147 results about "Denoising algorithm" patented technology

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 provides a Fresnel optical image target detection and identification method based on event stream data enhancement, and the method comprises the steps: carrying out the data preprocessing of collected Fresnel optical image event data through employing a denoising algorithm based on time-space continuity, and screening the original event stream data; converting the screened event stream data of the target area into pixel data under an image coordinate system by adopting a dual threshold limiting strategy of event quantity and accumulation time; existing image features are enhanced, and pixel-level fusion, feature-level fusion and decision-level fusion are carried out on event stream data of an event camera and RGB camera data; and the target detection module is used for performing target detection and recognition on the fused data by using a YOLO target detection model. According to the method, the detection of the Fresnel optical image target is realized, so that the detection and recognition capability of the Fresnel optical image target in a complex environment is improved, and the method is particularly suitable for scenes such as seaborne landing guidance and the like.

A deep learning fan fault diagnosis method based on double wavelet fusion and CEEMDAN decomposition comprises the following steps: S1, collecting vibration signals of a fan motor driving end, a fan driving end and a non-driving end, and constructing a data set; s2, performing improvement on the basis of wavelet packet noise reduction to form a double-wavelet fusion noise reduction algorithm; s3, carrying out noise reduction processing on the acquired signals; s4, carrying out sample division on the noise reduction signals according to a fixed time window, and carrying out classification according to measurement points; s5, randomly dividing the samples into a training set, a verification set and a test set; s6, decomposing all the samples by using CEEMDAN, and extracting IMFs; s7, sending the sample into the time-frequency domain joint feature extraction model of the corresponding measuring point for modeling; s8, training to obtain a diagnosis model and exporting a weight file; and S9, loading the model detection test set, and comprehensively judging the operation state of the fan. According to the invention, the signal processing quality and the fault identification accuracy are improved.

The invention provides a comprehensive noise reduction performance evaluation method for a nonlinear ultrasonic detection signal noise reduction algorithm, and the method comprises the steps: firstly constructing a multi-working-condition noise reduction performance pre-screening mechanism based on error band analysis based on early-stage experimental data, and then introducing a radar map as an evaluation tool for the comprehensive noise reduction effect of the signal noise reduction algorithm. The noise reduction effects of a moving average method (MA), a spectral subtraction method (SS), a short-time Fourier transform method (STFT), a wavelet transform method (WT) and an orthogonal matching pursuit algorithm (OMP) are compared, and finally a signal noise reduction algorithm with the optimal comprehensive efficiency is screened out. On the basis of algorithm optimization, quantitative mapping rules between microcrack three-dimensional geometric parameters and relative nonlinear coefficients are analyzed through regression modeling, and the significant level of the correlation degree of the relative nonlinear coefficients and microcrack size parameters is effectively improved; and a high-confidence theoretical support is provided for quantitative nondestructive detection of the microcracks in engineering practice.

The invention provides a Prony algorithm-based cable segment wave velocity acquisition method, device and system, and a medium, and the method comprises the steps: testing a to-be-tested cable, and obtaining a cable signal transfer function; on the basis of the cable signal transfer function, in combination with cable joint distribution, constructing a cable signal approximation function based on a Prony method; constructing a linear prediction model of the cable signal based on the cable signal approximation function; performing denoising processing on the linear prediction model based on a singular value decomposition method to obtain a denoised linear prediction model; solving the denoised linear prediction model to obtain an attenuation coefficient; and calculating the segmented wave velocity of the cable based on the attenuation coefficient. According to the method, the Prony estimation method is combined with the singular value decomposition noise reduction algorithm, high-precision extraction of attenuation constants and segmented wave velocity decoupling are achieved, and then the electrical distance positioning precision is improved.

The invention discloses a method for detecting tea saponin in a tea extract, and relates to the technical field of chemical analysis and computer science, and the method comprises the following steps: eliminating high-frequency noise of tea extract data by using a Gaussian filtering algorithm, eliminating baseline drift by using a baseline correction algorithm, and generating a tea extract sample data stream; the method comprises the following steps: performing dynamic gradient separation through a chromatographic signal simulation algorithm on the basis of a tea extract sample data stream to generate a virtual chromatographic peak sequence, calculating a peak area through a sliding window integration algorithm on the basis of the virtual chromatographic peak sequence, and generating chromatographic signal data by using a wavelet transform threshold denoising algorithm, the chromatographic signal data are input into the deep convolutional neural model, the feature pyramid network is obtained through the multi-scale feature fusion algorithm, the feature information of the chromatographic peaks is extracted from different scales by constructing the deep convolutional neural model and combining the multi-scale feature fusion algorithm, and therefore the recognition precision of the feature peaks is improved.

The invention belongs to the field of image processing and computer vision in computer intelligent information processing, and discloses an industrial product unsupervised anomaly detection method and system based on a conditional control diffusion model, and the method comprises the steps: constructing a diffusion model based on conditional control and a multi-scale double-attention mechanism, reconstructing the to-be-detected sample image by using a diffusion model based on condition control and a multi-scale double-attention mechanism to obtain a reconstructed image with the same size as the to-be-detected sample image; and extracting feature maps of different scales from the to-be-detected sample image and the reconstructed image through the same pre-trained feature extraction model. Performing a two-stage background denoising algorithm on each scale feature map to obtain an initial abnormal feature map; calculating an average value of a plurality of initial abnormal feature maps obtained by using the same method to obtain a final abnormal feature map; and processing the final abnormal feature map by adopting a thermodynamic diagram generation algorithm to generate a final abnormal thermodynamic diagram, thereby realizing the anomaly detection of the to-be-detected sample image.

The invention discloses a method and a system for detecting internal defects of a bathroom glass door based on multispectral imaging. The method comprises the following steps: carrying out multi-view and multiband image acquisition on the bathroom glass door by adopting a combined spectrum light source; performing adaptive reflection correction on the acquired image to generate a multi-band basic image; optimizing the multi-band basic image through a hybrid denoising algorithm to generate a multi-band detection image; performing multi-modal feature extraction and fusion on the multi-band detection image to obtain a multi-modal fusion feature vector, inputting the multi-modal fusion feature vector into a multi-feature fusion model combining a convolutional neural network and a support vector machine, and identifying and grading defect types by adopting a dual-stage feature enhancement-classification architecture; and outputting a defect type and severity grading result to a terminal in real time. The method is used for solving the technical problems of visual angle and spectrum blind areas, insufficient interference suppression capability, wave band crosstalk, feature deficiency and poor small sample generalization in existing bathroom glass door defect detection.

The invention relates to the technical field of infrared imaging, in particular to an infrared image self-adaptive denoising algorithm and system based on an FPGA (Field Programmable Gate Array), and is characterized in that an adaptive blind pixel searching algorithm module captures data of a certain frame number at an FPGA end, performs multi-directional gradient detection on each pixel point, sets conditions to judge to obtain a blind pixel position, and marks the blind pixel position; the convolution kernel self-adaptive blind pixel replacement module performs self-adaptive blind pixel replacement multi-time module multiplexing at the PL end of the FPGA through a convolution kernel sliding block and gradient comparison to ensure complete coverage of a large blind pixel group, and the self-adaptive median filtering module intelligently selects and replaces pixels meeting requirements through automatic sorting of center pixels and surrounding pixels; according to the method, surface fuzzy filtering is carried out through a surface-blue algorithm module, so that a smooth area in an image is effectively smoothed, details of an edge area are reserved, blind pixels in a video stream can be effectively detected and compensated, and meanwhile traditional noise is suppressed and image details are reserved through multi-stage filtering.

The invention discloses a flammable and explosive gas remote monitoring method and system based on big data. The flammable and explosive gas remote monitoring method comprises the following steps: acquiring an original data set acquired by a multi-source sensor in real time; performing signal preprocessing on the original data set by adopting a denoising algorithm, and performing normalization processing to obtain a first multi-parameter data set; generating a second fusion feature set according to the first multi-parameter data set; if the parameter value of the second fusion feature set exceeds a preset threshold value, identifying a gas leakage event by adopting a classification model, and generating leakage probability and range data in combination with a diffusion model; performing compression coding on the leakage probability and range data, and dynamically adjusting the transmission frequency and the size of a data packet according to network state parameters to form a stable transmission stream; and decoding the stable transport stream at the monitoring center, generating a dynamic visual interface by adopting a visual engine, and judging whether the reliability of the system meets a preset standard or not according to interface parameters. According to the invention, real-time monitoring, intelligent identification and early warning of gas leakage are realized.

The invention relates to the technical field of tunnel construction quality detection, in particular to a tunnel initial spraying concrete spraying quality detection method based on a visual technology, which comprises the following steps: S1, arranging an automatic cleaning device at the front end of equipment through multi-modal sensor fusion and a self-adaptive exposure denoising algorithm in combination with a multi-information joint correction technology, and performing multi-modal sensor fusion and self-adaptive exposure denoising; through original point positioning calibration, high-precision robust acquisition of three-dimensional coordinates and apparent vector data of the spraying and mixing surface in a complex environment is carried out. Through multi-modal sensor fusion, anti-interference processing and accurate calibration, high-precision robust acquisition of the spraying and mixing surface data in a complex tunnel environment is ensured, a reliable data basis is provided for subsequent detection and control, the spraying and mixing surface flatness control precision is effectively improved, the material loss rate is reduced, and the working efficiency is improved. And the strategy can be adaptively adjusted according to geological conditions and construction states, and the construction efficiency and safety are both considered.

The invention relates to the technical field of operator safety management, and discloses a non-contact physiological status rapid screening method and system and a medium, and the method comprises the steps: collecting an RGB three-channel space-time image sequence of a facial region of interest; carrying out illumination correction on the RGB three-channel space-time image sequence by adopting an adaptive gamma correction algorithm, carrying out end-to-end noise reduction processing on the illumination correction image sequence by utilizing a one-dimensional convolutional neural network auto-encoder noise reduction algorithm, and separating a remote photoelectric volume pulse wave tracing pulse signal; calculating physiological indexes of heart rate, heart rate variability and oxyhemoglobin saturation based on the remote photoplethysmography pulse signal; and performing grading judgment on the physiological indexes based on the physiological index grading standard, determining a final post suitability judgment grade in combination with the wooden barrel effect model, and generating a grading early warning signal. According to the method, accurate and rapid evaluation of the physiological state of the operator in a complex environment can be realized, the individual health degree can be comprehensively and objectively considered, and decision support is provided for reasonable distribution of posts and efficient configuration of personnel.

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 damage identification method and device for an in-service steel wire rope type horizontal lifeline, and belongs to the technical field of high-altitude operation safety facilities. The method comprises the following steps: synchronously acquiring a magnetic flux leakage signal and a surface image of an in-service steel wire rope through magnetic flux leakage detection equipment and a high-definition camera which are carried on a steel wire rope inspection robot; preprocessing the acquired magnetic flux leakage signal, wherein the preprocessing comprises singular value elimination and trend term removal processing; carrying out de-noising processing on the pre-processed signal by adopting an improved wavelet threshold de-noising algorithm fused with a Sigmoid function; extracting a characteristic value for representing the damage of the steel wire rope, and performing normalization processing to form a characteristic vector; and inputting into a BP neural network identification model optimized by a genetic algorithm for identification, and outputting an assessment result of the damage type and positioning of the steel wire rope. The method can realize automatic and quantitative detection and accurate identification of internal and external damages of the steel wire rope.

The present application belongs to the field of seismic data processing and data optimization, in particular to a method and device for seismic prestack data optimization based on improved BEMD algorithm. The method of the present application uses improved BEMD algorithm and adaptive denoising algorithm to decompose prestack gathers into characteristic signals of different scales. Then, orthogonal wavelet transform denoising based on threshold is carried out on each component to remove most of the noise. Then, the correlation coefficient between each component and the original data is calculated, and the data is reconstructed based on the correlation coefficient. The effective signal is retained to the greatest extent, the interference of noise signal is removed, the signal-to-noise ratio of prestack gathers is improved, and a good data basis is provided for subsequent seismic prediction algorithms.

The invention discloses a tea processing quality real-time monitoring system and monitoring method based on machine vision, and relates to the technical field of tea processing quality monitoring. According to the method, tea processing images are continuously collected from multiple angles, color, shape and texture changes of tea are comprehensively captured, the images are preprocessed by using an advanced denoising algorithm, the definition and accuracy are improved, high-quality data are provided for subsequent feature extraction, the problems that traditional monitoring information is not completely acquired and subjective judgment errors are large are effectively solved, and the method is suitable for large-scale popularization and application. The monitoring accuracy and reliability are improved; meanwhile, environment data and dynamic feature vectors extracted by a convolutional neural network are fused, the fermentation degree, the drying effect and other quality indexes are accurately analyzed, a parameter adjusting module intelligently judges the processing process deviation and adjusts parameters in real time, the processing stability and consistency are ensured, and the processing efficiency is improved. The problems that traditional monitoring lacks real-time performance and predictability and is difficult to adjust in time are solved, and quality monitoring, regulation and control in the tea leaf processing process are achieved.

A deep learning segmentation method for asphalt mixture CT images relates to the technical field of aggregate image segmentation. CT images of asphalt mixtures are collected and voids are removed. Image equalization is performed using a gamma beam hardening correction algorithm. A filter window is selected, and an adaptive bilateral filtering denoising algorithm based on local image information is used to denoise the image. The U-Net model is improved, with the Inception convolution module replacing the standard convolution operation and residual connections replacing skip-layer connections. A spatial attention mechanism is introduced, and a joint loss function is used. Image samples are selected, aggregates are labeled, and data augmentation is used to increase the sample size to form a training set for model training. After training, this set is used to segment other images. By equalizing and denoising CT images and using the improved U-Net model, the method can achieve accurate segmentation of aggregate-mortar boundary information, avoiding aggregate adhesion issues.

The invention discloses a multi-angle three-dimensional measurement device and a point cloud denoising method. The device comprises a projection device, an image acquisition device, a connecting piece for fixedly connecting the projection device and the image acquisition device, a two-dimensional displacement platform group for adjusting a measurement distance and an electric rotating platform for rotating a measured object, all parts are linked by a controller, and one-button automatic calibration and multi-angle measurement are realized. The method is based on the device, and noise is effectively filtered out from multi-view point clouds collected by rotating around the same axis through nearest neighbor search and statistical judgment. Through combination of displacement and the rotating platform, multi-angle data are automatically obtained, and metal reflection interference is effectively inhibited; through an innovative rotation consistency denoising algorithm, the point cloud quality is remarkably improved, and the method is suitable for multi-angle three-dimensional measurement including high-reflection workpieces.

The application provides an intelligent image processing system and method for a car based on a perception algorithm model, and the method comprises the following steps: S1. spatio-temporal reference double anchoring and dynamic intrinsic extrinsic parameter calibration of a vehicle-mounted image acquisition node; S2. multi-node image heterogeneous domain normalization and adaptive preprocessing based on a self-adaptive kernel regression non-local mean denoising algorithm; S3. hierarchical image feature extraction and semantic anchoring based on a graph neural network dynamic feature interaction network; S4. cross-node and cross-frame feature mutual checking and pseudo-feature elimination; S5. full-scene semantic completion and dynamic target trajectory prediction based on a variational autoencoder trajectory prediction model; S6. dynamic lightweight adaptation and algorithm power adaptive scheduling of the perception algorithm model; and S7. risk scene grading identification and image targeted enhancement output based on semantics and trajectories. The application provides stable, accurate and efficient vehicle-mounted image perception support for intelligent driving of a car, and improves the safety and adaptability of environmental perception of intelligent driving.

This invention discloses a vibration signal denoising and feature extraction method based on wavelet adaptive thresholding, belonging to the field of vibration sensor signal processing technology. The method includes the following steps: Step 1, performing multi-level decomposition of the signal using a filter bank constructed based on discrete wavelet transform; Step 2, adaptively calculating the noise threshold; Step 3, implementing wavelet coefficient shrinkage denoising based on the SureShrink algorithm; Step 4, reconstructing the vibration signal based on quasi-discrete wavelet transform; Step 5, automatically extracting signal features based on a multi-layer recurrent neural network. This invention, through signal decomposition based on discrete wavelet transform and an adaptive thresholding denoising algorithm, reduces the error impact caused by the vibration sensor's own noise and environmental disturbances, retains most of the features of the original vibration signal, and establishes a multi-layer recurrent neural network for the denoised vibration signal to achieve automatic extraction of time features, providing an important foundation for the subsequent development of fault diagnosis and health monitoring algorithms.

The invention discloses a self-adaptive anti-interference water quality detection method, system, equipment and medium, belongs to the technical field of water quality detection, and aims to solve the technical problem of low accuracy of a water quality detection result in the prior art. The method comprises the steps of task parameter loading, global path cruising, pollution core area identification, pollution core area sampling, digital twin map generation, map transmission and signal denoising. When the pollution core area is identified, suspected pollution points are screened preliminarily, the comprehensive pollution index of the suspected pollution points is calculated, and the pollution core area with the highest pollution degree is locked; and when map transmission and signal denoising are carried out, the signal receiver carries out denoising and purification treatment on the transmitted signal through a Dig Flow denoising algorithm, and uploads the signal to a shore-based platform after the signal is verified to be correct, so that water quality detection is completed. A pollution assessment scheme of multi-parameter algorithm fusion, CPI comprehensive assessment and a DigFlow denoising algorithm is adopted, the pollution identification accuracy is improved, the traceability error is small, the signal error rate is reduced, and the signal-to-noise ratio is improved.

In order to de-noise medical imaging data, a first imaging data set (HE) and a second imaging data set (LE) are decomposed according to a spatial frequency band to generate a high-frequency data set (HE-HF, LE-HF) corresponding to a high-frequency band and a low-frequency data set (HE-LF, LE-LF) corresponding to a low-frequency band. A trainable denoising algorithm (7) is trained by performing an optimization which uses at least one parameter of the denoising algorithm (7) as an optimization variable and an objective function (L1, L2, L3, L4, L5) which depends on the denoised high-frequency data set (HE-HF ') and the high-frequency data set (LE-HF) of the second imaging data set. A denoised high-frequency dataset (HE-HF ') is generated by applying a denoising algorithm (7) to the high-frequency dataset of the first imaging dataset (HE-HF). A trained denoising algorithm (7) is applied to the high frequency dataset of the first imaging dataset (HE-HF) to generate a final denoised high frequency dataset (HE-HF ').

In accordance with a method of detecting a pressure induced sensor artifact (PISA) in an analyte trace, a measured analyte trace having a plurality of data samples obtained over a period of time from an analyte sensor is received. A reconstructed analyte trace and an associated confidence window is generated from the measured analyte trace using a Bayesian denoising algorithm that includes a model that models the measured analyte trace as a sum of an unknown true analyte trace and a measurement error. The measured analyte trace is compared to the reconstructed analyte trace to identify data samples in the measured analyte trace that are located outside of the confidence window as being associated with a PISA.

A method for segmenting and denoising a triangle mesh, the method comprising: reading triangle mesh data containing N triangular patches, determining the noise level of the triangle mesh data, and optimizing data at a noise level higher than a preset value; segmenting the triangle mesh data by using a region growing segmentation algorithm, such that a plurality of sub-regions of the triangle mesh data are formed; optimizing the segmented triangle mesh data by using a hole-filling algorithm; and filtering the segmented triangle mesh data by using a denoising algorithm.

The invention relates to the technical field of terminal quality detection, in particular to a wire harness terminal quality detection method and system based on image processing. The method comprises the following steps: acquiring a gray terminal image; calculating a gray level change factor, analyzing angular point features and gradient change features in the neighborhood of each edge line, and calculating a feature extraction factor in combination with the gray level change factor; constructing a weight mapping factor based on the feature extraction factor; using the weight mapping factor to improve an attenuation parameter in an NLM denoising algorithm to obtain an improved attenuation parameter; based on the improved attenuation parameter, using an NLM denoising algorithm to obtain a denoised gray-scale terminal image, and using a neural network to perform quality detection on a terminal in the gray-scale terminal image; the wire harness terminal quality detection efficiency is improved, and the wire harness terminal quality detection effect is improved.

The application provides an audio signal denoising method based on improved variational mode decomposition and a related product. The method comprises the following steps: obtaining a target input audio signal, and using an improved particle swarm optimization algorithm with a local optimal solution jump-out mechanism and a fuzzy entropy search variational mode decomposition to search for a target component number and a target penalty factor; decomposing the target input audio signal according to the target component number and the target penalty factor to obtain a plurality of first target components; calculating the fuzzy entropy of each first target component, and using a wavelet threshold denoising algorithm to denoise the first target components with fuzzy entropy higher than a preset fuzzy entropy threshold to obtain first components; and obtaining a target output audio signal according to the first target components with fuzzy entropy not higher than the preset fuzzy entropy threshold and the first components. The application can solve the problem of poor denoising effect caused by unreasonable parameter setting in the prior art, thereby realizing more efficient and accurate audio signal denoising processing.

The application discloses a point cloud registration method based on a curved surface characteristic region constraint, comprising a point cloud downsampling module, a characteristic point extraction module, a characteristic point description and characteristic matching module, and a registration module based on a curved surface characteristic constraint region, and is used in a registration task with noise and unordered point cloud data. The point cloud downsampling module is subjected to denoising treatment by adopting a denoising algorithm based on KD-tree. The characteristic point extraction module adopts an extraction detection method based on a curved surface change index. The characteristic description and characteristic matching module uses a fast point feature histogram and a random sampling consistency matching algorithm to complete initial pose transformation. The registration module based on the curved surface characteristic constraint region adopts a method of adding a curved surface characteristic region constraint, accelerates the matching rate of icp characteristic point pairs by a KD-tree algorithm, completes point cloud registration, and improves the accuracy of point cloud registration. The application can be applied to model registration tasks in a complex part digitization detection process.

The invention discloses a point cloud denoising optimization method based on no-reference geometric quality evaluation. The method comprises the following steps: constructing a point cloud quality evaluation data set comprising an objective evaluation data set and a subjective evaluation data set; constructing and training a point cloud sorting quality evaluation network, and extracting multi-scale geometric features of the point cloud and predicting a quality score by using a twin network architecture and a sorting learning mechanism; based on the trained evaluation network, constructing a quality-guided denoising optimization framework, and introducing the predicted non-reference quality score as an auxiliary loss function into the training process of the denoising network; and by dynamically adjusting the de-noising intensity, optimizing de-noising network parameters, and outputting a final de-noising point cloud. According to the method, the problem that a traditional evaluation index is inconsistent with human eye visual perception is solved, a non-reference quality evaluation measure is provided, the measure is used for guiding the optimization of a denoising algorithm, and the geometric detail retention capability and the visual perception quality of the denoised point cloud are remarkably improved.

The invention relates to the technical field of agricultural / forestry pest monitoring, in particular to a trunk borer early warning detection method based on audio signal time-frequency feature fusion, which comprises the following steps: acquiring an audio signal generated by boring vibration of trunk borers on the surface of a trunk, sequentially carrying out noise reduction, pre-emphasis, framing and windowing treatment on the audio signal, and sending the audio signal to the trunk borer. Generating a bidirectional logarithmic Mel spectrogram through a bidirectional Mel filter; wherein the noise reduction adopts a wavelet threshold noise reduction algorithm. According to the method, audio frequency domain and time domain features are respectively extracted through the double-branch network, time frequency information complementation is realized after fusion, the complex features of the trunk borer audio can be better captured compared with a single model, the early weak signal recognition rate is improved, the model parameter quantity is reduced to 0.3 M or below through convolutional layer pruning, lightweight convolution (DWConv + GConv) and parameter-free fusion, the calculation amount is greatly reduced, and the method is suitable for large-scale popularization and application. The method can be deployed in an embedded edge device, and the problems of'recalculation power and difficulty in landing 'of a traditional model are solved.

The application discloses a current signal denoising method and system for fault arc detection and a storage medium, comprising: collecting the working current of the load at the current time, and analyzing to obtain the noise estimation spectrum of the working current at the current time; wavelet decomposing the noise estimation spectrum and the working current of the load at the subsequent time respectively to obtain first wavelet coefficients and second wavelet coefficients; correcting the second wavelet coefficients based on the first wavelet coefficients to obtain third wavelet coefficients; and wavelet reconstructing the third wavelet coefficients to obtain a time domain enhanced signal. The improved wavelet threshold function improves the defects of poor continuity and constant deviation of the traditional wavelet threshold function, and the current signal processed by the denoising algorithm not only suppresses the existence of noise, but also improves the arc characteristics of the signal, so that the detection performance of the fault arc detection algorithm is significantly improved.