Search

In response to the specific requirements of space-borne gravitational wave detectors (SGWD) for laser frequency stability, this work systematically analyzes and investigates the frequency noise of Fabry-Pérot (FP) cavity-based ultra-stable lasers (USL). Aiming at the detection frequency band of SGWD (0.1 mHz $\sim$ 1Hz), a refined laser frequency noise model of USL is proposed. Based on the model, corresponding simulations and experiments yield an approach to identify the primary noise sources within segmented frequency bands, revealing the overlooked role of temperature fluctuations on cavity noise, and emphasizing the importance of the control loop. The proposed model will provide valuable guidance for improving laser frequency stability in space gravitational wave detectors.

Assessing the invasiveness of thyroid nodules, particularly whether they have metastasized to lymph nodes, is crucial for guiding treatment decisions. Current diagnostic methods, including ultrasound imaging, are limited by operator dependence and interpretative variability, complicating accurate evaluation of nodule invasiveness. To address these limitations, this study introduces the UTV-ST Swin Transformer, a deep learning model that combines ultrasound video data with standardized clinical information to predict the invasiveness of thyroid nodules. The model classifies nodules into three categories: non-invasive, central lymph node metastasis (CLNM), and central plus lateral lymph node metastasis (CLNM+LLNM). By analyzing ultrasound video features using the Video Swin Transformer and clinical data using a text analysis module based on the KAN network, and then fusing these features, the model achieves a classification accuracy of 82.1% and an average AUC of 94.2%. These results surpass the performance of traditional methods, particularly in distinguishing different degrees of invasiveness, even under noisy conditions. This study highlights the potential of the UTV-ST Swin Transformer model in improving the accuracy of thyroid nodule assessment, reducing reliance on operator expertise, and providing a more consistent and automated method for evaluating nodule invasiveness.

The widespread adoption of battery electric vehicles (BEVs) has highlighted the critical importance of precise energy consumption prediction models to address the problem of range anxiety among drivers. This study aims to enhance the accuracy of such models by combining real-time traffic state recognition and velocity prediction, thereby mitigating range anxiety and enhancing the driving experience. Consequently, we propose an improved Fuzzy C-Means (FCM) clustering algorithm that use historical traffic data and dynamic traffic information accurately identify traffic conditions. In addition, a Fuzzy-Markov-based velocity prediction model is developed to generate future velocity profiles under diverse traffic scenarios. In the energy consumption prediction stage, a particle swarm optimization-radial basis function neural network (PSO-RBFNN) model is employed to estimation the energy consumption. Simulation results show a significant improvement in prediction accuracy, with the Mean Absolute Percentage Error (MAPE) reduced to below 3.2% under diverse traffic scenarios.

With the rapid development of information communication and mobile device technologies, smart devices have become increasingly popular, providing convenience to households and enhancing the level of intelligence in daily life. This trend is also driving innovation and progress in various fields, including healthcare, transportation, and industry. However, as technology continues to proliferate, network security concerns have become increasingly prominent, making the protection of digital life and data security an urgent priority. Intrusion detection has always played an important role in the field of network security. Traditional intrusion detection systems predominantly rely on anomaly detection technology to identify potential intrusions by detecting abnormal patterns in network traffic. With technological advancements, machine learning-based methods have emerged as the cornerstone of modern intrusion detection, enabling more precise identification of abnormal behaviors and potential intrusions by learning the patterns of normal network traffic. In response to these challenges, this paper introduces an innovative intrusion detection model that amalgamates the Attention-CNN-BiLSTM (ACBL) and Temporal Convolutional Network (TCN) architectures. The ACBL and TCN models excel in processing spatial and temporal features within network traffic data, respectively. This integration harnesses diverse neural network structures to elevate overall model performance and accuracy. Furthermore, a unique approach inspired by dung beetles’ natural behavior, incorporating Tent mapping-enhanced Dung Beetle Optimization Algorithm (TDBO), is leveraged for both optimizing feature selection parameters and searching for optimal model hyperparameters. The feature selection parameters obtained from TDBO are then combined with the importance ranking from the Random Forest algorithm, ensuring optimal features can be better selected to enhance model performance. This paper introduces a novel intrusion detection model, the TDBO-ACBLT model, and validates its performance using the UNSW-NW15 dataset. TDBO excels in feature selection compared to common algorithms and achieves superior parameter optimization accuracy over Harris’s Hawk Optimization (HHO), Particle Swarm Optimization (PSO), and Dung Beetle Optimization (DBO). The proposed model achieves higher accuracy than prevalent machine learning models.

The integration of traditional Chinese medicine (TCM) diagnostics with modern artificial intelligence (AI) techniques has emerged as a promising approach to enhance the objectivity and accuracy of disease assessment. Tongue diagnosis, a non-invasive and unique TCM practice, plays a critical role in evaluating health status but is often limited by the subjective judgment of practitioners. This study addresses these limitations by developing an intelligent tongue diagnosis system using the Cv-Swin Transformer architecture. The system processes a diverse dataset of 5,365 tongue images, classifying them into ten categories based on TCM diagnostic standards. Key findings indicate that the Cv-Swin Transformer model achieves an average accuracy of 87.37% in tongue image classification, demonstrating superior performance compared to traditional models. The system effectively captures complex tongue features related to various diseases, providing precise health assessments and personalized treatment recommendations. This research represents a significant advancement in integrating AI with TCM, offering a robust tool for objective diagnostics and supporting the modernization of traditional practices.

Waiting is an indispensable and inevitable part of man-machine voice interaction. The voice user interface (VUI) feedback mechanism is a key factor affecting voice interaction’s waiting experience. The feedback time of most available voice interfaces is fixed or decided by the processing time of hardware and software, which has not been designed and cannot offer users a good interaction experience. In this paper, the speech rate of user-machine voice interaction is collected through prototype experimentation. Besides, users’ time perception of different voice interfaces’ feedback time settings is studied based on time psychology theories. Moreover, users’ emotional changes are described after a specific feedback time with the distribution of two-dimension arousal-valence emotion space. Users’ time perception and subjective emotions are differently influenced by different VUI feedback times. The experimental results show that 750 ms is the optimal VUI feedback time point at which the best users’ subjective feelings and psychological experiences are reached, and the threshold limit time spent by users in waiting for the VUI feedback is 1,850 ms which will lead to user emotions with low levels of arousal and valence after being exceeded. Based on that, a linear regression model is proposed to define the optimal feedback time of VUI. The user experience VUI research results show that the calculated feedback time parameters can make users produce time perception in line with their expectations in interacting with voice interfaces.

The Sequentially-Switched Delay Lines (SSDL) consists of a set of RF switches and transmission lines of equal length which provides magnetless nonreciprocity with no theoretical bandwidth limit. SSDL circulator architectures are presented in this paper on a GaN MMIC. The SSDL circulators presented here demonstrate nonreciprocity from 10 MHz to 1.2 GHz with isolation between the transmitter and receiver which is greater than 20 dB for most of the measured frequency range. The insertion loss of the GaN MMIC SSDL circulator is reduced to below 3 dB from 10 MHz to 800 MHz, by using matching circuits at the gates of the switching transistors. The spurious-free operation of the SSDL circulator is also verified from 10 MHz to 1 GHz. It is also demonstrated that with passive bootstrapping, SSDL insertion loss is further reduced to about 2 dB and P1 dB performance can be significantly enhanced by an additional 10 dB. The time-modulation/switching strategy to achieve broadband magnetless nonreciprocity has the potential to be used in future STAR and full duplex communication systems.

Accurate soil structure models are crucial references for substation grounding system design. Typically, inversion algorithms are employed to obtain uniform or horizontal layered soil models based on measured apparent resistivity. However, soil resistivity outlier distribution areas can affect the accuracy of these inversion algorithms, particularly when these areas are near the surface. Traditional algorithms do not account for the outlier distribution of soil resistivity, leading to significant discrepancies between the calculated results of the design scheme and actual operation, thereby impacting the safety and economy of the grounding system. Therefore, this paper proposes an inversion method for soil structures with outlier distribution characteristics based on deep belief networks (DBNs). Firstly, we introduce a statistical criterion for identifying the outlier distribution characteristics of soil resistivity. Subsequently, we construct a database of soil models with outlier distribution characteristics to train the DBN. Finally, we verify the inversion accuracy of the optimal DBN using apparent resistivity data measured in a 220 kV substation and the Qinghai-Tibet Railway. The results demonstrate that the inversion accuracy of the method proposed in this paper is comparable to that of the traditional method for horizontally layered soil but exhibits a remarkable improvement of approximately 40% when dealing with soil apparent resistivity exhibiting outlier distribution characteristics.

In this paper, a general approach to enhance the beamwidth of microstrip antennas is proposed for wide beam coverage in both E- and H-planes. A microstrip antenna loaded with two arrays of capacitive via fences is propounded and systematically studied. By introducing the vertical currents brought by the capacitive metalized vias, the half-power beamwidth (HPBW) is effectively broadened compared with the regular microstrip antennas. In addition, by utilizing the air medium with low loss, the microstrip antenna can be supported by the two arrays of via fences, maintaining a high radiation efficiency. To validate the proposed design, a prototype is fabricated and tested. The measurement results agree well with the simulated ones, with enhanced HPBWs of 100° and 90° in E- and H-planes, respectively. Compared with the existing 2.4-GHz antennas, the propounded antenna is with the advantages of wide beamwidth and high radiation efficiency, exhibiting the potential applications for space-limited mobile devices with wide coverage requirement.

Arc plasma, characterized by its high current density, high energy concentration, and elevated gas enthalpy, is widely utilized in fields such as metallurgy, chemical industry, energy and environmental protection, as well as aerospace. Plasma treatment of solid waste is one of the advanced environmental protection technologies at the international forefront. The rapid development of plasma technology has led to continuous updates and upgrades of arc plasma power sources. This paper provides a comprehensive review of the research progress in arc plasma technology in China, with a particular focus on two key aspects: structure and control methods. The objective is to provide a novel perspective and insights for the international academic community, thereby driving further development in the field of arc plasma power sources. Furthermore, it discusses and analyzes the future development of arc plasma power sources with a focus on achieving higher system reliability, increased power output, environmentally friendly solutions, and intelligent operation.

Large scale integration of photonics devices requires low power consumption devices. In this paper, we demonstrate a low power consumption polymer/silica hybrid thermo-optic switch based on racetrack resonator. With the high index-contrast between SU-8 core, silica buffer and PMMA cladding, a compact racetrack resonator with a small bending radius of 120 μm and a coupling length of 1765 μm is fabricated through simple and low-cost contact lithography technology. An extinction ratio of 16.83 dB is achieved while the power consumption applied is 14.69 mW. The energy efficiency of the switch is 12.07 pm/mW. The rise/fall time the switch is 174 μs/182 μs.

The surface defect detection of automobile pipe joints based on computer vision faces technical challenges. The tiny-sized and smooth surfaces with processing textures will undermine the defect detection accuracy. In order to solve this problem, a new method was proposed, which combines wavelet decomposition and reconstruction with the canny operator to detect defects, and then uses the multi-channel fusion convolutional neural network to identify the types of defects. Firstly, illumination compensation technology is used to obtain a more uniform gray distribution of the original image. Then, the wavelet decomposition and reconstruction are used to remove noises and processing textures. Furthermore, the defect regions are segmented using the canny operator and hole filling from the image. Finally, the multi-channel fusion convolutional neural network of decision-level is used to identify the surface defect types. This method provides an idea for the surface defects detection of automobile pipe joints with serious interference, such as smooth surface, random noises, and processing textures. The experimental results reveal that the method can effectively eliminate the influence of uneven illumination, random noises, and processing textures and achieve high defect classification accuracy.

(6+1)×1 pump and signal combiner with high power handing capability, negligible beam quality degradation and bi-directional signal transmission ability has been developed by employing signal-fiber non-taper fabrication approach and in-line active splicing process. The influences of lateral core offset and angular misalignment between the signal input and output fibers in combiner have been studied theoretically, showing that M2 factor rather than signal efficiency is more suitable to be the evaluation criterion. Three splicing methods have been compared experimentally, revealing that the M2-based active splicing is the most effective to preserve beam quality and obtain good efficiency simultaneously. Signal laser transmitted backward through 25-μm fiber core has shown efficiency of 95.5% and M2 factor deterioration no more than 10% at 1 kW of power. Combined pump from laser diodes into 400-μm fiber clad has reached beyond 5 kW of power at efficiency of 98.2%, with the maximum temperature on combiner being 70.5°C, which was tested in the condition without cooling on fiber pigtails and with flat-cleaved fiber termination aggregating thermal load. Both results indicate the effectiveness of the proposed fabrication technique to make bi-directional (6+1)×1 combiner for high power and superior beam quality fiber laser systems.

Objective: With the rapid growth of high-speed deep-tissue imaging in biomedical research, there is an urgent need to develop a robust and effective denoising method to retain morphological features for further texture analysis and segmentation. Conventional denoising filters and models can easily suppress the perturbative noise in high-contrast images; however, for low photon budget multiphoton images, a high detector gain will not only boost the signals but also bring significant background noise. In such a stochastic resonance imaging regime, subthreshold signals may be detectable with the help of noise, meaning that a denoising filter capable of removing noise without sacrificing important cellular features, such as cell boundaries, is desirable. Method: We propose a convolutional neural network-based denoising autoencoder method — a fully convolutional deep denoising autoencoder (DDAE) — to improve the quality of three-photon fluorescence (3PF) and third-harmonic generation (THG) microscopy images. Results: The average of 200 acquired images of a given location served as the low-noise answer for the DDAE training. Compared with other conventional denoising methods, our DDAE model shows a better signal-to-noise ratio (28.86 and 21.66 for 3PF and THG, respectively), structural similarity (0.89 and 0.70 for 3PF and THG, respectively), and preservation of the nuclear or cellular boundaries (F1-score of 0.662 and 0.736 for 3PF and THG, respectively). It shows that DDAE is a better trade-off approach between structural similarity and preserving signal regions. Conclusions: The results of this study validate the effectiveness of the DDAE system in boundary-preserved image denoising. Clinical Impact: The proposed deep denoising system can enhance the quality of microscopic images and effectively support clinical evaluation and assessment.

Detecting defects, which is a branch of target detection in the field of computer vision, is widely used in factory production. To solve the problems in existing detection algorithms that relate to their insensitivity to large or medium defect targets on bearing covers, their difficulty in detecting subtle defects effectively and their lack of real-time detection, in this work, we establish a large-scale bearing-cover defect dataset and propose an improved YOLOv3 network model. The proposed model is divided into four submodels: the bottleneck attention network (BNA-Net), the attention prediction subnet model, the defect localization subnet model, and the large-size output feature branch. To test the generality, robustness and practicability of the new model, we design a comparative experiment under abnormal illumination conditions. We design an ablation experiment to verify the validity of the proposed submodules. The experimental results show that our model solves the problem of the YOLOv3 algorithm's insensitivity to medium or large targets and satisfies real-time detection conditions. The mAP result is 69.74%, which is 16.31%, 13.4%, 13%, 10.9%, and 7.2% more than that of YOLOv3, EfficientDet-D2, YOLOv5, YOLOv4, and PP-YOLO, respectively.

The shock wave phenomenon has been very common in the high-pressure fuel spray. In this paper, the effect of the shock waves on the spray development and the variations of the flow field parameters were investigated using the Schlieren imaging coupled with a numerical simulation. Results showed that the shock wave contributed to the increase in the spray tip penetration. The mixing characteristics of the spray also improved in the shock-wave state. Numerical simulations were used to investigate the flow characteristics of the shock wave and the effect of the shock wave on the flow field parameters. Results showed that the simulated shock wave characteristic parameters were consistent with the experimental data. In addition, the flow field parameters were affected by the shock wave propagation. The maximum density ratio, pressure ratio and temperature ratio after and before the shock wave are 2.46, 2.01 and 1.15, respectively, under the fuel injection pressure of 320MPa.

We propose and demonstrate a low-cost and simple fiber optic cantilever accelerometer with a spherical tip based on Gaussian beam focusing. The accelerometer consists of ceramic ferrule, ceramic sleeve, receiving fiber and emitting fiber, where both fibers are single mode fiber, the ferrule and sleeve have characteristics of high precision, which reduce the difficulty of optical alignment. The end of the emitting fiber is made into a spherical tip for focusing the Gaussian beam to improve sensitivity. When the accelerometer is in operation, the emitting fiber acts as a cantilever beam, the acceleration can be measured by detecting the transmission power. Further, our experimental results show that the spherical fiber tip can improve the acceleration sensitivity by 67% over 10 Hz–1000 Hz without reducing the working bandwidth. In addition, it is found that the fiber accelerometer has a high signal-to-noise ratio (SNR) up to 60 dB, and a low harmonic distortion of better than -30 dB, rendering a quasi-8-shaped directionality at the working frequency ranging from 10 Hz to 1200 Hz. This clever sensor structure may have potentials for developing high-performance and cost-effective accelerometers and hydrophones.

Some recent studies have suggested using Generative Adversarial Network (GAN) for numeric data over-sampling, which is to generate data for completing the imbalanced numeric data. Compared with the conventional over-sampling methods, taken SMOTE as an example, the recently-proposed GAN schemes fail to generate distinguishable augmentation results for classifiers. In this paper, we discuss the reason for such failures, based on which we further study the restrained conditions between $G$ and $D$ theoretically, and propose a quantitative indicator of the restrained structure, called Similarity of the Restrained Condition (SRC) to measure the restrained conditions. Practically, we propose several candidate solutions, which are isomorphic (IWGAN) mirror (MWGAN) and self-symmetric WGAN (SWGAN) for restrained conditions. Besides, the restrained WGANs enhance the classification performance in AUC on five classifiers compared with the original data as the baseline, conventional SMOTE, and other GANs add up to 20 groups of experiments in four datasets. The restrained WGANs outperform all others in 17/20 groups, among which IWGAN accounted for 15/17 groups and the SRC is an effective measure in evaluating the restraints so that further GAN structures with G-D restrains could be designed on SRC. Multidimensional scaling (MDS) is introduced to eliminate the impact of datasets and evaluation of the AUC in a composite index and IWGAN decreases the MDS distance by 20% to 40%. Moreover, the convergence speed of IWGAN is increased, and the initial error of loss function is reduced.

Cardiovascular disease is a severe threat to human health. The assessment of the degree of cardiovascular stenosis is very important for the diagnosis of cardiovascular disease. The segmentation of coronary angiographic images is the basis of analyzing the degree of stenosis. In this paper, a method that evaluates explicitly the degree of stenosis of blood vessels of interest is proposed. This method first extracts the vessels of interest by interactive segmentation, then thins it, and then calculates the diameter by edge intersection method. The average Dice Similarity Coefficient (DSC) value of the proposed segmentation method exceeds 92%, and the average Jaccard Similarity (JAC) value is over 86%. The area under the ROC curve value of the method is larger than the u-net method and the multiscale Hessian method. The results indicate that the interactive method can have good segmentation results and meet the general segmentation requirements. The qualitative and quantitative evaluation of the diameter measurement effect also shows that the effect of the corresponding diameter measurement method can reflect the change of the thickness of the blood vessel and basically meet the needs of clinical use.

For storage and recovery requirements on large-scale seismic waveform data of the National Earthquake Data Backup Center (NEDBC), a distributed cluster processing model based on Kafka message queues is designed to optimize the inbound efficiency of seismic waveform data stored in HBase at NEDBC. Firstly, compare the characteristics of big data storage architectures with that of traditional disk array storage architectures. Secondly, realize seismic waveform data analysis and periodic truncation, and write HBase in NoSQL record form through Spark Streaming cluster. Finally, compare and test the read/write performance of the data processing process of the proposed big data platform with that of traditional storage architectures. Results show that the seismic waveform data processing architecture based on Kafka designed and implemented in this paper has a higher read/write speed than the traditional architecture on the basis of the redundancy capability of NEDBC data backup, which verifies the validity and practicability of the proposed approach.

With the development of intelligence in air confrontation, the demand for cooperative engagement of manned/unmanned aerial vehicle (MAV/UAV) is becoming more intense. Deep reinforcement learning (DRL), which combines the abstract representation capability of deep learning (DL) and the optimal decision-making and control capability of reinforcement learning (RL), is an appropriate application for dealing with this problem. In the case of continuous action space, the dynamics model of UAV and the basic structure of one of the most popular DRL methods called deep deterministic policy gradient (DDPG) are built firstly. To establish the framework of intelligent decision-making of MAV/UAV, typical intentions including Head-on attack, Fleeing, Pursuing and Energy-storing, corresponding to four optimization models, are introduced secondly. Then the neural network is trained by means of reconstructing the replay buffer of DDPG algorithm. Finally, simulation results show that UAV is able to learn intelligent decision-making throughout the intention guiding of MAV. Compared with original DDPG algorithm, the improved method can achieve a better performance in convergence and stability. Furthermore, the level of intelligent decision-making in air confrontation can be improved by self-learning.

104 W supercontinuum (SC) covering 370 nm-2400 nm is reported. High power visible SC is generated by pumping cascaded photonic crystal fibers (PCF) with 1064 nm picosecond pulse. Cascaded PCFs is fabricated by fusion splicing two different PCFs together with a coupling efficiency of 75%. For the purpose of avoiding fiber fuse phenomenon in PCF, the discharge time and intensity of fusion splicer should be precisely controlled. Two components of cascaded PCFs are pumped separately by 1064 nm picosecond pulse for comparing with the impact of cascaded PCFs. As a result of pumping cascaded PCFs, 104 W SC covering 370 nm-2400 nm is generated. Visible spectra (380 nm-780 nm) accounts for 50.4% (52.4 W) by numerical integration. To best of our knowledge, it's the first time that a visible SC with over a hundred watt output and 370 nm short-wave edge is obtained experimentally.

Theoretical studies of ultra-thin silicon solar cells with cylindrical, conical and parabolic surface nanostructures inherited from natural self-assembled anodic alumina oxide (NSA-AAO) were performed by finite-difference time-domain (FDTD) method. All nanostructured solar cells obtained an optimized efficiency enhancement as high as more than 33% comparing with that of the anti-reflective (AR) one. Numerical results reveal that the range of efficient structural parameters for the nanostructured (e.g., cylindrical) solar cell can be effectively enlarged as the period of the nanostructure changes from 0.1 μm to 0.5 μm. Moreover, the improvements of absorption photocurrent density (Jph) in conical and parabolic nanostructured solar cells are comparable with the cylindrical nanostructured one but less sensitive to the fill factor and structural height in the whole simulation region of 0.1-0.9 and 0-0.25 μm, respectively. Equivalent refractive index models were used to analysis the antireflection performance of surface nanostructures from the point of view of sidewall profiles. Resonance modes induced through nanostructures have greatly improved the absorptance of solar cells in broadening wavelength bands which consequently raised the Jph. This study serves as a way for the practical design and application of AAO nanostructure based high-efficiency ultra-thin solar cells.

Functional overlaps and conflicts between Electronic Power Steering (EPS) and Torque Vectoring Control (TVC) for Distributed Drive Electric Vehicles (DDEVs) are issues that need to be addressed urgently. This paper deals with the interaction between EPS and TVC from the view of the influence mechanism and control objectives. The disturbance steering torque caused by the differential torque is estimated and a unified steering characteristic is proposed. To validate the analysis results, a torque compensation strategy is modified based on the exiting control system. And a robust vehicle state estimator is built thanks to the aligning torque estimation to provide the necessary information. Typical ground test results show that the torque oscillation on the steering wheel is suppressed. A faster and more linear steering response can be seen, which means that the proposed disturbance compensation strategy can comfort the contradiction between EPS and TVC, also improve the handling performance of the vehicle.

Dance emotion recognition technology is of great significance for the digitalization, virtual performance, inheritance and protection of folk dance. Based on the mechanism that emotion expression in dance performance can be fully expressed through the strength and rhythm of dance movements, a novel dance emotion expression method is proposed to train hybrid deep learning neural network, to effectively identify the seven basic dance emotions of fear, anger, boredom, excitement, joy, relaxation and sadness. First, in order to fully express the emotions contained in the dance movements, this paper defines a dance emotion expression method through Laban Movement Analysis (LMA) method, which includes the characteristic parameters of the three aspects of body structure, spatial orientation and force effect, and converts the original dance movement data into three characteristic expression parameters to obtain dance emotion data. Then, Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) hybrid neural network models are used to test and train dance emotion data. Finally, in order to verify the applicability of the CNN-LSTM model, decision tree, random forest, CNN and LSTM are established and compared for accuracy. The results show that it is feasible to identify dance emotion from the perspective of dance movement, and the CNN-LSTM model is of high accuracy.

This paper presents a nonlinear error feedback controller for a pneumatic soft bionic fin based on an extended state observer. Manufacturing process is shown for a bionic stingray with the pneumatic soft bionic fin. A test experiment is carried out for a driving part in the pneumatic soft bionic fin by upper chamber inflated and lower chamber inflated with different internal pressures. A dynamic nonlinear system is established for the pneumatic soft bionic fin by test experiment results and kinematic modeling. Total disturbances of the dynamic nonlinear system are estimated and compensated by the extended state observer and the nonlinear error feedback controller, respectively. Moreover, both the convergence of the extended state observer and stability of the dynamic nonlinear system are proved by Lyapunov approach. Finally, simulation results are shown to verify the effectiveness of the proposed control method for the pneumatic soft bionic fin.

This paper proposes a functional safety and information security protection mechanism based on blockchain technology. The design of the basic level and integration level blockchain structure of CPS distributed architecture and related functional safety and information security measures are introduced. An effective communication judgment mechanism based on functional safety error threshold is proposed, which is stored and judged by smart contract. And a refund transaction with a clock is proposed to ensure the effective execution of the functional safety error threshold mechanism. The article takes cyber physical machine tool system as an example to describe the SIL judgment method of CPS physical equipment and functional safety loop and combines SIL with fault diagnosis and risk protection. Finally, the rationality of our proposed mechanism is proved by information security, functional security, real-time and maintainability.

A stress-strength model usually has more than one failure mode since the component suffers at least two types of stresses, complicating the expression of the likelihood function and increasing the computational complexity of the parameter estimation for general distributions (non-exponential distributions). A phase-type distribution (also known as a PH distribution) is dense and has closure properties, which makes it suitable to reduce the computational complexity of the stress-strength model. The traditional expectation-maximization (EM) method for estimating the parameters of the PH distribution cannot be used directly when the strength changes over time since the PH distribution is a continuous-time Markov process that must satisfy the relevant properties of the infinitesimal generator in the Markov state-space. Therefore, a parameter estimation method based on extending the Markov state-space with variable transition rates for the stress-strength model is proposed. Both failure and censored samples are considered. First, the stress-strength model based on the PH distribution is briefly introduced, and the likelihood functions for different failure modes are derived. Subsequently, the principle of the method is described in detail, the derivation process of the relevant equations is provided, and the limitations of the method are discussed. The performance of the method is evaluated using two simulation cases.

The traditional shock model generally describes the magnitude of the cumulative damage caused by a random shock sequence and compares the magnitude with a predetermined threshold to obtain the failure time of a component. There are two limitations in this kind of models in practice: First, the statistical characteristics of the damage due to a single shock may be difficult to obtain, which means the magnitude of the damage may not be described by an appropriate distribution; Second, the cumulative shock magnitude may be difficult to measure, or it may be difficult for a failure mode to be described by a threshold, meaning that the magnitude of the damage and the threshold may not be compared with each other. Considering both failure and censored samples, a reliability modeling method is proposed in this work to address the above problems. The shock model is first established by using both continuous and discrete phase-type (PH) distributions. Then the parameter estimation method of the shock model is derived based on EM method and the identifiability of the parameters in PH distributions is also given. Finally, the adaptability of the model is analyzed using three different types of simulation cases.

This paper proposes a new methodology for designing wideband Class J Doherty power amplifiers (DPAs). The paper begins by presenting a network analysis that leads to a set of design equations to govern the synthesis of the combiner network parameters that satisfy Class J load requirements. The combiner network produces a complex-to-complex Doherty load modulation capable of avoiding potential transistor voltage clipping due to a mismatch between the fundamental and harmonic impedances. Consequently, this method improves the high power AM-AM characteristic and reduces the DPA's peak power variation versus frequency. A wideband Class J DPA was designed as a proof-of-concept demonstrator to operate from 2.7 GHz to 4.3 GHz. Under continuous wave stimuli, over the entire band, the measured AM-AM distortion in the Doherty region of the fabricated DPA was found to be lower than 1.2 dB with a relatively constant output power of 38.9 ± 0.3 dBm at saturation. Moreover, good drain efficiencies of about 42% and 54% were recorded at 6-dB output back-off and saturation powers, respectively. In addition, the linearizability of the fabricated DPA was confirmed under both intra- and inter-band carrier-aggregated signal stimuli. In fact, measurements using an 80 MHz inter-band carrier-aggregated signal revealed that the proposed DPA could deliver an adjacent channel leakage ratio of better than -48 dBc after digital pre-distortion with a good average drain efficiency of 45% - 49% over the entire band of interest.

Underground target detection technology has been widely used in urban construction and resource exploration. With the development of industrial modernization, the demand for underground target detection is becoming more specific, such as the material and shape of underground targets. Therefore, it is necessary to classify the properties of underground targets. In this paper, sensitivity analysis was performed on the spheroid model and the approximate forward model at first, and the influence of the target properties on the model output is obtained. Secondly, we utilized the fitting algorithm to obtain the model parameters of the simulation data (model response of targets with varying shapes and materials), and analyzed the influence of the fitting algorithm on the classification results at different SNR. Finally, eight machine learning algorithms: support vector machine(SVM), neural network(NN), quadratic discriminant analysis (QDA), Gaussian process (GP), decision tree (DT), random forest (RF) and AdaBoost were used in this study to compare the obtained results. From the above analysis, we found that the shape (radius) have a greater influence on the model than the material (permeability) in the spheroid model. According to the approximate forward model, we found that it is not feasible to classify targets when the orientation is unknown. The influence of the fitting algorithm on the classification performances is related to the noise level. The obtained results using neural network demonstrated that the proposed method outperformed in material-based classification and shape-based classification. In the material-based classification, the classifier generally has a weaker ability to distinguish between permeable materials.

This paper investigates a finite-time formation control problem for multiple networked quadrotors. A novel distributed control approach is presented under the leader-follower formation framework, and the approach is developed based on the finite-time Lyapunov theory and the homogeneous system theory such that all quadrotors form and maintain a desired geometric pattern within finite time while tracking a reference trajectory. The designed control law is composed of a dynamic observer, a position controller and an attitude controller, in which the observer is adopted to provide estimates of the leader quadrotor information for each follower quadrotor, and the controllers are in a cascade structure. It is shown that the finite-time leader-follower formation of a group of quadrotors can be achieved via the distributed control approach, and the cascade control architecture conforms to quadrotor dynamic characteristics. The constructive procedures on how to synthesize such a control law are also given. The effectiveness of the proposed control approach is verified by the simulation.

The deep convolutional neural networks (CNNs) have been shown excellent performances for image denoising. However, the denoising CNN model trained with a specific noise level cannot deal with the images which have spatiotemporally variant random noise and low signal-to-noise ratio (SNR), such as seismic images. To this end, we propose a patch-based denoising CNN method, namely PDCNN. Specifically, we cluster the overlapping patches of noisy image into K classes where the image patches have close noise levels in each class, and then choose a suitable model for denoising the corresponding class from a series of well-trained CNN models. By embodying the structural statistics, we propose a CNN model selection criterion with a structural-dependent parameter. In contrast to the manual model selection process, the more accurate CNN model is chosen automatically and effectively. The capability of the PDCNN is demonstrated on synthetic and field seismic images. Experimental results show that the proposed method largely benefits from using multiple CNN models to jointly denoise, and leads to the satisfactory denoising performance in spatiotemporally variant seismic random noise reduction and structural signal preservation.

It is important to find an effective method for power quality (PQ) disturbance recognition under the challenges of increasing power system pollution. This paper proposes a PQ disturbance signal recognition method based on Multiresolution S transform (MST) and decision tree (DT). For improving recognition accuracy, adjustment factors are introduced to obtain a controllable time-frequency resolution. On this basis, five feature statistics are obtained to quantitatively reflect the characteristics of the analyzed power quality disturbance signals, which is less than the traditional S-transform-based method. As the proposed methodology can effectively identify the PQ disturbances, the efficiency of the DT classifier could be guaranteed. In addition, the noise impacts are also taken into consideration, and 16 types of noisy PQ signals with a signal-to-noise ratio (SNR) scoping from 30 to 50 dB are used as the analyzed dataset. Finally, a comparison between the proposed method and other popular recognition algorithms is conducted. The experimental results demonstrate that the proposed method is effective in terms of detection accuracy, especially for combined PQ disturbances.

China's soybean supply and demand are seriously imbalanced. It is crucial to improve the level of soybean breeding. Hundred-grain weight is one of the most essential phenotypic parameters for crop breeding. Accurate soybean seed counting is a key step for 100-grain weight. There are several seed counting methods, which have their own limitations one way or the other. Among these, manual counting is time-consuming, electronic automatic seed counter devices are expensive and their counting speed is very slow, and the traditional digital image processing techniques are not suitable for seed counting based on individual pod images. This paper attempted to develop a method that would combine the density estimation-based methods and the convolution neural network (CNN)-based methods to accurately estimate the seed count from an individual soybean pod image with a single perspective. In this paper, we first introduced a new large-scale seed counting dataset, named Soybean-pod. The dataset contains 500 annotated pod images with a total of 32 126 seeds and is the largest annotated dataset for soybean seed counting so far. Simultaneously, we used annotation information to generate a ground-truth density map by convolving a Gaussian kernel and, then, devised a simple but effective method that would elucidate pod images to a seed density map using a two-column CNN (TCNN) and thus accomplish seed counting ultimately. We conducted relevant experiments from three aspects on the new dataset to verify the effectiveness of our model and method, which provided 13.21 mean absolute error (MAE) and 17.62 mean squared error (mse). In addition, our research results showed that deep learning techniques can be easily adapted to precision tasks for plant phenotyping and breeding purposes.

To improve the efficiency of deep reinforcement learning (DRL)-based methods for robotic trajectory planning in the unstructured working environment with obstacles. Different from the traditional sparse reward function, this paper presents two brand-new dense reward functions. First, the azimuth reward function is proposed to accelerate the learning process locally with a more reasonable trajectory by modeling the position and orientation constraints, which can reduce the blindness of exploration dramatically. To further improve the efficiency, a reward function at subtask-level is proposed to provide global guidance for the agent in the DRL. The subtask-level reward function is designed under the assumption that the task can be divided into several subtasks, which reduces the invalid exploration greatly. The extensive experiments show that the proposed reward functions are able to improve the convergence rate by up to three times with the state-of-the-art DRL methods. The percentage increase in convergence means is 2.25%-13.22% and the percentage decreases with respect to standard deviation by 10.8%-74.5%.

This paper presents two novel neural network models for radio-frequency (RF) power amplifiers (PAs): vector decomposed time-delay neural network (VDTDNN) model and augmented vector decomposed time-delay neural network (AVDTDNN) model. In contrast to conventional neural network-based models, VDTDNN and AVDTDNN comply with the physical characteristics of RF PAs by employing carefully designed network structures. In particular, the nonlinear operations are conducted only on the magnitude of the input signals, while the phase information is recovered with the linear weighting. Linear terms with shortcut connection, as well as high-order terms, can be used to further boost the modeling performance. The complexity analysis shows that the proposed models have significantly lower complexity than the existing neural network models. A wideband GaN RF PA excited by the 40- and 60-MHz OFDM signals were employed to evaluate the performance. The extensive experimental results reveal that the proposed VDTDNN and AVDTDNN models can achieve better linearization performance with lower computational complexity compared with the existing neural network-based models.

We propose and demonstrate a quasi-distributed fiber sensor based on cascaded microfiber Fabry-Perot interferometers (MFPIs) for simultaneous refractive index (RI) and temperature measurement. By employing MFPI that is fabricated by taper-drawing the center of a uniform fiber Bragg grating on standard fiber into a section of microfiber, dual-parameters including RI and temperature can be detected through demodulating envelop and resonant wavelength of the reflection spectrum of the MFPI. Then, wavelength-division-multiplexing is applied to realize quasi-distributed dual-parameter sensing by using cascaded MFPIs with different Bragg wavelengths. A prototype sensor system with five cascaded MFPIs is constructed to experimentally demonstrate the sensing performance. The quasi-distributed dual-parameter sensing system has great significance on monitoring gradient parameter-variations in chemical and biological sensing applications.

Characterizing acoustic waveguide channels is becoming important for the development of communication and signal processing applications across diverse fields ranging from urban water supply systems to oil and gas distribution pipeline networks. These applications include sonar and transmission systems in support of leak detection, blockage location, sensing, monitoring, and signaling for example. In this paper, we provide experimental results and models for the wideband channel characterization of acoustic waveguides formed from gas and water pipelines over the 1-50 kHz frequency band. Experimental results are provided for two straight pipe systems comprising an acrylic pipe filled with air and a steel pipe filled with water. A mode-based analytical model for predicting acoustic wave propagation in rigid and elastic pipes is proposed with deterministic and stochastic characteristics both considered. Good matching is demonstrated between the model predictions and experimental results in terms of dispersion curves, channel spectrograms, and delay spread. A key finding is that acoustic waveguides filled with water should be treated as elastic pipes, and they have significantly different characteristics from those filled with gas, which can usually be treated as rigid pipes. Furthermore for the steel-water waveguide pipeline, link budget calculations and noise power spectral density measurements reveal that a communication range of more than 50 m can be obtained.

In this paper, we have designed an ultra-wideband electromagnetic (EM) absorber based on the concept of metasurface Salisbury screen (MSS), which features low profile, light weight, simple configuration, and robust angular performance. The metasurface with extremely simple patch pattern is utilized to generate diverse controllable reflection phases, in place of the non-dispersive metallic plate used in conventional absorbers, thus achieving a multi-octave ultra-wideband EM wave absorption. Equivalent circuit model is established to analyze the performance of the MSS elements, and then the genetic algorithm and simulated annealing algorithm are employed to optimize the MSS element geometries and their spatial distribution. The proposed and fabricated MSS, with a polarization-insensitive absorption over 88% from 3.74 to 18.5 GHz verified by experiments, shows a considerable bandwidth improvement compared with the conventional Salisbury screen of same thickness which has 88% absorption band from 4.8 to 11.5 GHz. Furthermore, the MSS can still provide ultra-wideband absorption with high efficiency for large incident angle, for example, higher than 82% for 45° incidence. The proposed concept could provide opportunities for flexibly designing ultra-wideband EM absorbers, exhibiting promising potentials for many practical applications, such as electromagnetic compatibility, stealth technique, and so on.

A load balancing mechanism can adjust the load distribution among access points (APs) and improve resource utilization for dense wireless local area networks (WLANs). In this paper, we propose a semi-matching-based load balancing scheme for the IEEE 802.11 dense WLANs. The proposed scheme runs in a centralized controller. The controller judges whether the load is unevenly distributed according to the collected channel busy time ratio information of the entire network, and triggers the load balancing mechanism accordingly. In order to realize load balancing among APs and maximize the overall network throughput, we model the station to AP association problem as a weighted bipartite graph matching problem and find the optimal semi-matching using the Kuhn-Munkres (K-M) algorithm. Simulation results show that the proposed scheme achieves performance improvement comparing with traditional schemes.

This paper introduces the POMA-C (Partial Observable Model for Anesthesia Control) framework, developed to address the challenge of anesthesia management in environments with incomplete physiological monitoring, such as low-resource settings where critical indicators like the Bispectral Index (BIS) are often unavailable. Unlike traditional methods that rely on fully observable data, POMA-C frames the problem of anesthesia control under incomplete observability within a Partially Observable Markov Decision Process (POMDP), enabling the precise control of anesthesia despite missing data. By establishing a formal correspondence between the anesthesia control process and POMDP, this framework provides a theoretical foundation for modeling anesthesia control in uncertain environments. The framework employs the POMCPOW (Partially Observable Monte Carlo Planning with Observation Weighting) algorithm, which integrates Monte Carlo Tree Search (MCTS) and particle filtering to estimate the patient’s true physiological state and guide optimal anesthetic decisions. Through comprehensive ablation experiments—where key observation dimensions are systematically reduced to simulate missing data—POMA-C demonstrates significantly higher decision accuracy and cumulative reward optimization compared to methods like Q-learning and human expertise, even in data-constrained environments. This work not only provides a robust solution for anesthesia control under incomplete observability but also bridges the gap between MDP and POMDP models, offering a foundation for future research in automated anesthesia management.

Feature selection is an important preprocessing step in machine learning to remove irrelevant and redundant features. Due to its ability to effectively maintain the discriminability of extracted features, Trace Ratio Linear Discriminant Analysis (TR-LDA) has become the foundation for many feature selection algorithms. As is known, TR-LDA is a challenging problem to solve because of its trace-ratio form, and it also faces the scale invariance problem. These two drawbacks of TR-LDA significantly reduce the performance of feature selection algorithms based on it. To overcome these drawbacks, this paper proposes the sparse LDA with constant between-class distance (SLDA-CBD) model to select relavant features. This model first transforms TR-LDA into a non-trace ratio problem with a constant between-class distance constraint, and then imposes row constraints on the projection matrix to implement feature selection. Since the SLDA-CBD model is rooted in TR-LDA, it ensures the discriminative performance of the selected features. The constant between-class distance constraint successfully avoids the scale invariance problem. Additionally, due to the non-trace ratio form of the SLDA-CBD model, it is easily solvable. The experimental results show that the proposed method has better performance compared to the baseline and six state-of-the-art relative methods, with improvements of over 1% on image datasets and over 2% on video datasets in most cases, while also demonstrating high stability, proving its effectiveness and advantage in practical applications.