147 results about "Soft thresholding" patented technology

The invention relates to a remote sensing image change detection system and method based on multi-modal deep learning. According to the system, the detection precision is improved by constructing a twin network fusing a CNN, a Transform and an attention mechanism; a lightweight backbone network is constructed by adopting multi-size depth separable convolution, and noise is adaptively suppressed and feature expression is enhanced in combination with a soft thresholding channel space attention mechanism; harr wavelet transform downsampling is innovatively introduced to reserve high-frequency details, and traditional pooling operation is replaced to reduce information loss; the twinborn branch semantic deviation is relieved through cross-channel feature exchange, and a Transform codec is used for modeling a long-range dependency relationship; in the decoding stage, edge feature recovery is enhanced by adopting feature splicing and a progressive up-sampling strategy; according to the system, model parameters are reduced, meanwhile, the precision and the anti-interference capability of remote sensing building change detection are remarkably improved, and the system has the advantages of high efficiency and detail keeping.

The invention discloses a laboratory detection data processing method and system based on a machine learning technology, and relates to the technical field of data processing. According to the method, the signal is decomposed through discrete wavelet transform, the noise standard deviation is calculated based on the median of the highest frequency coefficient, the high signal-to-noise ratio data is reconstructed through the dynamic threshold and the soft threshold function, and the signal quality is improved; a sliding window is used for extracting time sequence signal statistics and FFT frequency domain features, spatial features are extracted in combination with an SIFT algorithm, high-correlation features are reserved through mutual information screening, and redundancy is reduced; constructing a graph convolutional network anomaly detection model and an XGBoost-LightGBM weighted regression model, and eliminating pollution data through an anomaly probability threshold to obtain a normal regression predicted value; predicting a compensation amount according to the environmental parameters by using an LSTM network, and obtaining calibration laboratory data according to the normal regression predicted value and the predicted compensation amount; according to the method, the problems of noise sensitivity, feature splitting, model isolation and environment drifting of multi-source data are solved, and the laboratory analysis precision and robustness are remarkably improved.

The invention relates to a sleep respiration detection method, a sleep respiration detection system and a sleep respiration detection cloud platform for apnea syndrome, and the sleep respiration detection system for apnea syndrome is used for collecting a respiration signal caused by thoracic movement of a wearer, and processing and uploading the respiration signal through a microcontroller. According to the sleep respiration detection method, original signals are denoised based on wavelet transform and a soft threshold function, and an apnea event is judged through the combination of a dynamic amplitude threshold and a time threshold. The sleep respiration detection cloud platform adopts an SVM and LSTM-TCN fusion model to identify respiration abnormity, calculates risk scores in combination with blood oxygen decline, event duration and frequency, and triggers early warning in a grading manner. The system supports federated learning and model issuing, and realizes personalized recognition and cloud edge collaboration. The sleep apnea monitoring system realizes continuous monitoring and intelligent identification of sleep apnea, and is suitable for home health management and remote medical treatment.

The invention relates to the field of image processing, and particularly discloses a rural highway pavement disease intelligent identification and positioning system comprising an image standardization module used for obtaining a standardized grayscale image; the wavelet decomposition module is used for obtaining a low-frequency sub-band and a plurality of high-frequency sub-bands; the high-frequency processing module is used for carrying out soft threshold processing on the high-frequency coefficient and retaining high-variance region features; the inverse transformation module is used for reconstructing the de-noised image; the edge enhancement module is used for highlighting crack and pit slot target contour information; the binarization module is used for segmenting a foreground disease candidate region; and the edge filling module is used for separating the disease from the background. According to the method, the core contradiction between background impurity removal and disease feature retention in rural highway tiny disease recognition is effectively solved, a traditional denoising algorithm either smooths tiny disease features or cannot thoroughly remove background impurities, and the system achieves the balance of the background impurity removal and the disease feature retention through cooperation of multiple modules.

The invention relates to the technical field of test and measurement, and discloses a series fault arc detection method, which comprises the following steps: collecting current data of fault arc waveforms and normal arc waveforms of a plurality of different loads; performing wavelet transformation on the preprocessed current data, dynamically extracting detail coefficients of a target layer number, calculating a standard deviation, a normalized energy ratio and an energy entropy based on the wavelet coefficients, splicing peak-to-peak values to form a four-dimensional feature vector, adding a normal or fault label, and converting the feature vector into a three-dimensional tensor; the residual shrinkage module is used for constructing a deep residual shrinkage network model based on a one-dimensional convolutional neural network and introducing an attention threshold generation and soft thresholding mechanism; training the model and monitoring the performance by adopting an early stop method; and inputting label-free sample data to the model, and outputting a detection result. The problems that in the prior art, deep features cannot be reflected, the training cost is high, and overfitting is prone to occurring are solved, and the purposes of improving the accuracy, being high in stability and efficient in detection are achieved.

The invention provides a feature extraction method for an electroacoustic background interference signal of operation power transformation equipment, which belongs to the technical field of power transformation equipment, and comprises the following steps: acquiring an electroacoustic signal, preprocessing the electroacoustic signal, and generating a controlled interference signal at the same time; and performing time-frequency analysis on the preprocessed electroacoustic signal and the controlled interference signal, and optimizing the interference signal parameter to enable the interference signal parameter to be highly similar to the electroacoustic signal. And constructing a wavelet basis function library based on the optimized controlled interference signal, and performing wavelet packet decomposition on the electroacoustic signal to obtain a plurality of frequency band sub-signals. A self-adaptive threshold model is established by using a controlled interference signal, and soft threshold denoising processing is performed on sub-signals. Time-frequency features of the denoised sub-signals and the controlled interference signals are extracted, a feature mapping relation is established, and an initial feature set is obtained; and performing nonlinear dimensionality reduction on the initial feature set by adopting principal component analysis to obtain a dimensionality-reduced feature set. And an improved support vector machine model is adopted to evaluate the importance of dimension reduction features, and an optimal feature set is selected as an electroacoustic background interference signal feature.

The invention provides a bearing signal expansion method and system based on a chaotic particle swarm algorithm and a generative adversarial network, and the method comprises the following steps: firstly, carrying out the preprocessing of a collected bearing acceleration signal, and carrying out the denoising through employing the discrete wavelet transform (DWT) in combination with a Bayesian soft threshold method; secondly, determining a proper signal sample length according to a Nyquist sampling theorem, slicing the signal by adopting a sliding window strategy, inputting the sliced data into a chaos particle swarm algorithm for optimization, and searching an optimized vector which is closest to the structural similarity of the sliced data; and finally, inputting slice data of a real sample into a discriminator, and superposing the optimized feature vector with random disturbance to serve as initial input of a generator. In the training process, the parameters of the generator and the discriminator are mutually confronted and updated until a preset training round is reached. And the small sample problem and the class imbalance problem in bearing fault diagnosis are effectively relieved.

The invention relates to the technical field of marketing operation based on big data analysis, and discloses a marketing operation method based on big data analysis. Preprocessing the user-commodity interaction data to construct an initial interaction matrix; constructing a co-occurrence counting matrix; automatically calculating a sparse penalty coefficient and a convergence threshold; setting an optimization objective function and a soft threshold function; initializing a projection matrix and iteratively solving a sparse projection matrix through a coordinate descent algorithm; constructing a commodity sparse co-occurrence adjacency set; and generating a personalized recommendation result in combination with user historical behaviors. The analysis quality is improved through data preprocessing and denoising, commodity potential correlation is mined through co-occurrence modeling, adaptability is enhanced through automatic parameter adjustment, a projection matrix optimization result is sparse and interpretable, the structure sensing capacity of recommendation is improved through an adjacent structure, and finally high-precision and high-correlation recommendation is achieved. The structure definition, the calculation efficiency and the model generalization ability are excellent, and an efficient and stable personalized marketing recommendation system is constructed.

The invention relates to the technical field of signal processing, and discloses a flip chip vibration signal denoising method and system, and the method comprises the steps: obtaining a vibration signal responded by a to-be-detected flip chip under the action of external ultrasonic excitation, carrying out the segmentation processing of the vibration signal, constructing an initial sparse dictionary according to the segmented vibration signal, and iteratively updating the sparse dictionary; in the iteration updating process, sparseness parameters are adjusted in real time along with the number of iterations, and soft threshold denoising is conducted on residual terms; introducing a feature projection matrix to perform joint modeling on each section of vibration signal, and constructing a joint optimization model according to the updated sparse dictionary and the corresponding sparse coefficient; alternately optimizing the sparse coefficient and the feature projection matrix in the joint optimization model to obtain an optimal solution of the sparse coefficient; and combining the updated sparse dictionary and the optimal solution of the sparse coefficient to reconstruct the vibration signal to obtain a denoised vibration signal. According to the method, transient features can be effectively extracted, expression of key features is enhanced, noise is suppressed, and denoising robustness and reconstruction precision are improved.

The invention discloses a broadband signal parameter identification method based on wavelet noise reduction and an improved VMD-Prony algorithm, and the method comprises the following steps: carrying out the noise reduction preprocessing of a broadband signal based on a soft threshold wavelet noise reduction method, so as to reduce the interference of the noise to a measurement process, and building an evaluation index according to a Pearson's correlation coefficient and a signal-to-noise ratio for evaluating the noise reduction effect; parameters of a mutual information entropy and an energy entropy optimization VMD algorithm are introduced, the noise-reduced signal is decomposed by using the parameter optimization VMD algorithm, a plurality of modes with center frequencies are decomposed, and the complexity of broadband signal processing is reduced; performing parameter estimation on each mode by using a Prony algorithm, selecting a proper model order p according to the characteristics of the modes, performing numerical calculation by constructing a plurality of functions, obtaining frequency, amplitude and phase information of each mode, and realizing parameter identification of the broadband signal; according to the invention, the problems of noise interference and insufficient identification precision of the broadband oscillation signal in the parameter identification process are solved.

The invention relates to the field of data anomaly detection, and particularly discloses a multi-stage anomaly detection system based on multivariable time sequence data, which comprises an input data acquisition module used for acquiring time sequence multivariable data; the MTS-Mixer spatio-temporal feature extraction module is used for extracting multi-level time sequence correlation features through a feature mixing technology and generating low-rank compression prediction features; the time sequence anomaly detection module is used for realizing anomaly positioning and classification by adopting alternate stacking of an Angle-Attention layer and a feed-forward network, expanding anomaly differences in combination with prior constraints and adversarial learning and fusing reconstruction errors and association differences to generate anomaly scores; and the attention mechanism soft thresholding module is used for suppressing noise, enhancing key signals and outputting high signal-to-noise ratio representation. According to the method, multiple technologies are integrated, and real-time anomaly detection and interpretable positioning of the high-dimensional time series data are realized through multi-stage collaborative optimization.

The invention discloses an ocean electromagnetic signal denoising method based on wavelet transform, which comprises the following steps of: performing wavelet decomposition on a noisy ocean electromagnetic signal to obtain a wavelet coefficient of each layer; carrying out threshold calculation on the high-frequency wavelet coefficient and obtaining soft and hard threshold functions for processing the wavelet coefficient; and performing self-adaptive threshold processing on the high-frequency wavelet coefficient, namely quantifying the high-frequency wavelet coefficient by adopting a self-adaptive threshold function which dynamically adjusts the characteristics of a soft threshold and a hard threshold by adjusting a parameter k belonging to [0, 1], and when k is equal to 0, the self-adaptive threshold function is equivalent to the soft threshold function, when k is equal to 1, the self-adaptive threshold function is equivalent to the hard threshold function, and when k is equal to 0, the self-adaptive threshold function is equivalent to the hard threshold function. To eliminate discontinuity and constant deviation at the threshold; 4, performing inverse wavelet transform on the processed wavelet coefficient, and reconstructing to obtain a de-noised ocean electromagnetic signal; not only are the signal-to-noise ratio and definition of the signals significantly improved, but also a more reliable and accurate data basis is provided for real-time monitoring, accurate analysis and subsequent scientific research of the marine electromagnetic signals.

The invention discloses a hydroelectric generating set vibration signal noise reduction method based on SW-EEMD (Single Wave-Ensemble Empirical Mode Decomposition) and Hansel-SVD (Hanker-Singular Value Decomposition). The method comprises the following steps: processing an original noisy signal by adopting an SW-EEMD method; screening effective IMF components through correlation coefficients; respectively constructing Hankel matrixes for the screened IMF components and residual components; respectively carrying out singular value decomposition on the constructed Hankel matrix; carrying out soft threshold processing on the obtained singular value; the method aims at restraining the end effect of EEMD, noise in the vibration signals can be effectively filtered out, signal features are enhanced, more real and effective signal components can be obtained easily, and more reliable data are provided for state operation and maintenance of the hydroelectric generating set.

The invention discloses a signal processing method and device for a wind measurement laser radar, and the method comprises the steps: carrying out the windowing interception of an original laser echo signal, and obtaining an intercepted signal; performing dual-tree complex wavelet decomposition transformation on the intercepted signal to obtain each level of approximation wavelet and each level of detail wavelet; performing correlation operation and soft threshold filtering on each level of detail wavelet to obtain a corrected detail wavelet; combining the highest-level approximation wavelet with the corrected detail wavelet to obtain a preliminary noise reduction signal; performing Bayesian maximum posteriori estimation filtering on the preliminary noise reduction signal to obtain a final noise reduction signal; performing dual-tree complex wavelet inverse transformation on the final noise reduction signal to obtain a noise reduction time domain signal; performing fast Fourier transform on the denoised time domain signal to obtain a frequency domain signal; and performing frequency spectrum correction on the frequency domain signal, and then extracting to obtain Doppler frequency. According to the invention, noise filtering of the signals of the wind measurement laser radar is realized, and the Doppler frequency extraction accuracy is improved.

The invention discloses a blind separation and time-frequency noise reduction combined comprehensive pulsar time establishing method. The method comprises the following steps: selecting a pulsar meeting a preset requirement; processing the timing residual error of the selected pulsar; establishing a mixed signal model, iteratively solving an anti-mixing matrix through an approximate negentropy contrast function, and separating independent components from the multi-dimensional features; carrying out Pearson correlation analysis on each separated independent component and reference time; and performing multi-scale decomposition on the selected common-mode clock difference signal by adopting a plurality of wavelet bases, and performing Bayesian soft threshold processing to obtain the final comprehensive pulsar time. According to the method, the long-term stability of the comprehensive pulsar is improved, the common problems of aliasing and modal mixing in a high-noise environment in the prior art are solved, and the extracted signal is more stable.

The invention provides a rolling bearing fault signal noise reduction and feature extraction method based on adaptive semi-soft threshold set noise reconstruction empirical mode decomposition (ASTENEEMD) and (Sort-time Fourier Transform, STFT), and belongs to the field of intelligent diagnosis of rolling bearing faults. The rolling bearing fault signal noise reduction and feature extraction method based on the adaptive semi-soft threshold set noise reconstruction empirical mode decomposition (ASTENEEMD) and the Sort-time Fourier Transform (STFT) is characterized in that the rolling bearing fault signal noise reduction and feature extraction method based on the adaptive semi-soft threshold set noise reconstruction empirical mode decomposition (Sort-time Fourier Transform (STFT) is provided. According to the method, firstly, noise reduction of an original vibration signal of the rolling bearing is achieved through self-adaptive semi-soft threshold set noise reconstruction empirical mode decomposition, then, the noise reduction signal is converted into a time-frequency image through short-time Fourier transform, and time-frequency feature extraction of a non-linear and non-stable vibration signal is achieved. According to the method, the rolling bearing fault signal quality is enhanced, the noise reduction effect can be remarkably improved, the fault features in the signal are clearer, rich time-frequency features can be extracted, and reliable input is provided for subsequent fault diagnosis by using a deep learning model.

The present invention provides a method and device for real-time monitoring of the working state of a gearbox, relating to the technical field of gearbox testing. In the present invention, all characteristic information in time domain and frequency domain and other scale spaces is integrated through wavelet transform. Then, during the training process of the artificial intelligence model, methods such as attention mechanism and soft thresholding are adopted to screen the characteristics of the time domain information and frequency domain information, realize noise reduction, obtain key characteristics related to the fault type, and train an artificial intelligence model based on the key characteristics and fault type. By using the thus obtained artificial intelligence model to conduct real-time monitoring on the gearbox, the noise in the vibration signal can be effectively reduced, the influence of the noise on the gearbox fault prediction can be weakened, and the accuracy and reliability of the shape-integrated gearbox fault diagnosis can be improved.

The invention discloses an acoustic signal partial discharge detection system of a distribution network transformer, which is characterized by comprising a multi-sensor acquisition unit, a synchronous clock control unit, a signal preprocessing unit, an edge calculation processing unit and a communication and storage unit, the edge calculation processing unit integrates a cross-band denoising module, a multi-scale feature extraction module, a lightweight neural network diagnosis module and a diagnosis result fusion module, wherein the cross-band denoising module is used for performing joint denoising on audible sound and ultrasonic frequency band signals. According to the invention, a set of partial discharge detection scheme for a distribution network transformer scene is established, and the scheme flow comprises denoising, feature extraction, neural network training and diagnosis. According to the scheme, audible audio frequency band and ultrasonic frequency band data are covered, and interference generated by mechanical vibration of the transformer in the ultrasonic frequency band is removed through a soft threshold value of wavelet transformation and a pulse detection denoising algorithm.

The invention discloses a buoy-based water body apparent spectrum real-time acquisition method, device and equipment, a medium and a product, and relates to the field of marine environment intelligent monitoring, and the method comprises the following steps: preprocessing original marine spectrum data to generate high-quality marine spectrum data with time-space alignment; reconstructing a three-channel spectral data matrix through multi-scale discrete wavelet decomposition and soft threshold quantization compression, inputting the three-channel spectral data matrix into a dynamic weight model, and identifying a spectral abnormal mode; obtaining an abnormal feature mark set, calculating spectral shape parameters, and generating a shape parameter sequence; and fusing the abnormal feature mark set and the shape parameter sequence to generate a spectral morphological feature curve, integrating the three-channel spectral data matrix, the abnormal mode and the shape parameter sequence, and outputting a real-time water body apparent spectrum acquisition report. Through deep fusion of the dynamic weight model and the multi-dimensional morphological features, the accuracy and reliability of spectrum anomaly detection in a complex marine environment are remarkably improved.

The invention discloses a state monitoring and diagnosing method and system for a marine main engine, and the method comprises the steps: collecting the operation state parameters of the marine main engine, and carrying out the preprocessing; embedding a dynamic soft thresholding layer in each residual module of the residual shrinkage network, and executing channel-by-channel soft thresholding on the feature map; constructing a double-flow attention network, fusing the time-frequency domain characteristics of the vibration signals and the temperature field data of the space heat conduction model, and capturing a causal sequential relationship of multi-modal data through a gating mechanism; carrying out embedded optimization on the model by adopting depth separable convolution, adaptive load normalization and knowledge distillation; the model is carried on a shipborne industrial personal computer, and whether model updating needs to be triggered or not is judged by calculating the weight difference between the shipborne model and the global model of the cloud platform. According to the method, a whole-process closed loop from data acquisition to model deployment is realized, the anti-noise capability, causal association analysis, resource optimization and privacy protection are integrated, and the intelligent operation and maintenance requirements of the marine main engine are met.

The present invention relates to a remote sensing image change detection system and method based on multimodal deep learning. The system improves detection accuracy by constructing a twin network that integrates CNN, Transformer and attention mechanism; adopts multi-scale depthwise separable convolution to construct a lightweight backbone network, and combines the soft thresholding channel space attention mechanism to adaptively suppress noise and enhance feature expression; innovatively introduces Harr wavelet transform downsampling to retain high-frequency details, replacing traditional pooling operations to reduce information loss; alleviates twin branch semantic bias through cross-channel feature exchange, and uses Transformer codec to model long-range dependencies; feature splicing and progressive upsampling strategies are used in the decoding stage to enhance edge feature recovery; while reducing model parameters, the system significantly improves the accuracy and anti-interference ability of remote sensing building change detection, and has the advantages of high efficiency and detail preservation.

The invention discloses an education course session recommendation method and device, equipment and a storage medium, and relates to the technical field of education resource recommendation, and the method comprises the steps: collecting and preprocessing student course selection information, constructing a hypergraph model, and employing Laplacian to normalize a bidirectional weighting adjacency matrix; message passing, sparse attention calculation and soft threshold comparative learning optimization are performed by adopting a gated hypergraph neural network, and finally a recommendation result is generated by comparing with a global course feature library, so that the high-order relationship between courses is effectively captured, the noise influence is inhibited, the accuracy and individuation level of course recommendation are improved, and the course recommendation efficiency is improved. And more accurate learning resource recommendation is provided for students.

The invention discloses a central air-conditioning system energy consumption prediction method based on a WTD hybrid algorithm, and relates to the technical field of central air-conditioning systems, and the method comprises the following steps: S1, carrying out the wavelet transform decomposition processing of original cold load data, carrying out the denoising through a soft threshold function, and reconstructing the data into a low-frequency trend component and a high-frequency detail component; s2, inputting the denoised data into a hybrid model comprising a Transform module and an LSTM (Long Short Term Memory) module, extracting global correlation characteristics of input variables by the Transform module through a multi-head self-attention mechanism, and enhancing time sequence information by using position coding; the LSTM module models a long-term dependency relationship of a time sequence through a gating mechanism of an input gate, a forgetting gate, and an output gate. According to the invention, through the improved wavelet transform decomposition technology, in combination with the Daubechies wavelet basis and the adaptive soft threshold function, accurate filtering of noise components and effective signal reconstruction are realized, and interference of non-stationarity on model input is reduced.

The invention discloses a visual target detection method and device used in a strong noise interference environment, and belongs to the technical field of target detection, and the method comprises the steps: replacing a C2f module with a deep residual shrinkage module DRSB in a backbone network of YOLOv8; soft thresholding processing is introduced, and noise features are removed; an attention mechanism module WSW is designed in front of an SPPF module of a backbone network of the YOLOv8, multi-head self-attention is calculated in windows of a feature map and between the windows in sequence by using sliding window operation, the improved backbone network of the YOLOv8 is connected with a feature fusion network and a detection head of the backbone network of the YOLOv8, a complete model is established, noise data of different types and different grades are prepared, and the noise data of different types and different grades is obtained. Training and testing are carried out, and the effectiveness and universality of the visual target detection method in the strong noise interference environment are verified; the method is suitable for target detection in a strong noise interference environment, and has the advantages of high precision, high speed, low power consumption and the like.

The present application relates to a kind of unmanned aerial vehicle signal identification method and system, belong to unmanned aerial vehicle signal identification technical field.It includes: step 1: adaptive noise base estimation;The input unmanned aerial vehicle IQ signal, i.e.unmanned aerial vehicle signal is carried out short-time Fourier transform, and time-frequency matrix is obtained, generates unmanned aerial vehicle time-frequency diagram, estimates noise power spectrum, calculates the power spectral density of each frequency point;Step 2: time-frequency feature enhancement;5 order Butterworth band-pass filter is used to frequency domain filtering to signal;Soft threshold domain denoising of soft short-time Fourier transform result is used;Step 3: multi-feature fusion regression;Energy ratio feature, spectral peak sharpness feature are calculated, and deep learning feature is extracted, feature fusion and regression;Step 4: through the unmanned aerial vehicle signal classification model after training, unmanned aerial vehicle signal identification is realized.The present application method is high in precision in complex environment signal-to-noise ratio estimation, and unmanned aerial vehicle classification precision and generalization are excellent.

The present invention discloses a method for denoising GNSS monitoring data of open-pit mine slopes by integrating multiple source indicators. The method comprises the following steps: S1, collecting GNSS monitoring data of the slopes and detecting and interpolating large gross errors using the 3σ method, then performing wavelet decomposition to obtain low-frequency coefficients and high-frequency coefficients; reconstructing the trend term Q using the low-frequency coefficients, solving the mean square error m using the first and second layer high-frequency coefficients, and then detecting gross errors again; S2, performing empirical mode decomposition on the GNSS monitoring data of the slopes to obtain a number of slope IMFs; screening to obtain high-noise modes, low-noise modes, and residuals, reconstructing the low-noise modes and residuals to obtain a preliminary denoised slope GNSS signal result EMD_A; and S3, processing the high-noise modes of the slopes using the interval soft threshold method and synthesizing them with the slope EMD denoising result EMD_A to obtain the final slope GNSS denoising result. The present invention achieves accurate denoising of GNSS monitoring data of open-pit mine slopes, providing more reliable data support for slope stability monitoring in mining areas.

The present invention discloses a method for detecting and removing stripe noise in satellite-borne spectral images. The method comprises an image classification and characteristic analysis module, a stripe detection and analysis module (including a stripe detection unit, a characteristic analysis unit, and a degradation modeling unit), an image transformation module, and a soft threshold signal decomposition module. Based on independent analysis of the original space-spectral domain and the transform domain, the present invention comprehensively integrates the space-spectral domain analysis method and the transform domain analysis method to simultaneously detect and remove stripe noise. The method also fully analyzes the image characteristics and wavelet distribution characteristics, deploys an image decomposition method in the wavelet domain, effectively analyzes the stripe components, and retains image detail information. The present invention has the advantages of being adaptable to the complexity of spectral image sources and the diversity of application environments, taking into account the differences in different stripe noise distributions, and having strong generalization and robustness.

This invention discloses an underwater acoustic anti-interference transmission method and system based on sparse time-frequency feature mapping. The method includes: at the transmitting end, adaptively generating a fractional-order linear frequency-modulated waveform with a specific time-frequency shear slope based on the Doppler state of the underwater acoustic channel, achieving physical-layer focusing of transmitted energy; at the receiving end, constructing a hybrid observation model containing wide-block sparse channel components and narrow-block sparse burst noise components, and using a dual-channel variational Bayesian algorithm to jointly iteratively infer the posterior probability distribution of environmental burst noise in the fractional-order domain; finally, recovering the original signal through soft-threshold interference cancellation and fractional-order channel equalization. This invention effectively solves the communication failure problem caused by high-dynamic Doppler diffusion and marine biological impulse noise interference in underwater acoustics by actively mapping the waveform and using heterogeneous sparse joint inference at the receiving end, significantly improving transmission reliability in harsh underwater acoustic environments.

The present application relates to the technical field of transformer fault diagnosis, and in particular to a transformer body vibration signal analysis and fault diagnosis method, device and medium. The present application combines the kurtosis characteristics of the vibration signal, adopts a semi-soft threshold function wavelet denoising method based on the threshold selection method of the 3sigm rule, and achieves better denoising effect. Through the fault diagnosis model of the organic fusion of the two improved HHT transforms-mobilenetV2 model, combined with different feature extraction methods, it is more conducive to retaining the effective features of the vibration signal; the improved mobilenetV2 model designs a multi-scale deep convolution model, introduces a channel attention mechanism before the channel-by-channel convolution, and after multi-scale deep convolution feature extraction, a multi-source data attention mechanism is introduced; without affecting the safe and reliable operation of the transformer, the intelligent diagnosis of the vibration state fault of the transformer is realized.

The invention discloses a design method for ISTA sparse channel estimation based on a single-transmitting and single-receiving OFDM (Orthogonal Frequency Division Multiplexing) system. According to the method, the signal can be recovered under the condition of unknown sparseness, accurate reconstruction of the signal is realized, and a sparse channel is reconstructed through a small amount of training data, so that the training overhead is reduced, the spectrum efficiency is improved, and signal reconstruction is carried out by utilizing gradient updating and a soft threshold threshold function. A simulation experiment shows that the algorithm can realize better channel reconstruction, and meanwhile, the algorithm also has higher convergence speed and lower calculation complexity.