Search

Numerous advanced and lightweight signal processing methods have been presented for single-channel speech enhancement (SE). It is imperative to carefully explore how to efficiently combine, integrate, and balance these methods. This paper proposes a more effective and less resource-intensive SE system, focused on the integration and adaptation of several approaches, especially the temporal cepstrum smoothing (TCS). First, a more robust fundamental frequency estimator is employed within TCS, mitigating the performance limitations caused by the inaccuracy of the original estimator. Additionally, a harmonic enhancement mechanism is introduced, effectively recovering the weak harmonic components. By incorporation of the modified TCS in the a posteriori speech presence probability estimation, the unbiased minimum mean square error noise power spectral density estimator can be refined. The modified TCS is also utilized for the a priori signal-to-noise ratio estimation. Moreover, this paper enhances the log-spectral amplitude estimator by applying both super-Gaussian speech priors and speech presence uncertainty for further improvement. Experimental evaluations demonstrate that the proposed method yields an improvement in speech quality while maintaining modest computational and storage requirements. Furthermore, the proposed system exhibits comparable performance to several baseline systems based on lightweight deep neural networks. [ABSTRACT FROM AUTHOR]

Traditional turbulence models suffer from low accuracy and weak applicability when predicting complex separated flows, such as those that occur in shock boundary layers. To overcome this problem, the present paper considers a cavity-ramp structure and calibrates the turbulence model parameters using a deep neural network (DNN) surrogate model and a genetic algorithm (GA). The non-intrusive polynomial chaos expansion method is used to quantify the uncertainty of the shear stress transport (SST) turbulence model parameters and determine the effects of these parameters on the wall pressure, allowing suitable feature identification parameters to be selected for the DNN turbulence surrogate model. The DNN is compared with the traditional polynomial chaos expansion method, and the results highlight the advantages of using the DNN method to construct the surrogate model. Finally, a GA is used to optimize and calibrate the SST turbulence model parameters based on the surrogate model and experimental data. Experimental results show that the DNN turbulence surrogate model is highly accurate, with the predicted wall pressure, achieving a coefficient of determination above 0.998. The DNN has higher precision, stronger feature extraction ability, and faster prediction times than the traditional polynomial chaos expansion method. The calibrated SST turbulence model produces wall pressures that are close to the experimental data, verifying the feasibility of the proposed method. It is expected that the approach proposed in this paper will improve the calculation accuracy of the SST turbulence model. [ABSTRACT FROM AUTHOR]

This paper presents a free-field method for inverse estimation of acoustic porous material parameters from sound pressure measurements above small rectangular samples. The finite sample effect, the spherical propagation of the sound field, and a potential lateral material reaction are considered. Using an extensive series of systematically varied finite element simulations, neural network models are developed to replace computationally expensive simulations as a forward model for the calculation of the complex sound pressure above small samples in the inverse optimization. The method is experimentally validated using various porous material samples. The results show that the influence of the finite sample size is successfully removed and thus, the acoustic properties of the materials can be estimated from the determined porous parameters with high accuracy, even based on a single sound pressure measurement over small samples with pronounced edge diffraction. The poroacoustic parameters hence derived can be used directly, e.g., in simulation applications, or to calculate complex surface impedances or absorption coefficients. [ABSTRACT FROM AUTHOR]

Noise is a consistent problem for x-ray transmission images of High-Energy-Density (HED) experiments because it can significantly affect the accuracy of inferring quantitative physical properties from these images. We consider experiments that use x-ray area backlighting to image a thin layer of opaque material within a physics package to observe its hydrodynamic evolution. The spatial variance of the x-ray transmission across the system due to changing opacity serves as an analog for measuring density in this evolving layer. The noise in these images adds nonphysical variations in measured intensity, which can significantly reduce the accuracy of our inferred densities, particularly at small spatial scales. Denoising these images is thus necessary to improve our quantitative analysis, but any denoising method also affects the underlying information in the image. In this paper, we present a method for denoising HED x-ray images via a deep convolutional neural network model with a modified DenseNet architecture. In our denoising framework, we estimate the noise present in the real (data) images of interest and apply the inferred noise distribution to a set of natural images. These synthetic noisy images are then used to train a neural network model to recognize and remove noise of that character. We show that our trained denoiser network significantly reduces the noise in our experimental images while retaining important physical features. [ABSTRACT FROM AUTHOR]

Gradient-based aerodynamic optimization methods involve complex gradient calculations with high computational costs. To address these problems, this paper proposes a rapid deep learning-based method for predicting the aerodynamic force gradients and establishes an aerodynamic optimization framework. A deep neural network is used to determine the latent mapping relationship between the near-wall flow field information and the aerodynamic force gradients, and the resulting gradient prediction model is integrated into an aerodynamic optimization platform. Combining grid deformation techniques with gradient-based optimization enables the aerodynamic optimization of airfoils. The deep neural network model can rapidly predict gradients that are basically consistent with those from the traditional discrete adjoint method, and decouples the gradient computation from the flow field solution process. The proposed method is validated through the aerodynamic optimization of the airfoil. Under various optimization objectives and initial conditions, the deep neural network model significantly improves the aerodynamic performance of the airfoil and substantially enhances the efficiency of optimization compared with traditional methods. [ABSTRACT FROM AUTHOR]

Artificial neural network is considered to be one of the effective ways to enable soft robots to achieve high-performance control due to their significant advantages, such as massively parallel processing and distributed storage of information, adaptivity, and fault tolerance. Artificial neural networks are composed of microelectronic components connected together, of which the most basic units are artificial neural synaptic units, such as atomic switches, memristors, and synaptic transistors. This paper first introduces the research status of soft robots and artificial neural synapses, predicts the demand of soft robots for artificial neural synapses, summarizes the difficulties and problems that may be encountered in the application of artificial neural synapses to soft robots, and finally points out the importance and feasibility of artificial neural synapses in the research and development of soft robots. [ABSTRACT FROM AUTHOR]

This document is a preface to a special collection in honor of Freeman Dyson, a renowned mathematician and theoretical physicist. Dyson made significant contributions to various fields, including quantum electrodynamics, statistical mechanics, many-body quantum mechanics, stability of matter, quantum entropy, trace inequalities, and random matrix theory. The collection includes papers related to these topics, written by collaborators and former colleagues of Dyson. Dyson received numerous prestigious prizes and awards throughout his career. The preface also mentions a paper in the collection that applies mathematics to neuroscience, which Dyson would have appreciated. [Extracted from the article]

Data-driven approaches have made preliminary inroads into the area of transition–turbulence modeling, but are still in their infancy with regard to widespread industrial adoption. This paper establishes an artificial neural network (ANN)-based transition model to enhance the capacity of capturing the crossflow (CF) transition phenomena, which are frequently identified over a wide range of aerodynamic problems. By taking a new CF-extended shear stress transport (SST) transition-predictive (SST-γ) model as the baseline, a mapping from mean flow variables to transition intermittency factor (γ) is constructed by ANN algorithm at various Mach and Reynolds numbers of an infinite swept wing. Generalizability of the resulting ANN-based (SST- γ ANN ) model is fully validated in the same infinite swept wing, an inclined 6:1 prolate spheroid, and a finite swept wing in extensive experiment regimes, together with two effective a priori analysis strategies. Furthermore, the calculation efficiency, grid dependence, and performance of the present model in non-typical transitional flow are also assessed to inspect its industrial feasibility, followed by the elucidation of rationality behind the preliminary success and transferability of present framework. The results manifest that the SST- γ ANN model aligns well with the benchmark SST-γ model, and both can capture the CF transition accurately compared with their experiment counterpart, completely breaking through the disability of original SST-γ model without CF correction. In addition, good properties of efficiency, robustness, and generalizability are achieved for the ANN-alternative transition model, together with the usability of present framework across various transitional flows. [ABSTRACT FROM AUTHOR]

The rapid development of deep learning has significant implications for the advancement of computational fluid dynamics. Currently, most pixel-grid-based deep learning methods for flow field prediction exhibit significantly reduced accuracy in predicting boundary layer flows and poor adaptability to geometric shapes. Although graph neural network models for unstructured grid-based unsteady flow prediction have better geometric adaptability, these models suffer from error accumulation in long-term predictions of unsteady flows. More importantly, fully data-driven models often require extensive training time, greatly limiting the rapid update and iteration speed of deep learning models when facing more complex unsteady flows. Therefore, this paper aims to balance the demands for training overhead and prediction accuracy by integrating physical constraints based on the finite volume method into the loss function of the graph neural network. Additionally, it incorporates a twice-message aggregation mechanism inspired by the extended stencil method to enhance the unsteady flow prediction accuracy and geometric shape generalization ability of the graph neural network model on unstructured grids. We focus particularly on the model's predictive accuracy within the boundary layer. Compared to fully data-driven methods, our model achieves better predictive accuracy and geometric shape generalization ability in a shorter training time. [ABSTRACT FROM AUTHOR]

The functional networks of the human brain exhibit the structural characteristics of a scale-free topology, and these neural networks are exposed to the electromagnetic environment. In this paper, we consider the effects of magnetic induction on synchronous activity in biological neural networks, and the magnetic effect is evaluated by the four-stable discrete memristor. Based on Rulkov neurons, a scale-free neural network model is established. Using the initial value and the strength of magnetic induction as control variables, numerical simulations are carried out. The research reveals that the scale-free neural network exhibits multiple coexisting behaviors, including resting state, period-1 bursting synchronization, asynchrony, and chimera states, which are dependent on the different initial values of the multi-stable discrete memristor. In addition, we observe that the strength of magnetic induction can either enhance or weaken the synchronization in the scale-free neural network when the parameters of Rulkov neurons in the network vary. This investigation is of significant importance in understanding the adaptability of organisms to their environment. [ABSTRACT FROM AUTHOR]

Nonlinear dynamical systems with control parameters may not be well modeled by shallow neural networks. In this paper, the stable fixed-point solutions, periodic and chaotic solutions of the parameter-dependent Lorenz system are learned simultaneously via a very deep neural network. The proposed deep learning model consists of a large number of identical linear layers, which provide excellent nonlinear mapping capability. Residual connections are applied to ease the flow of information and a large training dataset is further utilized. Extensive numerical results show that the chaotic solutions can be accurately forecasted for several Lyapunov times and long-term predictions are achieved for periodic solutions. Additionally, the dynamical characteristics such as bifurcation diagrams and largest Lyapunov exponents can be well recovered from the learned solutions. Finally, the principal factors contributing to the high prediction accuracy are discussed. [ABSTRACT FROM AUTHOR]

Eddy current thermography (ECT) is a non-invasive testing method that combines electromagnetic induction and infrared thermography to identify flaws in materials that conduct electricity. However, ECT faces difficulties in accurately locating and classifying defects owing to its low signal-to-noise ratio and complex defect patterns. In this paper, we propose a new method that integrates an improved faster region-convolutional neural network (R-CNN) and an adaptive Canny algorithm to enhance the defect detection performance of ECT. An improved faster R-CNN is a deep neural network that can automatically detect and locate multiple defects in a single ECT image, whereas the adaptive Canny algorithm is an edge detection technique that can identify defect boundaries. The proposed method was tested using a dataset of ECT images with different types of defects. The results demonstrated that our method achieved better accuracy, precision, and speed than existing methods. [ABSTRACT FROM AUTHOR]

Centrifugal pumps are important equipment in industrial production. At present, vibration signals are often used to diagnose cavitation in centrifugal pumps, but the vibration signals are easy to be disturbed and the fault characteristics are unstable to be detected. In this paper, a single stage centrifugal pump is taken as the study object, and the vibration signals of various parts of the centrifugal pump cavitation state are collected under different flow conditions. The short-time Fourier transform and one-third octave analysis are performed on the filtered signals, and the characteristic frequency of cavitation and the energy near the characteristic frequency with the development of cavitation are obtained. Based on vibration signals, the vibration root mean square (rms) and kurtosis values of different cavitation states are obtained. Flow state, kurtosis, and rms are used as input variables in the double-layer backpropagation neural network model to identify and classify the cavitation states of centrifugal pumps. The results show that the trained neural network model can accurately identify and classify the cavitation state of the centrifugal pump under the conditions of low flow rate, rated flow rate, and large flow rate, and the accuracy is more than 99.5%. This study provides a new technique for diagnosing cavitation in centrifugal pumps. [ABSTRACT FROM AUTHOR]

Understanding the aerodynamic hysteresis loop phenomenon is essential when assessing aerodynamic performance and designing aircraft control systems. This phenomenon is a result of time delay effects in unsteady flow. Traditional methods of predicting unsteady aerodynamic forces using computational fluid dynamics have drawbacks, such as long cycles and low efficiency. In this paper, we focus on predicting the aerodynamic hysteresis loop of the NACA (National Advisory Committee for Aeronautics) 0012 airfoil in transonic flow using a new model called LIDNN (Latin hypercube sample input deep neural network). This model integrates input signals and optimization methods to improve upon traditional neural network models. Based on the example validation, the LIDNN model is authenticated as an accurate and efficient method in predicting the unsteady aerodynamic hysteresis loop of the NACA 0012 airfoil in transonic flow, and another significant advantage of the proposed model is its ability to solve multivariable problems effectively, even under varying Mach numbers. [ABSTRACT FROM AUTHOR]