106 results about "Wavelet reconstruction" patented technology

The invention provides a power distribution network insulator early fault detection method and system considering an interference event. The method comprises the steps of obtaining a zero-sequence current of a power distribution network insulator, and calculating a zero-sequence current mutation rate based on the zero-sequence current; if the zero-sequence current abrupt change rate is greater than a set threshold value, judging that the power distribution network has a fault or a disturbance event; obtaining original recording data of the insulator of the power distribution network, and filtering the original recording data based on a wavelet reconstruction algorithm; for the filtered signal, identifying a peak waveform based on a fault phase voltage extreme point; the non-linear distortion waveform is identified based on zero-sequence current sine fitting, and based on an identification result, when a set first criterion or a set second criterion is satisfied, it is judged that an early fault of the insulator may occur; and calculating the wavelet energy entropy of the signal, if the wavelet energy entropy is greater than a threshold value of the set entropy, determining that the power distribution network insulator has an early fault, otherwise, determining that the power distribution network insulator has an interference event.

The invention relates to the technical field of laser spectrum detection, in particular to an LIBS (Laser-induced Breakdown Spectroscopy) spectrum noise reduction method, system and equipment based on adaptive threshold wavelet transform and a storage medium. Acquiring an original spectral signal of the laser-induced breakdown spectroscopy; performing five-layer multi-layer wavelet decomposition on the original spectral signal by adopting a db4 wavelet basis function to obtain a high-frequency coefficient and a low-frequency coefficient of each layer; calculating a noise intensity standard deviation based on the detail coefficient of the highest decomposition layer; dynamically determining the optimal value of the regulation factor through a double-layer optimization strategy combining a grid search method and a golden section iterative optimization method; constructing an adaptive threshold value based on the noise intensity standard deviation and the adjustment factor; carrying out threshold value processing on the high-frequency coefficient by adopting a self-adaptive threshold value; and performing wavelet reconstruction on the processed high-frequency coefficient and low-frequency coefficient, and outputting a denoised spectral signal. While the LIBS spectral signal-to-noise ratio is remarkably improved, the spectral feature form is completely reserved, and reliable technical support is provided for laser-induced breakdown spectroscopy detection in a complex industrial environment.

The invention is applied to the field of mechanical fault diagnosis signal processing, and particularly provides a pulse neural network mechanical fault diagnosis method based on STFT dimension transformation, and the method comprises the steps: collecting a one-dimensional mechanical vibration signal, carrying out the wavelet decomposition, carrying out the wavelet reconstruction of a low-frequency component and a denoised high-frequency component, and carrying out the wavelet reconstruction of the low-frequency component and the denoised high-frequency component; obtaining a denoised one-dimensional vibration signal; performing short-time Fourier transform, and converting the time-frequency two-dimensional matrix into a time-frequency two-dimensional matrix; inputting the time-frequency two-dimensional matrix into an improved HH threshold neuron model, carrying out Poisson sparse coding on the time-frequency two-dimensional matrix, and only carrying out pulse response on signal significant features; constructing a suprathreshold coding convolutional network with residual connection, inputting a sparse coding matrix, training by adopting an unsupervised learning rule based on STDP, and adaptively adjusting a network synaptic weight; and inputting to a trained above-threshold coding convolutional network, and obtaining pulse emission activity of neurons of an output layer through network forward propagation to determine a fault diagnosis result.

The invention relates to the technical field of power system power distribution network protection, in particular to a self-adaptive current differential protection method for a power distribution network with branch loads. Firstly, current data acquisition is carried out, intelligent electric meters are installed at power distribution network nodes, and timestamps are synchronized. Current phasor estimation is carried out, and the current phasor of each node is obtained through data preprocessing, soft threshold processing, wavelet reconstruction and phasor estimation; then, differential current is calculated, and the vector difference between adjacent nodes is determined; and determining an adaptive threshold according to historical fault data and real-time power grid load. The fault is judged by comparing the differential current with a threshold value, and the fault type is recognized by adopting a mode recognition algorithm. And finally, generating a protection decision based on a result, and taking measures of cutting off branches, selecting a protection mode according to a resistor, adjusting load distribution or limiting current and the like for short circuit, grounding, overload and other faults to improve fault detection accuracy, accurately identify fault types, reasonably generate and execute the protection decision and guarantee safe and stable operation of the power distribution network.

The invention discloses a cell capacity combination model prediction method, which belongs to the technical field of battery management, and comprises the following steps: firstly, performing characteristic correlation analysis, then specifically adopting corresponding data of selected characteristics, and training the data. The method comprises the following steps: acquiring training data, processing the data by utilizing a wavelet transform algorithm, and performing prediction processing by respectively using LightGBM, improved DDPG and XGBoost so as to respectively acquire a prediction error A, a prediction error B and a prediction error C; according to the prediction error A, the prediction error B and the prediction error C, a prediction error quadratic sum reciprocal method is used for analysis to obtain the weight of a single model corresponding to each prediction error; weighting operation is carried out on all the models, so that a combined prediction model of WT + XGBoost + DDPG + LightGBM is formed; and finally, carrying out wavelet reconstruction and reverse normalization processing on the combined prediction model to obtain a prediction result. The technical problems that in the prior art, errors caused by a single prediction model are large, and the prediction model cannot be applied to production practice are solved.

The invention discloses a low signal-to-noise ratio conditional modulation identification method based on wavelet transform and channel attention. The method comprises the following steps: acquiring a received signal; and inputting the received signal into an identification model of a pre-trained neural network, and outputting a classification identification result and classification accuracy. In the recognition model based on signal reconstruction and the neural network, a wavelet threshold estimation module and a wavelet reconstruction module form a signal reconstruction path, a multi-scale feature extraction module and a prediction classification module form a recognition path, and the two paths are mutually enhanced, so that the combination of the advantages of a digital signal processing technology and the neural network is realized; a wavelet threshold estimation module is introduced to predict a denoising threshold through a neural network, and network parameters of the wavelet threshold estimation module are updated according to back propagation; a signal reconstruction path is combined with a recognition path, the interpretability of the network is enhanced, and the accuracy of modulation recognition in a low signal-to-noise ratio scene is remarkably improved by using multi-scale modulation characteristics.

The invention discloses an electric leakage detection method based on wavelet transform harmonic suppression, and the method comprises the steps: synchronously collecting a current signal through a dual-channel differential sampling circuit, and obtaining two paths of differential voltages; after common-mode interference is eliminated through differential amplification and impedance matching, low-pass filtering and digital conversion are carried out; extracting signal multi-scale energy characteristics by using wavelet decomposition, and setting a threshold value to separate leakage low-frequency components from high-frequency noise; a pure electric leakage signal is obtained through inverse wavelet reconstruction, and a high-speed switching circuit compares a preset threshold value to trigger a protection action. According to the scheme, differential sampling, impedance matching and wavelet analysis technologies are combined, the signal-to-noise ratio and the detection precision are effectively improved, the problems that in the prior art, electric leakage signals are prone to interference and slow in response speed are solved, and the reliability and the real-time performance of electric leakage protection are remarkably improved.

The invention relates to the technical field of image processing, in particular to an insulator image enhancement method based on unmanned aerial vehicle infrared image denoising. The method comprises the following steps: preprocessing an original infrared image acquired by an unmanned aerial vehicle, and performing multi-scale decomposition on the preprocessed image by using wavelet transform to obtain a high-frequency detail coefficient and a low-frequency approximation coefficient; a high-frequency detail coefficient is processed by adopting an adaptive threshold value, Wiener filtering is carried out on a low-frequency approximation coefficient, and a de-noised image is obtained after wavelet reconstruction and guide filtering processing; based on the de-noised image, an edge detection operator is adopted to perform preliminary edge extraction, non-maximum suppression and dual-threshold processing are combined to obtain edge information, and a morphological edge reconstruction technology is adopted to connect edge gaps; and generating a final image after feature enhancement through adaptive contrast enhancement based on local temperature distribution and image fusion processing. According to the invention, efficient denoising and accurate feature enhancement of the insulator infrared image collected by the unmanned aerial vehicle can be realized.

The invention relates to a drill rod fatigue crack ultrasonic weak signal enhancement method based on sparse wavelet reconstruction, and belongs to the field of drill rod crack nondestructive testing. A multi-array-element ultrasonic data test is adopted for detecting the defect state of the drill rod, target features are enhanced through wavelet denoising result analysis, and crack recognition features are obtained based on a self-adaptive threshold value. The method comprises the following steps: carrying out wavelet decomposition and threshold processing on a detected signal, removing a noise signal, and reconstructing a wavelet coefficient of each layer after decomposition so as to obtain a de-noised ultrasonic detection signal. According to the method, the problem of weak crack signal decoupling under the strong noise background is solved, and accurate positioning and quantitative evaluation of the root defect of the drill rod thread are realized, so that the reliability of coal mine underground drill rod microcrack detection is improved, and a core technical support is provided for constructing an intelligent gas extraction safety monitoring system.

The invention discloses an esophageal cancer nuclear magnetic diagnosis enhancement system based on artificial intelligence, which belongs to the field of image processing and comprises an image acquisition module, a nuclear magnetic image optimization module, an esophageal cancer nuclear magnetic diagnosis model building module and a diagnosis enhancement module. The method specifically comprises the steps of performing multi-scale wavelet decomposition, designing an initial position of chaotic sine correlation mapping, constructing two neighborhoods based on a historical optimal position, completing updating of an individual position in combination with cosine similarity segmentation, finding an optimal threshold value, performing threshold value processing on a detail coefficient, and performing wavelet reconstruction to obtain an optimized nuclear magnetic image. Carrying out primary feature extraction to obtain a multi-scale feature map, carrying out two-time different aggregation on advanced features to extract global features, extracting local features from the first feature map, carrying out selective fusion on the global features and the local features to obtain final features, and outputting a diagnosis result to improve the quality of the nuclear magnetic image. And the accuracy and reliability of esophageal cancer diagnosis are further improved.

The invention discloses an edge-guided self-adaptive fusion remote sensing image change detection method based on spatial frequency domain interaction. The method comprises the following steps: preparing data of front and rear remote sensing images and real change labels; preprocessing based on the previous and later remote sensing images and labels to obtain a standardized data set; an edge-guided adaptive fusion network based on spatial frequency domain interaction is constructed, feature enhancement is carried out through a spatial and frequency double-branch structure, and an adaptive orthogonal fusion module is designed to carry out multi-construction interactive fusion on double-time features. Introducing an edge driving wavelet reconstructor as an auxiliary supervision branch to guide the feature change region edge modeling capability; and finally, a high-precision change prediction map is generated by inputting a double-time-phase remote sensing image. The method aims at solving the problems that edge prediction details are insufficient and background noise is difficult to restrain in a traditional change detection technology, the detection performance is effectively improved, the advanced change detection performance is shown on a data set, and the method has high competitiveness and practical value.

The invention discloses a cloud resource use prediction method and system based on wavelets and an attention mechanism. The method comprises the following steps: firstly, acquiring a multi-dimensional resource use sequence of a target server and dividing the multi-dimensional resource use sequence into a training subset and a test subset; generating an input sample set by adopting a sliding window mechanism; then constructing a cloud resource prediction model comprising a wavelet attention network, a multi-attention module, a wavelet reconstruction module and a memory enhancement self-attention module; decomposing the time window sample into a low-frequency feature component and a high-frequency feature component through a wavelet attention network; respectively generating a low-frequency feature vector and a high-frequency feature vector by using a multiple attention module; fusing into a reconstructed sample sequence through a wavelet reconstruction module; generating enhanced feature representation through a memory enhanced self-attention module; and finally, outputting a cloud resource use prediction sequence in a future time period through a full-connection projection layer. According to the method, the multi-frequency-domain time sequence characteristics can be effectively separated, the interaction relationship among the multi-element resources can be modeled, and the prediction precision and generalization ability are remarkably improved.

The present application relates to the technical field of image processing, in particular to an insulator image enhancement method based on infrared image denoising of a UAV. The original infrared image collected by the UAV is preprocessed, and the preprocessed image is decomposed in multiple scales by wavelet transform to obtain high-frequency detail coefficients and low-frequency approximation coefficients; the high-frequency detail coefficients are processed by adaptive thresholding, while the low-frequency approximation coefficients are subjected to Wiener filtering, and after wavelet reconstruction and guided filtering, a denoised image is obtained; based on the denoised image, an edge detection operator is used for preliminary edge extraction, edge information is obtained by combining non-maximum suppression and double thresholding, and the edge gap is connected by using morphological edge reconstruction technology; through adaptive contrast enhancement based on local temperature distribution and image fusion processing, the final image with enhanced features is generated. The present application can realize efficient denoising and accurate feature enhancement of the insulator infrared image collected by the UAV.

The invention belongs to the field of power distribution network fault detection, and provides a power distribution network fault traveling wave polarity discrimination method, which comprises the steps of wavelet decomposition, threshold value processing, wavelet reconstruction, differential absolute value calculation, filtering threshold value calculation, first maximum point extraction and polarity determination. According to the method, polarity judgment is performed by comparing the data subjected to wavelet denoising and differential processing with 0, so that the influence of the non-stationarity of the original data on the waveform polarity can be effectively eliminated, and the polarity judgment accuracy is improved.

The present invention discloses an improved wavelet denoising method based on a Sagnac fiber acoustic sensing system. The method includes: acquiring an original sound signal; decomposing the original sound signal by using a wavelet denoising method to obtain a decomposed sound signal; filtering the decomposed sound signal by using an improved wavelet threshold denoising algorithm, and then performing wavelet reconstruction to obtain a filtered noisy sound signal; and performing low-pass filtering on the filtered noisy sound signal to obtain a noise-reduced sound signal. The present invention can improve the quality of the acquired sound signal.

The present invention discloses a low dry-to-noise ratio interference detection method and a positioning method based on a multi-beam double-star system, including: determining the interference frequency according to the main satellite interference, and establishing an interference model of the main satellite interference and the secondary satellite interference based on this; performing matched filtering on the interference forwarded by the two multi-beam satellites to the ground receiving station to improve the dry-to-noise ratio between the interferences; performing wavelet decomposition on the interference after matched filtering to obtain wavelet coefficients of different scales and frequencies; then determining a threshold, performing threshold processing on the wavelet coefficients to remove the noise coefficients and retain the interference coefficients. Using the retained interference coefficients for wavelet reconstruction to obtain the denoised interference. The present invention breaks through the double-star positioning limitation condition that the double stars need to jointly cover the interference source, and can realize the positioning of the interference source in a vast area, providing an effective tool for the positioning and troubleshooting of the interference source in the multi-beam satellite communication system.

The present invention discloses a multi-energy consumption data completion method based on wavelet decomposition and Fourier transform, which includes the following steps: 1. performing wavelet decomposition on the collected carbon-containing characteristic energy consumption sequence to obtain an energy consumption cycle characteristic sequence and an energy consumption trend characteristic sequence; 2. obtaining a predicted energy consumption trend characteristic sequence based on curve fitting; 3. obtaining a predicted energy consumption cycle characteristic sequence based on Fourier series fitting; and 4. completing missing data through wavelet reconstruction based on steps 2 and 3. The present invention constructs a data completion model based on wavelet decomposition and Fourier transform, thereby enabling the completion of massive multi-energy consumption data for key emission-controlled enterprises.

The invention belongs to the technical field of petroleum and natural gas exploration and development, and discloses a method and a device for engraving a tiny fracture geologic body, electronic equipment and a medium, the method for engraving the tiny fracture geologic body comprises the following steps: based on a multi-wavelet decomposition result, carrying out wavelet energy optimization sorting and stripping a strong energy quantum wave, and then completing multi-wavelet reconstruction; seismic signals related to the tiny fracture system are highlighted, and the spatial form of the tiny fracture geologic body is obtained through engraving. The method effectively improves the spatial form engraving precision of the tiny fractured geologic body, and is suitable for spatial engraving of the tiny fractured geologic body.

This invention discloses a baseline removal method for multi-channel magnetocardiogram (MCC) signals based on wavelet low-frequency reconstruction and morphological secondary smoothing. The invention includes: acquiring multi-channel MCC signal data; performing power frequency notch filtering on each channel to suppress the power frequency fundamental frequency and at least one harmonic component; performing configurable IIR low-pass filtering on the notched signal; setting a discrete wavelet transform extension mode and performing discrete wavelet decomposition on the filtered signal; selectively retaining decomposition coefficients based on the retention layer set and setting the remaining coefficients to zero, then performing inverse wavelet reconstruction to obtain an initial baseline estimate; performing morphological secondary smoothing on the median-smoothed baseline signal to obtain a corrected baseline signal; subtracting the corrected baseline signal from the filtered signal to obtain the baseline-corrected channel signals; and assembling and outputting a multi-channel baseline-corrected signal matrix. This invention can improve the stability and cross-channel consistency of baseline estimation for multi-channel MCC signals, reduce the risk of over-correction and under-correction, and is applicable to multi-channel batch processing scenarios.

The invention discloses a method and a system for measuring trace ammonia by using an optical means, and belongs to the technical field of ammonia concentration measurement, and the method is characterized by comprising the following steps: acquiring a Raman scattering light signal and a photoacoustic signal of an ammonia-containing sample to be measured; reconstructing the Raman scattering light signal to obtain a first reconstructed signal, and reconstructing the photoacoustic signal to obtain a second reconstructed signal; according to the method, the problem of signal noise interference is effectively solved based on wavelet reconstruction, the influence of environmental factors on a measurement result is considered in real time through a Bayesian causal analysis method, and the accuracy of the measurement result is effectively judged.

The embodiment of the application provides a cascade anti-interference circuit suitable for an electrocardio monitoring device, which comprises a sliding mean filter module, a notch filter module, a lifting wavelet decomposition module, a threshold calculation module, a threshold processing module and a lifting wavelet reconstruction module; while ensuring a small circuit scale, the ECG signal collected by the wearable electrocardio monitoring device is subjected to hierarchical noise reduction processing, and high signal-to-noise ratio ECG signal output is realized; for the ECG signal collected by the wearable electrocardio monitoring device, first, sliding mean filtering is performed to filter out baseline drift noise caused by human respiratory movement and other activities; the power frequency interference caused by the wired and wireless connection of the electromagnetic environment around the device is suppressed by using the notch filter module; finally, wavelet noise reduction is realized by using the lifting wavelet decomposition and reconstruction and threshold noise reduction method, the electromyographic interference caused by the autonomous or unconscious movement of the wearer is suppressed, and high signal-to-noise ratio ECG signal is output, which is used for physiological parameter extraction and electrocardio diagnosis.

This invention provides a power load forecasting method based on time-frequency enhancement and multi-level wavelet modeling, comprising: constructing a multivariate historical power load time series and generating instance normalized sequences according to variable channels; performing time-domain enhancement and spectral enhancement on the instance normalized sequences to obtain enhanced feature sequences; performing multi-level discrete wavelet decomposition on the enhanced feature sequences to obtain a first-level approximation coefficient sequence and a detail coefficient sequence of wavelet coefficients at each level; performing independent resolution branch modeling based on the decomposition coefficient sequences, modeling and predicting the coefficient sequences corresponding to each resolution branch, obtaining predicted approximation coefficient sequences and predicted detail sequences at each level, and then performing multi-level wavelet reconstruction; and performing inverse normalization on the reconstruction results to obtain the power load forecasting results. Applying this method can coordinate time-domain and frequency-domain feature enhancement with multi-resolution modeling, improving the accuracy and stability of long-term power load forecasting while reducing computational complexity.

This invention relates to a wavelet threshold-based method and system for denoising cold test vibrations, comprising the following steps: constructing a vibration parameter database based on the acquired diesel engine cold test vibration signal; performing wavelet decomposition on the vibration parameters in the database according to the selected wavelet basis function and decomposition level to obtain wavelet coefficients; performing wavelet reconstruction on the vibration parameters based on a threshold function to obtain denoised cold test vibration parameters, specifically: shrinking the wavelet coefficients within the threshold range to zero, while keeping the wavelet coefficients outside the threshold unchanged; and processing the acquired diesel engine cold test vibration signal using the denoised cold test vibration parameters to obtain a denoised cold test vibration signal. The threshold function is used to solve the oscillation and distortion problems existing in the denoising process, thereby improving the denoising effect of the cold test vibration signal.

The invention discloses an image noise reduction method and device, equipment and a medium, which are used for realizing multi-channel wavelet high-frequency noise reduction. The method provided by the invention comprises the steps of performing wavelet decomposition on a to-be-denoised image to obtain a plurality of high-frequency components, and determining a plurality of high-frequency channel vectors; each high-frequency channel vector is obtained by combining high-frequency components of a plurality of channels of the to-be-denoised image at the same image coordinate position; for each high-frequency channel vector, determining a similar vector of the high-frequency channel vector, and generating a high-frequency similar vector matrix; on the basis of the high-frequency similar vector matrix, determining a parameter for performing noise reduction on the high-frequency channel vector; performing noise reduction processing on the high-frequency channel vector by using a parameter for performing noise reduction on the high-frequency channel vector to obtain a high-frequency channel vector after noise reduction; and on the basis of each denoised high-frequency channel vector, performing wavelet reconstruction by adopting a mode corresponding to wavelet decomposition to obtain a denoised image.

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.

The invention discloses a cardiac mechanical signal processing method and device based on wavelet reconstruction. The method comprises the following steps: performing trend removal and frequency band extraction on an original SCG signal, and reserving main mechanical energy components corresponding to a heart sound event; then wavelet decomposition and main scale reconstruction are carried out, and time domain features of key components such as S1 and S2 structures similar to PCG signals are strengthened, so that heart sound-like signals which are similar to PCG in form and clear in rhythm are generated. Compared with an original SCG signal, the heart sound-like signal obtained through reconstruction is remarkably improved in the aspects of rhythm definition, form consistency and peak stability. And further performing rhythm analysis on the reconstructed signal, introducing physiological constraints, and screening to obtain a cardiac trigger point sequence, thereby realizing high-precision cardiac recognition. The device can be integrated with medical image equipment and wearable equipment, and the functions of heart sound gating, synchronous triggering or auscultation auxiliary analysis and the like are achieved.

The present invention discloses a method for resisting mainlobe interference based on signal separation, comprising: performing de-skewing processing on an echo signal received by a radar system to obtain the real and imaginary parts of the de-skewing signal; performing L-layer wavelet decomposition on the real and imaginary parts of the de-skewing signal to obtain M sub-modes of the de-skewing signal; calculating the matching degree of the M sub-modes with the radar matching coefficient, and selecting the sub-mode corresponding to the target signal component; using the sub-mode corresponding to the target signal component as the mode to be reconstructed, and obtaining a reconstructed signal through wavelet reconstruction; performing an inverse de-skewing operation on the reconstructed signal to obtain a reconstructed echo signal. Aiming at the situation of low interference-to-signal ratio and low signal-to-noise ratio, the present invention divides the echo into interference and target components according to wavelet decomposition and retains the target component to reconstruct the echo, thereby removing the interference components entering the radar mainlobe, ensuring the integrity of the target information as much as possible, and reducing the energy loss of the target.

The present application relates to the technical field of fault detection, and relates to a relay contact fault identification method based on wavelet denoising and a support vector machine, comprising the following steps: S1, constructing a current signal acquisition and double-channel reconstruction circuit: a Rogowski coil is sleeved on an output loop of a relay contact, and a differential voltage signal is inducted and output by the Rogowski coil; S2, respectively adopting a first wavelet base and a second wavelet base to perform multi-scale wavelet decomposition, threshold denoising and wavelet reconstruction on a complete current digital sequence and a transient current digital sequence; S3, extracting time domain statistical features from the complete current digital sequence after denoising, extracting time-frequency energy features from the transient current digital sequence after denoising, and outputting a fault category of the relay contact after classification and decision by a support vector machine classifier. The present application overcomes the inherent contradiction that a single acquisition channel is difficult to simultaneously consider slow-changing contact components and transient disturbance components in terms of dynamic range and denoising strategy.

The invention belongs to the technical field of oil abrasive particle monitoring, and particularly relates to an abrasive particle feature extraction method based on differential signal band selection adaptive filtering, which comprises the following steps: carrying out common-mode rejection on an oil abrasive particle signal to obtain a differential signal to be detected; performing wavelet packet decomposition on the differential signal to be detected through an orthogonal wavelet basis function to obtain a multi-decomposition-scale feature vector; calculating a similarity index on each decomposition scale, and extracting a target scale feature vector; processing the target scale feature vector by adopting an improved minimum root mean square adaptive filter to obtain a filtering vector; performing wavelet reconstruction on the zero vector and the filtering vector to obtain a noise reduction signal; dividing the noise reduction signal into a plurality of segments, and calculating a composite identification index of each segment; calculating the weight of each segment according to the composite identification index, and weighting all segments to obtain an oil abrasive particle characteristic signal extraction result; according to the invention, the signal-to-noise ratio of the abrasive particle signal is improved, the self-adaptive capability is stronger, and the accurate detection of the oil abrasive particle characteristics is improved.

The application discloses a power load prediction method and system based on multi-stage time sequence preprocessing and double-branch wavelet Mamba, and relates to the technical field of intelligent power grids and time sequence analysis. In view of the problems that the existing power load prediction technology has insufficient long sequence modeling capability, low prediction robustness and precision, and cannot simultaneously consider global trend fitting and local mutation capturing when facing high-noise non-stationary data, a high-quality target sequence is first constructed through multi-stage time sequence preprocessing, and after channel independence and block embedding processing, the high and low frequency characteristic components are decoupled through discrete wavelet transform, the global long-range dependence is extracted through a bidirectional trend Mamba module, the local mutation characteristics are purified through a detail Mamba module, and noise is suppressed, and finally, the prediction result is output through inverse wavelet reconstruction. The application significantly improves the precision and robustness of long sequence power load prediction and reduces the computational complexity.