, and a standard deviation of 14.6 Hz and an average error of 7.6 Hz are obtained. The demodulation frequency data matrix obtained by the classical demodulation method has a demodulation error probability of 89.18%, which proves the widespread existence of demodulation errors in vibration signals. The fusion error of demodulation frequency is reduced to 0.25 Hz, the frequency demodulation accuracy reaches 100%, and the demodulation error caused by interference attenuation can be completely eliminated. This system based on fault-tolerant Kalman filter has the characteristics of simple multiplexing structure, interference fading resistance and stable demodulation performance. • A sensor information fusion algorithm based on fault-tolerant Kalman filter is proposed to achieve fading-free sensing. [ABSTRACT FROM AUTHOR]

Acoustic signal processing technology is a new discipline developed in recent years, which has been widely applied in various fields such as communication, radar, sonar, and so on. Due to the wide variety of sound sources and the high susceptibility of signals to various interferences, it is necessary to perform noise reduction processing on them. Blind source separation technology is an emerging signal processing method that can simultaneously separate signals from multiple sources, thus having significant advantages in noise reduction. On the basis of summarizing traditional noise reduction algorithms, this article proposes an improved adaptive filtering algorithm combined with blind source separation technology, and conducts simulation experiments on both the traditional algorithm and the improved algorithm. [ABSTRACT FROM AUTHOR]

As a vital mechanical sub-component, the health monitoring of rolling bearings is important. Vibration signal analysis is a commonly used approach for fault diagnosis of bearings. Nevertheless, the collected vibration signals cannot avoid interference from noises which has a negative influence on fault diagnosis. Thus, denoising needs to be utilized as an essential step of vibration signal processing. Traditional denoising methods need expert knowledge to select hyperparameters. And data-driven methods based on deep learning lack interpretability and a clear justification for the design of architecture in a "black-box" deep neural network. An approach to systematically design neural networks is by unrolling algorithms, such as learned iterative soft-thresholding (LISTA). In this paper, the multi-layer convolutional LISTA (ML-CLISTA) algorithm is derived by embedding a designed multi-layer sparse coder to the convolutional extension of LISTA. Then the multi-layer convolutional dictionary learning (ML-CDL) network for mechanical vibration signal denoising is proposed by unrolling ML-CLISTA. By combining ML-CDL network with a classifier, the proposed denoising method is applied to the explainable rolling bearing fault diagnosis. The experiments on two bearing datasets show the superiority of the ML-CDL network over other typical denoising methods. • The multi-layer convolutional LISTA (ML-CLISTA) algorithm is derived. • The multi-layer convolutional dictionary learning network is proposed. • The proposed method is applied to the explainable rolling bearing fault diagnosis. [ABSTRACT FROM AUTHOR]

Fault diagnosis based on deep learning (DL) has been a research hotspot in recent years. However, the current neural networks are getting larger and larger, with more and more parameters and insufficient noise resistance, making it difficult to effectively apply these methods to real working conditions. To address these issues, we propose a novel deep learning method with fewer parameters and better noise resistance based on transfer adversarial subnetwork (TAS) and channel-wise thresholds (CWT), namely, anti-noise transfer adversarial convolutions (ANTAC). In the proposed method, the original data and feature vectors are mapped to reproducing kernel Hilbert space (RKHS) and processed by maximum mean discrepancy (MMD) and Wasserstein distance (WD), which makes the method more capable to distinguish the similar features without producing any additional training parameters. Furthermore, white Gaussian noise (WGN) and the soft thresholding method with CWT are used to reduce data noise and improve the robustness and noise resistance of the network. Finally, the superiority of the proposed method is validated through experiments on different datasets, network structures and the data with different SNRs. The results show that the proposed method has better feature discrimination ability, noise resistance, and fewer parameters compared with other methods. The highest accuracy of the proposed method is 99.90% on the test set. • In order to enhance the network ability in distinguishing similar types of samples, TAS is proposed. Experiments show that the proposed method can effectively improve the network stability and the feature discrimination ability with fewer parameters. • The anti-interference ability and robustness of the network is enhanced by adding a certain disturbance into it. In addition. The soft threshold function with channel-wise threshold is used to improve the denoising ability of the network. • To enhance the network generalization and sensitivity to similar features, AFV loss is calculated by WD and controlled by scaling factor λ , and finally sent into RMSprop optimizer together with classification loss for optimization. • The method can directly process the original signals and the procedures of denoising and feature extraction are finished automatically in the method, which improves the fault diagnosis efficiency and provides a further practical possibility for the deployment of end-to-end fault diagnosis platforms. [ABSTRACT FROM AUTHOR]

EEG signal recordings are often contaminated by Ocular and Muscular artifacts. Such artifacts obscure the signal's important information and complicate the diagnosis. Thus, before utilising the recorded EEG signal for any kind of diagnosis, the signal must be pre-processed to eliminate the artifacts that have contaminated the signal. In this paper a hybrid approach for denoising the EEG signal is proposed. The approach is based on Canonical Correlation Analysis (CCA) as a Blind Source Separation (BSS) technique for elimination of muscular artifacts. In the next step, DFA thresholded EMD process is utilized for the rejection of ocular artifacts. The performance of the proposed hybrid approach is compared with existing techniques on the basis of different performance metrics. [ABSTRACT FROM AUTHOR]

Detecting damage in coated pipelines is a challenging and costly task. This study proposes a method for pipeline defect identification based on VMD-DWT noise reduction and GA-SENet-ResNet18. Combining wavelet transform to convert denoised defect signals into time-frequency representations enhances the model's ability to capture both time-domain and frequency-domain features of defect signals, thereby improving its recognition capability for different types of defects. The study analyzed the feature extraction capabilities of ALexNet, GooleNet, VGG16, ResNet18, SENet-ResNet18, and GA-SENet-ResNet18 models in pipeline defect recognition. Experimental results show that SENet-ResNet18 achieved an accuracy of 0.9591 on the training set in 9m38s, significantly outperforming the first four models. GA-SENet-ResNet18 achieved 96.83 % accuracy, 96.67 % precision, 96.73 % recall, and 96.68 % F1 score in pipeline defect signal recognition. Compared to ResNet18, it improved accuracy by 2.06 %, precision by 1.94 %, recall by 2.09 %, F1 score by 2.37 %, with a reduction in time by 1m1s. The study demonstrates that the combined improvement of GA and SENet enhances ResNet18 not only in feature selection and response enhancement but also significantly improves its performance compared to traditional ResNet18 networks, making it more effective in pipeline defect recognition tasks. This research is crucial for ensuring pipeline system integrity and preventing pipeline accidents. • VMD-DWT for pipeline defect signal denoising. • SENet improved ResNet18. • GA-SENet-ResNet18 for pipeline defect recognition. [ABSTRACT FROM AUTHOR]

Graph Neural Networks (GNNs) have drawn great attention in handling graph-structured data. To characterize the message-passing mechanism of GNNs, recent studies have established a unified framework that models the graph convolution operation as a graph signal denoising problem. While increasing interpretability, this framework often performs poorly on heterophilic graphs and also leads to shallow and fragile GNNs in practice. The key reason is that it encourages feature smoothness, but ignores the high-frequency information of node features. To address this issue, we propose a general framework for GNNs via relaxation of the smoothness regularization. In particular, it employs an information aggregation mechanism to learn the low- and high-frequency components adaptively from data, offering more flexible graph convolution operators compared to the smoothness-promoted framework. Theoretical analyses demonstrate that our framework can capture both low- and high-frequency information of node features, effectively. Experiments on nine benchmark datasets show that our framework achieves the state-of-the-art performance in most cases. Furthermore, it can be used to handle deep models and adversarial attacks. [ABSTRACT FROM AUTHOR]

Sequential recommender systems (SRSs) aim to suggest next item for a user based on her historical interaction sequences. Recently, many research efforts have been devoted to attenuate the influence of noisy items in sequences by either assigning them with lower attention weights or discarding them directly. The major limitation of these methods is that the former would still prone to overfit noisy items while the latter may overlook informative items. To the end, in this paper, we propose a novel model named M ulti-level S equence D enoising with C ross-signal C ontrastive L earning (MSDCCL) for sequential recommendation. To be specific, we first introduce a target-aware user interest extractor to simultaneously capture users' long and short term interest with the guidance of target items. Then, we develop a multi-level sequence denoising module to alleviate the impact of noisy items by employing both soft and hard signal denoising strategies. Additionally, we extend existing curriculum learning by simulating the learning pattern of human beings. It is worth noting that our proposed model can be seamlessly integrated with a majority of existing recommendation models and significantly boost their effectiveness. Experimental studies on five public datasets are conducted and the results demonstrate that the proposed MSDCCL is superior to the state-of-the-art baselines. The source code is publicly available at https://github.com/lalunex/MSDCCL/tree/main. [ABSTRACT FROM AUTHOR]

• A novel surface recovery approach for white light scanning interferometry, which can offer high-accuracy nano-scale surface profiling. • The ISGMD method is proposed as an extension of the SGMD method, with the introduction of spectrum matching. • The relationship between the height difference and the cross-power spectrum phase is established with a more accurate expression by using the Cor method. • Compared with traditional correlation method, the proposed ISGMDCor method offers superior repeatability of measurements. White light scanning interferometry (WLSI) is a widely applied technique for surface recovery, however, the interference signal is often corrupted by various environmental and system disturbances. In this paper, a novel surface recovery method named ISGMDCor is proposed, in which the improved symplectic geometry mode decomposition (ISGMD) for signal denoising and the correlation (Cor) method based on cross-power spectrum phase for surface recovery are combined for high precision measurement. The proposed method decomposes the noisy signal into the superposition finite symplectic geometric components. Moreover, the components with high matching degree to the spectrum distribution of illuminator are selected to reconstruct the denoised signal. The surface recovery achieved by originally positioning zero optical path difference is replaced by solving the cross-power spectrum phase of denoised signals at different pixels. The simulations are carried out under different levels of signal-to-noise ratio (SNR), which demonstrate that the proposed method could effectively reduce the root mean square error (RMSE) down to 0.78 nm when the SNR is 20 dB. In experiments, a step height standard (VLSI, 45.5 nm ± 1.2 nm) is measured to demonstrate the accuracy and the noise immunity of the proposed method, and the mean height of repeated step measurements 45.21 nm with repeatability 1.22 % is achieved, where measurements obtained by commercial profiler are also conducted as comparisons. At the last part of the experiments, two roughness standards with nominal Sa value 0.4 μm and 1.6 μm are tested to further verify the robustness of the proposed method. [ABSTRACT FROM AUTHOR]

Traditional leak location methods based on signal saltation points make it hard to locate the position of the weak leak event in the energy transportation system. To address it, a data-driven method based on signal attenuation mechanism is proposed in this paper. First, a wavelet adaptive threshold denoising procedure is proposed to obtain the pure signal from the collected data, and the decimal-coded genetic mechanism is designed to calculate the optimal threshold of wavelet decomposition for different layers. Second, the two-stage leak location method is proposed to calculate the position of the leak point through the combination analysis of pressure data change of both ends of the pipeline. Through the presented pressure signal attenuation mechanism, the leak location result is obtained from the data changes of both ends of the pipeline, not the time difference of arrived signal. With the above signal denoising and analysis way, the proposed method has the 1% location deviation of the leak point, which is superior to leak location methods based on signal saltation points such as wavelet transform. Finally, different experiment results are studied, and the analysis results show the effectiveness of the proposed method for the leak location problem. • This study proposes a wavelet adaptive threshold denoising procedure. • This study proposes a two-stage leak location method. • The leak location result is obtained from the data changes of both ends of the pipeline. [ABSTRACT FROM AUTHOR]

De-noising of ECG signal is very much necessary to monitor the heart health and to diagnose disease. This paper presents a real time de-noising system that efficiently wiped out the noises from corrupted ECG signal and improves its SNR which enhances the interpretation of features present in ECG signal. This proposed system is designed with three Infinite Impulse Response (IIR) elliptic filters in cascaded direct form II structure and is implemented on the FPGA platform. Total 376 ECG samples are taken from the open access public database physiobank. All the architectural level FPGA designs have been developed in Xilinx System Generator (XSG) and implemented on Xilinx Zynq 7000 board. Our design utilized only 3.87 % of the total available resource and consumed 0.321 watt on-chip power. Worst Negative Slack (WNS) for the critical path is 2.101 ns. The positive value of WNS signifies our hardware design implementation run successfully by meeting the required time constraint. To judge the performance of the proposed design, the measurement matrices SNR Improvement, RMSE, PRD, SSNR and SINAD are calculated and achieved 81.11% , 88.89% and 85.41% accuracy for ECG recordings of ECG-ID, MIT-BIH Normal Sinus Rhythm and MIT-BIH Arrhythmia database respectively. The novelty of this proposed denoising system is its robustness, which work satisfactorily for Baseline Wander (BW), Electrode Motion (EM), Power Line Interference (PLI), Muscle or Electromyography (EMG) and as well as Additive White Gaussian Noise (AWGN). • Design of a de-noising filter using Elliptic approximation. • To avoid coefficient quantization, we used direct from II cascaded filter structure. • Simulate the proposed filter model in Xilinx System Generator. • Hardware model is implemented in Xilinx FPGA ZedBoard Zynq-7000. • The model is tested on 376 ECG recordings and achieved maximum accuracy of 88.89%. • 3.87% of total available resource is utilized for FPGA implementation. [ABSTRACT FROM AUTHOR]

The paper addresses the issue of signal confusion between bubbles and particles in real-time online measurements using machine learning methods based on the principle of electrical sensing zone (ESZ). This is crucial for obtaining accurate particle size information, particularly in the process control of three-phase stirred reactor systems. For this propose, based on the previously developed and successfully applied Online Micron-sized Particle Analyzer (OMiPA), this paper presents a method for online measurement and real-time identification of microparticle types in multiphase systems. Following this approach, firstly, simulated signals are denoised and then real-time electrical signals are obtained, which generated while microparticles pass through the ESZ. Subsequently, by analyzing the differences, searching the peak points and extracting the pulse waveforms of the signals, the identification of microparticles and bubble signals within multiphase systems can be conducted. Two types of data input are designed in this study based on the differences in the pulse signals: one based on pulse signal features and the other based on pulse events, and then K-Nearest Neighbors (KNN) is chosen to perform the real-time identification work according to the comparison of several supervised learning models. Moreover, concentration gradient experiments are designed to validate the reliability of this approach, with a model classification accuracy of up to 97%. This method effectively addresses the issue of bubbles interfering during online microparticle measurements of OMiPA in multiphase systems. It not only provides the technical support for precise and intelligent control of multiphase reaction processes in the field of chemical engineering, but also offers inspiration for innovatively implementing microparticle type identification following by the ESZ principle. [Display omitted] • Novel real-time method for microparticle measurement in three-phase systems. • Advanced signal processing boosts particle size accuracy for process control. • KNN algorithm achieves 97% accuracy in real-time microparticle identification. • Widespread industrial use of precise microparticle measurement for optimization. • Future research to enhance IoT integration for precise industrial process control. [ABSTRACT FROM AUTHOR]

• A real-time lithology recognition method based on sound signals, which achieves rapid lithology recognition and reduces identification costs. • A novel joint denoising framework to reduce errors in sound signals reconstruction. • An efficient wavelet transform function, which achieves adaptive selection of wavelet threshold. • Simulation and experimental cases verify the feasibility and effectiveness of the proposed method. The present study proposes a joint denoising method for real-time drilling lithology identification based on drilling sound signals. The proposed method employs ensemble empirical mode decomposition (EEMD) for preliminary sound signal denoising while retaining the effective intrinsic mode function (IMF) components, which then undergo secondary denoising using the improved wavelet threshold function. Finally, the sound signal is reconstructed using the denoised IMF components. The experimental results show that the proposed method achieves a higher signal-to-noise ratio and lower root mean square error, reaching 12.94 and 0.0442, respectively. Furthermore, the proposed method demonstrated good performance in denoising rock drilling sound signals and successfully addressed issues including the susceptibility of sound signals to external interference, nonadaptive threshold selection in wavelet threshold denoising, and high reconstruction errors during EEMD in drilling environments. Simulation results show that the proposed method significantly improves the accuracy of lithology recognition by 23%, with a resulting accuracy of 72.58%. [ABSTRACT FROM AUTHOR]

[Display omitted] • Physiological signals from other body regions can interfere with EEG signals. • EMG and EOG artifacts are two of the most common artifacts in EEG signals. • GRU and self-attention are the state-of-art techniques for sequential data. • The model can reconstruct a clear EEG waveform with a decent SNR and RRMSE value. • The denoising algorithm can be applied to the preprocessing stage of the EEG signal. Electroencephalography (EEG) signals are nonlinear and non-stationary sequences that carry much information. However, physiological signals from other body regions may readily interfere with EEG signal capture, having a significant unfavorable influence on subsequent analysis. Therefore, signal denoising is a crucial step in EEG signal processing. This paper proposes a bidirectional gated recurrent unit (GRU) network based on a self-attention mechanism (BG-Attention) for extracting pure EEG signals from noise-contaminated EEG signals. The bidirectional GRU network can simultaneously capture past and future information while processing continuous time sequence. And by paying different levels of attention to the content of varying importance, the model can learn more significant feature of EEG signal sequences, highlighting the contribution of essential samples to denoising. The proposed model is evaluated on the EEGdenoiseNet data set. We compared the proposed model with a fully connected network (FCNN), the one-dimensional residual convolutional neural network (1D-ResCNN), and a recurrent neural network (RNN). The experimental results show that the proposed model can reconstruct a clear EEG waveform with a decent signal-to-noise ratio (SNR) and the relative root mean squared error (RRMSE) value. This study demonstrates the potential of BG-Attention in the pre-processing phase of EEG experiments, which has significant implications for medical technology and brain-computer interface (BCI) applications. [ABSTRACT FROM AUTHOR]

Mechanical system usually operates in harsh environments, and the monitored vibration signal faces substantial noise interference, which brings great challenges to the robust fault diagnosis. This paper proposes a novel attention-guided joint learning convolutional neural network (JL-CNN) for mechanical equipment condition monitoring. Fault diagnosis task (FD-Task) and signal denoising task (SD-Task) are integrated into an end-to-end CNN architecture, achieving good noise robustness through dual-task joint learning. JL-CNN mainly includes a joint feature encoding network and two attention-based encoder networks. This architecture allows FD-Task and SD-Task can achieve deep cooperation and mutual learning. The JL-CNN is evaluated on the wheelset bearing dataset and motor bearing dataset, which shows that JL-CNN has excellent fault diagnosis ability and signal denoising ability, and it has good performance under strong noise and unknown noise. • The feasibility of vibration signal denoising based on deep learning has been studied. • An excellent end-to-end CNN-based vibration signal denoising method is proposed. • A new solution to improve the noise robustness of the fault diagnosis model is proposed. • A new network framework based on joint learning is designed, which performs well under unknown noise. [ABSTRACT FROM AUTHOR]

Graph signal denoising problem is studied by estimating a piecewise constant signal over an undirected graph. A new Bayesian approach is proposed that first converts a general graph to a chain graph via the depth first search algorithm, and then imposes a heavy-tailed t -shrinkage prior on the differences between consecutive signals over the induced chain graph. Posterior computation can be conveniently conducted by fully exploring the conjugacy structure in the model. Posterior contraction rate is derived for a general class of Bayesian shrinkage estimators under suitable assumptions, including the proposed t -shrinkage prior as a special case, and it is shown that this rate is optimal up to a logarithmic factor, besides automatically adapting to the unknown edge sparsity level of the graph. The excellent empirical performance of the proposed method is demonstrated via extensive simulation studies and applications to stock market data and signal denoising over large real-world networks. • We propose a new Bayesian approach to study the graph signal denoising problem. • The posterior computation can be conducted using the conjugacy structure. • We derive the posterior contraction rate and show it is optimal up to a log factor. • Our method is demonstrated by simulation studies and two real-world applications. [ABSTRACT FROM AUTHOR]

• Improving DBO optimization through optimization strategies. • Introduction of random variables to improve Tent mapping instability. • The energy loss rate is used as the objective function of the algorithm. • Improvement of the threshold function by tuning factors. To accurately detect power quality disturbance (PQD) signals in noisy environments, a detection method for PQD signals based on parameter-optimized variational mode decomposition (VMD) combined with improved wavelet thresholding is introduced. Firstly, the VMD decomposition of the original signal is optimized by the Improved Dung Beetle Optimizer (IDBO) to obtain the intrinsic modal function (IMF). Next, the IMF is categorized into dominant and noisy components by correlation coefficients. Denoising of noise components is performed with wavelet thresholding. The Hilbert transform (HT) can acquire the amplitude information and the starting and ending times of the harmonic. Finally, denoising and dominant components are reconstructed to achieve noise reduction. The comparative experiment shows that IDBO is superior to the original algorithm, Gray Wolf Optimizer(GWO), and Sparrow Search Algorithm(SSA) in convergence speed and accuracy. Compared with Wavelet transform (WT) and Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), the proposed algorithm has the highest localization effect accuracy, with a maximum error of 1.2 %. SNR is improved by a maximum of 26.9 %, and the maximum error of amplitude detection is 3.8 %. [ABSTRACT FROM AUTHOR]

• A method of unsupervised learning based on sparse representation using a constructed dictionary is introduced to enhance the SNR of phase-sensitive OTDR system. • The SNRs for single point vibration and multi-point vibration tests can reach 37.95 dB and 35.76 dB, respectively, in a sensing length of 4 km. • The mean processing time of this kind of denosing method for each single OTDR trace is ∼ 1.24 s. In the past few decades, phase sensitive optical time domain reflectometry (Φ-OTDR)-based distributed optical fiber vibration sensing system has gained widespread recognition from both engineers and researchers because of its superior advantages. However, the Rayleigh backscattered traces received by this kind of sensing system are highly susceptible to various forms of noise, which can seriously deteriorate the signal-to-noise ratio (SNR) of the system. In order to effectively eliminate the unwanted noise and significantly enhance the SNR of the system, an unsupervised learning method based on sparse representation over constructed dictionary is introduced and studied in this work. Specifically, by designing a dictionary based on a symlet wavelet, we try to directly remove the unwanted noise from the differential OTDR traces by their counterpart sparse coefficients. Compared to the existing signal denoising approaches, the proposed method can adaptively and rapidly utilize the redundant textures of the received sensing data without the need for parameter settings or prior training. To validate the effectiveness of the proposed signal denoising approach, experiments based on the Φ-OTDR sensing system have been carried out. The experimental results show that the SNR of the single point vibration and the multi-point vibration tests can reach 37.95 dB and 35.76 dB, respectively. Moreover, the mean processing time of this kind of denosing method for each single OTDR trace is ∼ 1.24 s. Therefore, we believe that the proposed signal processing method has great potential for accurate vibration perception and detection in the Φ-OTDR-based system operating under harsh working conditions. [ABSTRACT FROM AUTHOR]

• Denoising and harmonic interference removal from the measurement signal is crucial for determining the actual parameters of the power generating unit model and requires various transient signals data recorded during a well-defined testing procedure. However, the recorded signals usually contain noise and harmonic interferences which are difficult to remove due to a presence of different kinds of discontinuities. • The paper presents a proposal of optimal denoising and harmonic interference removal method from transient measurement signals of 200 MW power generating unit. The method combines notch filters for harmonic rejection followed by SASS (Sparsity Assisted Signal Smoothing) algorithm for wideband noise rejection. Parameters of the notch filters as well as the SASS are determined using popular optimization algorithms (simulated annealing, Nelder-Mead simplex and a simple gradient algorithm). • Performance of the proposed method is compared to its direct Linear Time-Invariant (LTI) competitors such as low-pass and notch filters and to the optimal SASS method alone. The performance evaluation is based on a set of specially generated test signals which are using the developed mathematical model of the signals (to set the level of noise and interferences) as well as the mathematical model of the power generating unit (to provide the reference clean signals). The obtained results confirms that the optimal notch-SASS method leads to superior results than the LTI filters or the SASS algorithm alone. • The authors believe that the methodology presented in the paper and obtained results may encourage other scientists to try the SASS or the proposed optimal notch-SASS method. Especially that according to the authors knowledge, this is the first paper where the SASS method was used for this kind of signals, which makes the SASS method relatively unknown in this field. Denoising and harmonic interference removal from the measurement signal is crucial for accurate parameters calculation of a generating unit model. Determining the actual parameters values requires various unit transient signals data recorded during a well-defined testing procedure. However, the recorded signals usually contain noise and harmonic interferences which are difficult to remove due to a presence of different kinds of discontinuities. The paper presents a proposal of optimal denoising and harmonic interference removal method from transient measurement signals of 200 MW power generating unit. The method combines notch filters for harmonic rejection followed by SASS (Sparsity Assisted Signal Smoothing) algorithm for wideband noise rejection. Parameters of the notch filters as well as the SASS are determined using popular optimization algorithms (simulated annealing, Nelder-Mead simplex and gradient algorithm). Performance of the proposed method is compared to its direct Linear Time-Invariant (LTI) competitors such as low-pass and notch filters alone, notch and low-pass filters combined and to the optimal SASS method alone. The performance evaluation is based on a set of specially generated test signals which are using the developed mathematical model of the signals (to set the level of noise and interferences) as well as the mathematical model of the power generating unit (to provide the reference clean signals). The obtained results confirm that the optimal notch-SASS method might be a valuable noise removal tool for considered class of signals and leads to superior results than the LTI filters or the SASS method alone. [ABSTRACT FROM AUTHOR]

The second harmonic signal serves as the informational foundation for the inversion of the gas concentration in the tunable diode laser absorption spectroscopy (TDLAS) detection method. However, due to the impact of noise on the second harmonic signal, the accuracy and stability of the detection system are decreased. In this paper, a digital denoising method based on Long Short-Term Memory Denoising Autoencoder (LSTM-DAE) is proposed to address the issue of multiple noise interference in TDLAS detection system. In LSTM-DAE, the conventional DAE is combined with LSTM structure. The second harmonic signal has time correlation, and the LSTM structure can selectively remember the past node state, which helps the DAE reconstruct the original data. Additionally, CO 2 absorption spectrum at 2004 nm as an example, the effectiveness of this method is confirmed through simulated and real experiments. The experimental results demonstrate that LSTM-DAE, with the signal-to-noise ratio of 46.65 dB, the correlation coefficient of 0.9983, and the system detection limit of 6.1 ppb, outperforms other competitive methods in both qualitative and quantitative evaluations. [ABSTRACT FROM AUTHOR]

Different types of noise contaminating the surface electromyogram (EMG) signal may degrade the recognition performance. For noise removal, the type of noise has to first be identified. In this paper, we propose a real-time efficient system for identifying a clean EMG signal and noisy EMG signals contaminated with any one of the following three types of noise: electrocardiogram interference, spike noise, and power line interference. Two statistical descriptors, kurtosis and skewness, are used as input features for the cascading quadratic discriminant analysis classifier. An efficient simplification of kurtosis and skewness calculations that can reduce computation time and memory storage is proposed. The experimental results from the real-time system based on an ATmega 2560 microcontroller demonstrate that the kurtosis and skewness values show root mean square errors between the traditional and proposed efficient techniques of 0.08 and 0.09, respectively. The identification accuracy with five-fold cross-validation resulting from the quadratic discriminant analysis classifier is 96.00%. • A novel and efficient algorithm for identifying clean and noisy electromyogram signals is proposed. • It uses a minimal set of statistical features and a straightforward classifier. • It significantly reduces computational complexity and memory requirements. • It accurately identifies clean EMG signals and those contaminated by ECG interference, spike noise, or power line noise. • Results were demonstrated in real-time using the Arduino Mega 2560, a microcontroller board based on the ATmega2560 chip. [ABSTRACT FROM AUTHOR]

• The SVD across OR-PAM A-lines exhibited temporal-spatial photoacoustic features. • Unwanted features overlap among photoacoustic-related singular vectors. • Data-driven weighting matrices from temporal-spatial singular vectors were devised. • Enhanced image and signal quality were achieved in a short computation time. The prestige target selectivity and imaging depth of optical-resolution photoacoustic microscope (OR-PAM) have gained attentions to enable advanced intra-cellular visualizations. However, the broad-band nature of photoacoustic signals is prone to noise and artifacts caused by the inefficient light-to-pressure translation, resulting in poor image quality. The present study foresees application of singular value decomposition (SVD) to effectively extract the photoacoustic signals from these noise and artifacts. Although spatiotemporal SVD succeeded in ultrasound flow signal extraction, the conventional multi frame model is not suitable for data acquired with scanning OR-PAM due to the burden of accessing multiple frames. To utilize SVD on the OR-PAM, this study began with exploring SVD applied on multiple A-lines of photoacoustic signal instead of frames. Upon explorations, an obstacle of uncertain presence of unwanted singular vectors was observed. To tackle this, a data-driven weighting matrix was designed to extract relevant singular vectors based on the analyses of temporal-spatial singular vectors. Evaluation on the extraction capability by the SVD with the weighting matrix showed a superior signal quality with efficient computation against past studies. In summary, this study contributes to the field by providing exploration of SVD applied on A-line signals as well as its practical utilization to distinguish and recover photoacoustic signals from noise and artifact components. [ABSTRACT FROM AUTHOR]

Random telegraph signals (RTSs) are specific time-fluctuating signal patterns marked by a series of distinctive switching events between well-defined signal levels. These signals are ubiquitous in many electronic, chemical, and biological devices and systems. Analyzing RTSs unveils associated system structures and internal operation mechanisms, offering valuable insights into performance sensitivity. Therefore, accurate parameter quantification of RTSs is essential for understanding their origin and significance. While two-level RTS analysis is straightforward, complications arise at multiple levels, especially with unwanted background fluctuations. To address this challenge, we developed a novel denoising autoencoder model with U-Net and bidirectional long short-term memory (DAE UBL) for denoising multi-level RTSs degraded by Gaussian white and pink noise. DAE UBL extracts lower-dimensional latent features with its encoder and reconstructs denoised RTS with its decoder. Trained and validated with large datasets of noisy multi-level RTSs, our DAE UBL demonstrates superior and stable denoising performance compared to four classic models with lower average median root mean squared errors by over 78% and 63% for all RTS data accompanying various strengths of white noise and pink noise. Average median signal-to-noise ratios in the DAE UBL analysis are increased by over 65% and 56% for the white noise and pink noise datasets. In the time domain, DAE UBL effectively suppresses both local and global fluctuations, thereby successfully removing background noise. Our model exhibits robust performance in denoising multi-level RTSs with strong pink noise. We envision that our DAE UBL will be an attractive denoising methodology for the complex multi-level RTS analysis. • We build novel denoising autoencoder for complex random telegraph signal analysis. • Our denoising autoencoder effectively suppresses local and global background noise. • We successfully identify digitized telegraph signals from noisy multilevel data. • We give extensive performance results of many multilevel random telegraph signals. [ABSTRACT FROM AUTHOR]

[Display omitted] • An improved SSA-based wavelet threshold denoising method was proposed by introducing two adjustable factors. • The Hilbert Huang transform was used to decompose the infrasound signal, and the signal was preliminarily filtered based on Pearson correlation coefficient. • Three wavelet threshold parameters were determined in the infrasound simulation. • Experiments on debris flows with different volumes and velocities were conducted in Jiangjia Gully, and the effectiveness of the method was verified. The measurement and analysis of infrasound are widely utilized in debris flow monitoring and early warning. However, the infrasound signals are often contaminated by noise across various frequencies, posing significant challenges for feature extraction. The current methods for infrasound denoising are somewhat rudimentary and face several constraints. This study introduces a novel method for enhancing infrasound signals through denoising. This approach integrates empirical mode decomposition with an enhanced wavelet threshold algorithm, and improves the conventional wavelet threshold function. This study uses the Signal-to-Noise ratio, smoothness and correlation coefficient to fine-tune the wavelet parameters. Experiments in flumes with varying speeds and volumes were conducted to test the effectiveness of the denoising method. The results demonstrate that, compared to traditional methods, the proposed method improves the SNR by averages of 69.56%, 60.91%, and 55.63%. It offers a new alternative for the noise reduction of debris flow infrasound signals. [ABSTRACT FROM AUTHOR]

Engineering typically deploys multi-points to capture structural dynamic response signals which are vulnerable to noise interference. Non-stationary random vibration makes most signal feature-based denoising algorithms unsuitable for dynamic response signals' denoising. Truncated singular value decomposition is an extensively applied method in this field for fewer requested parameters and less dependence on signals, but originally designed for single-point and fails to take the spatial correlation of multi-points into account. This paper extended it and proposed a multivariate denoising algorithm suitable for structural dynamic response multi-points monitoring data. Firstly, singular value decomposition was processed on each signal element in multi-points monitoring data produced the decomposed signals. Then, singular value decomposition was performed again on the obtained decomposed signals in the time and spatial domains. Finally, it were determined that truncation points for multivariate signals denoising by hypothesis test on their singular value spectrums. The proposed algorithm's denoising performance was compared with four other denoising algorithms through several instance calculations. Instance calculations involved two different data sets of structural acceleration response monitoring data under seismic excitation: the simulated data of a frame structure and the measured data from a certain gravity dam. The proposed algorithm exhibited almost optimal denoising performance, particularly in processing colored noise. Compared with four other algorithms involved, the proposed algorithm averagely improved denoising assessment indicators by − 3.166%, 83.984%, 78.844% and 24.888% for white noise, as well as 30.562%, 103.311%, 103.695% and 35.124% for colored noise respectively. The proposed algorithm is effective and feasible, with significant advantages in the structural dynamic response multi-points monitoring data denoising. It extends the truncated singular value decomposition algorithm suitable for univariate denoising in structural dynamic response to multivariate denoising, expands its application scope, improves the algorithm's performance, and makes the selection of the parameter in the algorithm: the truncation points more reasonable. It is of great significance for improving the accuracy and reliability of structural health monitoring. • utilized the temporal and spatial multivariate signals to select truncation points. • determined the truncation points of signals denoising by hypothesis test. • a novel extension of the TSVD in multi-points monitoring signals denoising. • suitable for the structural dynamic response multi-point monitoring data denoising. [ABSTRACT FROM AUTHOR]

• Develop a self-supervised learning model (GCL) to learn representations from unlabeled ECG signals. • GCL combines the advantages of generative and contrastive learning. • Suitable for downstream classification and denoising tasks. • GCL is deployed on a heterogeneous SoC for designing portable cardiovascular disease detection devices. With the wide use of electrocardiogram (ECG) technology, more and more ECGs have been collected and stored. However, ECG labeling is costly and laborious, the utilization of unlabeled ECG is still a critical challenge. Self-supervised learning (SSL) is a way to deal with this problem, which can learn representations from unlabeled ECG and use them for downstream tasks. In this study, a novel SSL model that fuses generative learning (denoising autoencoder, DAE) and contrastive learning (CL) is proposed to learn robust representations from unlabeled ECG for downstream denoising and classification. The 12-lead ECGs are separated into one-dimensional single-lead ECGs and ECG-specific noises are added to the original ECGs as the data augmentation method. To improve the model's denoising and feature extraction abilities, the reconstruction loss and contrastive loss are combined during the pretraining phase. In the downstream classification task, a multi-branch network is used to enhance the correlation between different ECG leads. As a result, it improves the denoising performance over the standard DAE by an average of 5.20%. In the classification tasks for the Chapman, PTB-XL, and CPSC2018 databases, compared to existing work, the accuracies in linear probe are increased by at least 2.49%, 1.09%, and 2.61%, respectively. In addition, an SoC (system-on-a-chip) based heterogeneous deployment scheme is designed. It is 13.69–14.26 times faster in inference than pure software deployment and enables real-time detection of ECG signals. The proposed method provides a good way to utilize unlabeled ECG, and the designed deployment scheme has great potential for medical applications. [ABSTRACT FROM AUTHOR]

With the continuous use of various new radar systems and complex radar systems, the electromagnetic environment is extremely deteriorated. The traditional emitter recognition methods have been difficult to meet the requirements of recognition performance in the rapidly changing battlefield environment. To solve this problem, a radar electronic signal recognition algorithm based on wavelet transform and deep learning is proposed in this paper. Starting from the radar reconnaissance system, the causes of signal preprocessing are analyzed, and the methods of signal denoising, signal normalization, signal intra pulse modulation recognition, multipath signal detection and suppression are deeply studied. In particular, the denoising algorithm based on threshold wavelet transform is proposed, which significantly improves the reliability of the algorithm. Aiming at the individual feature extraction of emitter signal, the extraction methods of emitter signal time domain feature, frequency domain feature, fuzzy function slice feature and cyclic spectrum feature based on wavelet transform are studied and analyzed, which provides stable and reliable classification features for emitter signal recognition. According to the characteristics of radar emitter signal, an optimized convolution neural network is designed, and the feature fusion processing is carried out at the decision-making level, which greatly improves the recognition effect and enhances the robustness of the recognition system. Experiments on radar data show that the fusion recognition rate is higher than any single feature and has strong robustness. In addition, compared with the traditional SVM and elm networks, the CNN network proposed in this paper can extract detailed features more effectively and improve the recognition rate of electronic radar. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have been widely applied to machinery health management in recent years, whereas research on data-driven denoising methods is relatively limited. Therefore, this paper proposes a robust denoising method based on a non-local fully convolutional neural network (NL-FCNN). In this neural network, the Leaky-ReLU activation function is employed to maintain the information contained in the negative value of the signal. The wide kernel principle is also adopted to enlarge the receptive field. Lastly, the non-local means (NLM) is utilized to construct non-local block (NLB), which could efficiently enhance the long-range dependencies learning ability of the network. This block could enormously improve the denoising performance of the network. Moreover, the proposed method exhibits better performance compared with the three conventional denoising methods under multiple noise levels on the Case Western Reserve University (CWRU) motor bearing dataset. Ultimately, we also demonstrate its application to rolling bearing fault diagnosis. • A principle that uses a machine learning approach to deal with rolling bearing VSD is proposed. • A DL-based end-to-end network for the VSD task is presented. • The use of NLB in the VSD network is explored. [ABSTRACT FROM AUTHOR]

Rolling bearing are one of the most important rotating components in the rotating machine and they are prone to failure due to complex harsh working conditions. However, it's a difficult issue to extract the fault feature from heavy noisy signal. In this paper, a new multilayer hybrid denoising method called local mean decomposition (LMD)-based enhanced sparse coding shrinkage (SCS) (named LMD-EnSCS) is proposed for impulse component extraction from bearing vibration signals. First, LMD is used to decompose the signal into a set of product functions (PFs) to enable impulse component as prominent as possible. Moreover, the multiple criteria is proposed to select the effective PF components. The rest filtering layer is used to enhance and extract the impulsive feature through minimum entropy deconvolution (MED) and sparse coding shrinkage (SCS) respectively. MED is able to enhance the impulsive component of noisy signal, which improve the impulsive extraction ability of SCS effectively. The simulation signal and bearing signal are used to verify the effectiveness and superiority of LMD-EnSCS, and the results show that LMD-EnSCS can extract the fault feature and perform well for bearing fault diagnosis. [ABSTRACT FROM AUTHOR]

In this paper, we introduce convolutional proximal neural networks (cPNNs), which are by construction averaged operators. For filters with full length, we propose a stochastic gradient descent algorithm on a submanifold of the Stiefel manifold to train cPNNs. In case of filters with limited length, we design algorithms for minimizing functionals that approximate the orthogonality constraints imposed on the operators by penalizing the least squares distance to the identity operator. Then, we investigate how scaled cPNNs with a prescribed Lipschitz constant can be used for denoising signals and images, where the achieved quality depends on the Lipschitz constant. Finally, we apply cPNN based denoisers within a Plug-and-Play framework and provide convergence results for the corresponding PnP forward-backward splitting algorithm based on an oracle construction. [ABSTRACT FROM AUTHOR]

Electrocardiogram (ECG) signals are used to diagnose cardiovascular diseases. During ECG signal acquisition, various noises like power line interference, baseline wandering, motion artifacts, and electromyogram noise corrupt the ECG signal. As an ECG signal is non-stationary, removing these noises from the recorded ECG signal is quite tricky. In this paper, along with the proposed denoising technique using stationary wavelet transform, various denoising techniques like lowpass filtering, highpass filtering, empirical mode decomposition, Fourier decomposition method, discrete wavelet transform are studied to denoise an ECG signal corrupted with noise. Signal-to-noise ratio, percentage root-mean-square difference, and root mean square error are used to compare the ECG signal denoising performance. The experimental result showed that the proposed stationary wavelet transform based ECG denoising technique outperformed the other ECG denoising techniques as more ECG signal components are preserved than other denoising algorithms. • ECG signals contaminated with random noises are denoised using SWT is proposed. • Qualitative and quantitative analysis is performed on the MIT-BIH arrhythmia database. • The results demonstrate the superiority of the proposed SWT denoising methodology. • Periodic or quasi-periodic biomedical signals can be denoised using the proposed SWT. [ABSTRACT FROM AUTHOR]

Due to the excellent insulation and arc extinguishing properties of SF 6 gas, it is widely used in power equipment such as gas insulated switchgear or gas circuit breaker. Therefore, detection of SF 6 gas leakage with high sensitivity is necessary. Non dispersive infrared (NDIR) technology have advantages such as miniaturization, low power consumption, simple structure, low cost, and non-destructive testing, making it suitable for building large-scale sensor networks. However, NDIR sensors are susceptible to noise coming from the emitter, the detector and the circuits, which receives little research attention. In this study, we propose a new denoising method by combining improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) with Savitzky-Golay (S-G) to the noise suppression, which achieves a 7 dB signal-to-noise ratio improvement over the traditional wavelet threshold denoising, making the detection limit of the sensing system reach the level below 1 ppm. • Using noise reduction algorithms to improve the detection limit of NDIR sensors in SF 6 leakage detection. • A new denoising algorithm has been designed, which improves the signal-to-noise ratio by 7 dB compared to traditional wavelet threshold denoising. [ABSTRACT FROM AUTHOR]

[Display omitted] • Innovative flow cell for GASMAS sensing manufactured by 3D SLA technique. • Signal-to-noise ratio improvement based on spectral line-shape fitting and FFT denoising. • Reduced concentrations of oxygen detected employing 2f-WMS with good linearity. • Time-resolved monitoring of carbon dioxide diffusion in porous media was demonstrated. Signal denoising is a serious problem for in-situ laser diagnostics of gases dispersed in porous materials. An optical sensor system based on absorption spectroscopy of gases in a scattering environment was built using a 3D printed cell with reference samples of polystyrene foam. Selected A-band spectral lines of molecular oxygen were investigated using wavelength modulated spectroscopy with second harmonic detection. Quantitative information on the concentration of analyte dispersed in the porous medium was obtained at extremely low signal-to-noise ratio (SNR < 10). A spectral line shape fitting procedure based on the Gabor transform followed by a filtered inverse fast Fourier transform allowed to achieve a relatively high SNR with good linearity over a range of reduced oxygen concentrations in air. Finally, the applicability of the optical sensor system to monitor the diffusion of carbon dioxide into air dispersed in a Styrofoam sample and vice versa was successfully demonstrated. [ABSTRACT FROM AUTHOR]

• PASGWT considers both the optimization of design parameter and decomposition layers. • Combined kurtosis index is defined to characterize cyclic shocks and cyclostationary. • HPO is applied to find the optimal SGWT decomposition parameters. • Case studies prove the effectiveness and superiority of PASGWT. The signal noise can have a negative impact on the accuracy of engineering equipment condition assessment. However, the choice of decomposition layers and design parameters significantly impacts the outcomes of spectral graph wavelet transform (SGWT) methods. Conventional SGWT techniques often rely on predefined design parameters and decomposition layers, which can limit satisfactory decomposition results. To address this challenge, we present a novel parameter-adaptive SGWT method, referred to as PASGWT, designed for denoising vibration signals from wind turbines (WTs). PASGWT adaptively adjusts the optimal design parameters and decomposition layers to align with the signal characteristics. Furthermore, it introduces a novel evaluation metric to effectively identify fault-related features in signals corrupted by strong noise interference. Initially, the vibration signal is transformed into a directed graph representation. Subsequently, an SGWT with a warping function is constructed to process the signal. To guide the optimization process, we introduce a comprehensive evaluation index called the combination kurtosis index (CKI), which integrates periodic kurtosis and envelope spectrum kurtosis. The SGWT parameters are then optimized by the Hunter-Prey optimization (HPO) algorithm with maximum CKI value as the optimization objective. Finally, the denoised signal is reconstructed by selecting components with CKI values exceeding the average CKI value. The efficacy and practicality of the proposed method are validated through case studies involving simulated signals and two real-world fault signals from a scaled-down wind turbine test rig. Furthermore, comparative experiments highlight the superiority of the proposed method. [Display omitted] [ABSTRACT FROM AUTHOR]

A chirp cyclostationary (CCS) process is a nonstationary stochastic process, whose statistics have been defined and utilized in radar and communication signal processing. However, spectrum analysis of CCS signals, which provides a spectral view to characterize a signal, has not been performed. Classical spectrum analysis is based on the Fourier transform, which is not suitable for CCS signals because of the multiplicative chirp signals in a CCS model. In this study, we propose a generalized spectral representation of chirp cyclostationary signals associated with a linear canonical transform. The proposed representation shows that a CCS signal can be represented by a linear combination of a series of jointly stationary and equal-bandwidth processes, which bridge the nonstationary and stationary processes. Furthermore, the relationships between the second-order statistics of the CCS process and its stationary representations are introduced. Finally, based on the above observations, a linear time-varying Wiener filter was designed for CCS signal denoising and time-varying channel estimation. Finally, the relationships between the nonstationarity of the input and output and time-varying properties of the systems are discussed, which are helpful for determining the nonstationarity of the outputs or the time-varying property of a system. • A generalized spectral decomposition of chirp cyclostationary processes is proposed. • The CCS process is proven as a linear combination of joint stationary processes. • Statistics of the CCS process are linked with those of the stationary components. • A linear chirp periodically time-varying chirp cyclic Wiener filter is designed. [ABSTRACT FROM AUTHOR]

The application of deep learning in the field of power electronics fault diagnosis has garnered considerable attention and research interest. In the coil power supply of the HL-3 tokamak device, there exist practical challenges in dealing with fault data from 12-pulse thyristor power converters, including large data volume, variable frequency operation, and sensor noise, among others. To solve these problems, this paper proposes a method that combines wavelet threshold denoising and Meta Pseudo Labels. Firstly, discrete wavelet transform is used to transform the original voltage signals of the 12-pulse thyristor rectifier operating across a wide frequency range into the wavelet domain. Subsequently, a reasonable threshold is employed to filter out high-frequency noise, resulting in the output of wavelet threshold denoising. Then, Meta Pseudo Labels algorithm, an idiographic semi-supervised learning approach, is applied to learn the features of the denoised data, yielding classification models and experimental results. The model identifies three types of states in single thyristor branches of the rectifier circuit, which are normal, short-circuit fault, and open-circuit fault. The experimental results indicate that the accuracy and robustness of the model trained using the Meta Pseudo Labels algorithm outperform those of a convolutional neural network with the same network architecture. Especially, one of the experiments accomplishes a diagnosis accuracy exceeding 90% while training with the whole unlabeled training dataset and only 10% of stratified subsamples from the labeled training dataset. • Method that combines wavelet threshold denoising and Meta Pseudo Labels is proposed. • Wavelet threshold denoising can filter out high-frequency noise. • Meta Pseudo Labels helps mitigate the labeling challenges related to large dataset. • Accuracy of proposed method surpass those of CNN with the same network architecture. [ABSTRACT FROM AUTHOR]

The utilization of machine learning in analyzing ship radiated noise (SR-N) is undergoing rapid evolution. Because the omnipresent background noise strongly depends on the highly variable environment, the application of such techniques poses challenges. Furthermore, publicly available labeled datasets are scarce. Motivated by this, there has been a surge in the number of publications regarding the implementation of machine learning in the monitoring of SR-N within the past few years. This comprehensive survey delineates the state-of-the-art machine learning techniques applied to SR-N, with a specific focus on passive measurements. Recent developments are categorized into several sub-areas, namely; publicly available datasets, data augmentation, signal denoising, feature extraction, detection, localization, and recognition of SR-N. Additionally, future research directions are explored. • Machine learning (ML) in underwater sound analysis is rapidly evolving. • This survey overviews state-of-the-art ML techniques in Ship Radiated Noise. • Several preprocessing methods are discussed. • Automatic detection, localization, and recognition of ships are reviewed. • The survey ends with a forward-looking perspective for future research directions. [ABSTRACT FROM AUTHOR]

In recent years, a notable surge has occurred in the adoption of deep neural networks for addressing the challenges of seismic signal denoising. However, many existing methodologies are constrained by simplistic noise assumptions and exhibit limited generalisation capabilities, particularly when confronted with field seismic signals. Consequently, the need for a robust denoising solution tailored to the complexities of field seismic signals remains unresolved. To address this gap and effectively enhance the signal-to-noise ratio (SNR), we proposed a novel denoising approach. Central to our methodology is the incorporation of a custom-designed network structure that leverages the unique characteristics of seismic signals. Specifically, we introduced the swin-link-conv block, which amalgamates the modelling capabilities of the swin transformer block with the adaptive properties of the residual connection layer. This innovative building block was seamlessly integrated into the UNet architecture, which is a widely acclaimed framework for seismic signal processing. The resulting network architecture was a christened seismic swin-conv UNet (SSC-UNet). Extensive experiments conducted on both synthetic and field seismic signals underscore the efficacy of the proposed network structure. The SSC-UNet methodology achieved state-of-the-art denoising performance, demonstrating superior accuracy in removing additive white Gaussian noise (AWGN). This success not only validates the generalisation capabilities of SSC-UNet but also underscores its viability as a robust solution for field seismic signal denoising applications. • Combining the advantages of the advantagesswin-link-conv transformer block and UNet is used for seismic signal denoising. • The proposed method performs well in seismic signal denoising. • The proposed method can achieve denoising of seismic signals in different dimensions with different noise levels. In addition,seismic signals can also be denoised. [ABSTRACT FROM AUTHOR]

[Display omitted] • A novel vibration-based robotic ground and state classifier is proposed. • Multi-source vibration signal imaging for feature fusion. • Attention combines residual structures to characterize complex features. In agricultural environments, recognizing the walking ground and state of tracked mobile robots is a complex and challenging task, influenced by clay conditions and other external environmental disturbances. Therefore, this paper proposes a novel data processing method and efficient classifier. Firstly, the noise signals on the left and right sides of the collected robot are averaged, and a time-wavelet-time domain transformation is performed using the Mallat algorithm to achieve nonlinear enhancement of data features and signal denoising. Secondly, the Gramian Angular Summation Fields (GASF) is introduced to transform sequence data into single-channel images, capturing the periodicity and similarity of time series. Next, the images of three sets of sequences are stacked in the channel dimension in RGB format, thus achieving feature fusion of multi-source data. Finally, a supervised learning classifier named Attention-fused Residual Convolutional Neural Network (ANR-CNN) is proposed. Here, the combination of channel and spatial attention mechanisms captures important features in the feature map in both channel and spatial dimensions. The convolutional residual structure enhances feature transmission, improving the model's classification accuracy. Experimental results demonstrate that the proposed data augmentation method effectively enhances model performance, and the classification accuracy of ANR-CNN reaches 92.35%. This implies accurate recognition of the walking ground and state of tracked mobile robots in agricultural environments. [ABSTRACT FROM AUTHOR]

Due to uncertain production conditions and external disturbances, there are a lot of noise in the industrial process dataset. Traditional denoising models can only solve a single noise problem. However, the noise in industrial process data has diversity and inhomogeneity. Therefore, the noise adaptive filtering (NAF) model integrating spatio-temporal feature (ST) (ST-NAF) is proposed for the soft sensor, which achieves a balance between noise and data features to reduce the influence of noise on soft sensor results. The gated recurrent unit extracts temporal features of the industrial process data. Then, the dynamically gated graph attention extracts the interaction relationship between monitoring points to obtain spatio-temporal features. Moreover, the NAF converts time-domain features into frequency-domain features, and adaptively selects frequency-domain components with important information through the noise filtering matrix. The selected frequency-domain components are converted to the temporal domain to obtain the denoised spatio-temporal features. Finally, the self-attention learns the dynamic relationships between denoised spatio-temporal features and output variables to predict the key indicator. Compared with other denoising soft sensor models in a public dataset and an actual industrial process dataset, the ST-NAF achieves state-of-the-art results. And the prediction accuracy of the proposed model improved by 4.5% and 10.0% in terms of R2, respectively. Furthermore, the experimental results show that the ST-NAF can adaptively select useful information from industrial noise data to achieve accurate prediction of key indicators. [ABSTRACT FROM AUTHOR]

T 2 -weighted, two-point Dixon fast-spin-echo (FSE) is an effective technique for magnetic resonance neurography (MRN) that can provide quantitative assessment of muscle denervation. Low signal-to-noise ratio and inadequate fat suppression, however, can impede accurate interpretation. To quantify effects of principal component analysis (PCA) denoising on tissue signal intensities and fat fraction (FF) and to determine qualitative image quality improvements from both denoising and water-weighting (WW) algorithms to improve nerve conspicuity and fat suppression. Prospective. Twenty-one subjects undergoing MR neurography evaluation (11/10 male/female, mean age = 46.3±13.7 years) with 60 image volumes. Twelve subjects (23 image volumes) were determined to have muscle denervation based on diffusely elevated T 2 signal intensity. 3 T, 2D, two-point Dixon FSE. Qualitative assessment included overall image quality, nerve conspicuity, fat suppression, pulsation and ringing artifacts by 3 radiologists separately on a three-point scale (1 = poor, 2 = average, 3 = excellent). Quantitative measurements for FF and signal intensity relative to normal muscle were made for nerve, abnormal muscle and subcutaneous fat. Linear and ordinal regression models were used for quantitative and qualitative comparisons, respectively; 95% confidence intervals (CIs) and p -values for pairwise comparisons were adjusted using the Holm-Bonferroni method. Inter-rater agreement was assessed using Gwet's agreement coefficient (AC 2). Simulations showed PCA-denoising reduced FF error from 2.0% to 1.0%, and from 7.6% to 3.1% at noise levels of 10% and 30%, respectively. In human subjects, PCA-denoising did not change signal levels and FF quantitatively. WW decreased fat signal significantly (−83.6%, p < 0.001). Nerve conspicuity was improved by WW (odds ratio, OR = 5.8, p < 0.001). Fat suppression was improved by both PCA (OR = 3.6, p < 0.001) and WW (OR = 2.2, p < 0.001). Overall image quality was improved by PCA + WW (OR = 1.7, p = 0.04). WW and PCA-denoising improved nerve conspicuity and fat suppression in MR neurography. Denoising can potentially provide improved accuracy of FF maps for assessing fat-infiltrated muscle. [ABSTRACT FROM AUTHOR]

• A low-cost noncontact measurement system was developed for the hand motion evaluation. • Providing an automatic and remotely accessible monitoring system for hand motion analysis. • Validating the accuracy of the Leap Motion with Optotrak Certus. • A virtual environment (VE) was developed for hand and finger rehabilitation. Data collection and analysis are commonly used in a rehabilitation process to measure performances of the treatment. There is a lack of studies on the rehabilitation process monitored by a user-friendly interface. A low-cost system is developed in this research to assist users and therapists to measure hand motions and analyse important data of hand joints. The system consists of modules of data capturing, data analysis, and user interface. A Leap Motion sensor is used to capture joint positions of hand motions. Signal processing and wavelet de-noising methods are developed to improve accuracy of the data analysis. The user interface is designed using the Unity software to show graphical information of joint positions and motion parameters. The system has features of noncontact measurements, interactive environment, analysing and recording temporal data of motion parameters of hands. The system is validated by a gold standard motion capturing system. Case studies show effectiveness of the proposed system. [Display omitted] [ABSTRACT FROM AUTHOR]

The ill-posed problem of shock wave pressure (SWP) measurement comes from the influence of low-frequency noise components to measured responses and leads to an inaccurate result. To address this problem, an improved method, referred to as recursive empirical mode decomposition (REMD), is proposed for filtering low-frequency noises from the SWP measurement signals. By means of empirical mode decomposition (EMD), the measurement signal is adaptively decomposed into several intrinsic mode functions (IMFs) without any prior information. A mode classification scheme is firstly developed to select two mode indexes for separating the useful, mixed and noisy IMFs based on energy gradient and ringing amplitude ratio. Then, an adaptive diminishing white noise-assisted technology is presented to iteratively extract the remaining useful components from the mixed IMFs based on the damping ratio of SWP measurement signal. The final denoised result is achieved by a partial reconstruction with the useful IMFs and the useful components obtained from each extraction process. The effectiveness of the proposed method is verified through a series of simulated and real SWP measurement signals. Simulated results show that the REMD method always produces the largest SNR and the smallest RMSE and PRD in both single and mixed noise situations. Furthermore, the denoised results from real SWP measurement experiments with PR and PE sensors under different pressure conditions also demonstrate the superiority of the proposed method over the existing approaches in both denoising ability and signal integrity. • A low-frequency noise reduction method is proposed. • A mode classification is developed to separate different IMFs. • A recursion extraction method is proposed to separate the useful components. [ABSTRACT FROM AUTHOR]

The study of visual illusions has proven to be a very useful approach in vision science. In this work we start by showing that, while convolutional neural networks (CNNs) trained for low-level visual tasks in natural images may be deceived by brightness and color illusions, some network illusions can be inconsistent with the perception of humans. Next, we analyze where these similarities and differences may come from. On one hand, the proposed linear eigenanalysis explains the overall similarities: in simple CNNs trained for tasks like denoising or deblurring, the linear version of the network has center-surround receptive fields, and global transfer functions are very similar to the human achromatic and chromatic contrast sensitivity functions in human-like opponent color spaces. These similarities are consistent with the long-standing hypothesis that considers low-level visual illusions as a by-product of the optimization to natural environments. Specifically, here human-like features emerge from error minimization. On the other hand, the observed differences must be due to the behavior of the human visual system not explained by the linear approximation. However, our study also shows that more 'flexible' network architectures, with more layers and a higher degree of nonlinearity, may actually have a worse capability of reproducing visual illusions. This implies, in line with other works in the vision science literature, a word of caution on using CNNs to study human vision: on top of the intrinsic limitations of the L + NL formulation of artificial networks to model vision, the nonlinear behavior of flexible architectures may easily be markedly different from that of the visual system. [ABSTRACT FROM AUTHOR]

• In this paper, we develop a risk estimation framework for speech enhancement, where we optimize an unbiased estimate of the risk instead of the actual risk. The estimated risk is expressed solely as a function of the noisy observations and noise statistics. Hence, the denoiser obtained by minimizing the risk estimate does not require the clean speech prior. The stateof- the-art image denoising techniques optimize Steins unbiased risk estimate (SURE), which is an unbiased estimate of MSE, to obtain the optimum denoising function. Even though the MSE is a widely and successfully used distortion measure for signal denoising, in speech processing applications, distortion measures such as Itakura-Saito (IS), hyperbolic-cosine (cosh), weighted cosh, are known to be more perceptually relevant than MSE. Considering this into account, in this paper, we solve the speech denoising problem within the framework of perceptual risk estimation (wherein we derive unbiased estimates of speech-specific perceptual distortion measures and minimize them to obtain the corresponding denoising functions). We employ a DCT-domain pointwise shrinkage estimator for denoising where the optimum shrinkage estimator is obtained by minimizing the perceptual risk estimate. We evaluate the performance of the risk estimation-based techniques objective assessment in terms of segmental signal-to-noise ratio (SSNR), perceptual evaluation of speech quality (PESQ), short-time objective intelligibility (STOI), signal-to-distortion ratio (SDR), and subjective assessment by means of listening tests. Validation on several speech signals in real-world nonstationary noise scenarios and comparisons with benchmark techniques showed that, for input SNR greater than 5 dB, the proposed method results in better denoising performance than several benchmarking techniques. Among the risk estimation-based techniques, the quality of the denoised speech is higher (measured in terms of PESQ and subjective listening scores) for perceptual risk-based techniques than the MSE-based technique. Further, we want to emphasize that the proposed methodology is relatively simpler, from an implementation perspective since the shrinkage estimators are easy to compute, and does not require training making it ideal for deployment in practical applications, particularly for those involving hearing aids, mobile devices, etc. The goal in speech enhancement is to obtain an estimate of clean speech starting from the noisy signal by minimizing a chosen distortion measure (risk). Often, this results in an estimate that depends on the unknown clean signal or its statistics. Since access to such priors is limited or impractical, one has to rely on an estimate of the clean signal statistics. In this paper, we develop a risk estimation framework for speech enhancement, in which one optimizes an unbiased estimate of the risk instead of the actual risk. The estimated risk is expressed solely as a function of the noisy observations and the noise statistics. Hence, the corresponding denoiser does not require the clean speech prior. We consider several speech-specific perceptually relevant distortion measures and develop corresponding unbiased estimates. Minimizing the risk estimates gives rise to denoisers, which are nonlinear functions of the a posteriori SNR. Listening tests show that, within the risk estimation framework, Itakura-Saito and weighted hyperbolic cosine distortions are superior than the other measures. Comparisons in terms of perceptual evaluation of speech quality (PESQ), segmental SNR (SSNR), source-to-distortion ratio (SDR), and short-time objective intelligibility (STOI) also indicate a superior performance for these two distortion measures. For SNRs greater than 5 dB, the proposed approach results in better denoising performance — both in terms of objective and subjective assessment — than techniques based on the Wiener filter, log-MSE minimization, and Bayesian nonnegative matrix factorization. [ABSTRACT FROM AUTHOR]

• A denoising algorithm for hydroacoustic signal based on polynomial fitting is proposed to effectively eliminate noise and remove baseline drift. • A preprocessing algorithm for real-time attitude compensation of vector hydroacoustic signal is proposed and verified to have no effect on system stability and to have low time complexity, and finally, it has been proven to improve orientation accuracy. • A directional algorithm from single MEMS vector hydrophone based on weighted histogram statistics and cross-spectrum sound intensity is proposed, which can effectively suppress side-valve interference and realize higher precision directionality. In order to solve the problem of noise interference and attitude shift of vector hydrophone due to the influence of natural factors during underwater detection, it is necessary to develop a new method to solve the problem. In this paper designs an orientation algorithm of single MEMS vector hydrophone based on polynomial fitting and real-time attitude compensation. It is proved that the algorithm will not change the stability of the system, and the time complexity is low. In order to verify the effect of this algorithm, this paper compares this algorithm with other signal denoising methods and attitude correction methods, and proves that this algorithm has smaller processing error. Finally, the underwater acoustic signal actually collected in the lake is used as the input signal to calculate, and the orientation error is less than 1, which verifies the feasibility and practicability of the algorithm. [ABSTRACT FROM AUTHOR]

In recent times, predicting stock prices has garnered attention from both regulators and academic circles. However, the intricate nature of financial time-series data, with its nonlinearities, discontinuities, and sensitivity to noise, complicates the understanding and forecasting of financial movements. In our approach, we initially deploy an adaptive empirical modal decomposition on the primary data to enhance model precision. Subsequently, we sift the technical indicator data through the Boruta method, enhancing selected functionalities via an adaptive noise reduction technique. We then employ support vector regression (SVR) integrated with brain storm optimization algorithm (BSO) for effective data handling and forecasting target variables. Our results suggest that the composite model outlined in this paper outperforms the other eight comparison models in terms of reducing errors and improving regression scores. Additionally, when juxtaposed against these four models, the outcomes reinforce the efficiency of our proposed multiscale strategy and denoising technique in refining prediction accuracy. • Adaptive selection of technical indicators as input features using the Boruta feature selection algorithm. • Adaptive denoising of input features based on the proposed MIC-ICEEMDAN denoising algorithm. • The reconstruction uses the enhanced BSO-SVR model to predict the prediction results of each IMF and residual term Res. [ABSTRACT FROM AUTHOR]