33 results about "Generalized s transform" patented technology

The invention provides a cast-in-place pile quality online monitoring method and system based on an ultrasonic array, and the method comprises the steps: transmitting and receiving ultrasonic waves through the ultrasonic array arranged in a cast-in-place pile, and obtaining a multi-channel sound wave signal; performing generalized S transformation processing on the multichannel sound wave signals to obtain a time-frequency matrix, and calculating a cross correlation coefficient between each channel signal and the defect-free reference signal; an amplitude matrix and an instantaneous phase gradient matrix are calculated based on the time-frequency matrix, neighborhood amplitude energy and a phase consistency factor are fused in a time-frequency domain, and a defect energy indication diagram of the pile body section is generated; calculating the energy entropy of the defect energy indication diagram in a full frequency band; and extracting an actually measured energy arrival time sequence, comparing the actually measured energy arrival time sequence with a theoretical arrival time sequence to construct a space distortion vector, and setting a threshold value by combining energy entropy to judge defects.

The invention belongs to the field of power distribution network traveling wave echo analysis, and particularly discloses a power distribution network traveling wave echo identification method based on time-frequency enhancement and multi-evidence fusion. The method comprises a closed-loop process of adaptive frequency band generalized S transformation, low-rank SVD time-frequency enhancement, double-threshold hysteresis starting point detection, physical template GLRT discrimination, double-peak fixed interval scanning and multi-band consistency rechecking. Firstly, a clear time-frequency energy diagram is obtained through preprocessing and self-adaptive frequency band S transformation, then power frequency and a slowly-changing background are suppressed through low-rank reconstruction, and then a candidate starting point is locked on a time energy track by adopting a hysteresis mechanism of'low-threshold triggering-high-threshold confirmation '; carrying out GLRT statistical discrimination in the candidate window by using single-peak / double-peak / damped ringing three types of physical templates, and quantifying time delay between peaks through fixed interval scanning when branch reflection exists; and finally, taking the energy consistency of the low / medium / high bands as cross-band physical evidence, and eliminating false detection only caused by high-frequency ringing.

The invention discloses a section positioning method for a latent fault of an active medium-voltage distribution cable, relates to the technical field of power distribution network fault positioning, and solves the problems of short duration, small fault energy and difficulty in fault positioning of the latent fault of the cable in the prior art. The method comprises the following steps: firstly, detecting whether a latent fault occurs in real time, if so, collecting 1 / 4 period zero-sequence current data of each monitoring point of a fault line after the fault, and performing generalized S transformation; then, zero-sequence current transient energy value features, zero-sequence current transient energy polarity features and zero-sequence current amplitude time-frequency matrix correlation features of the sections are extracted; based on the three-dimensional coordinates of the sections, Euclidean distances among the sections are calculated, and a correlation matrix is established; and finally, calculating a comprehensive distance similarity coefficient between each section and other sections, and judging a fault section based on the comprehensive similarity coefficient of each section.

The invention belongs to the technical field of geophysical signal processing and mine micro-seismic monitoring, and relates to a micro-seismic signal first arrival pickup method based on sparse generalized S transformation. Aiming at the problems of low signal-to-noise ratio, complex noise type, weak first-arrival wave energy and difficulty in accurately picking up first arrival of mine micro-seismic signals, the method comprises the following steps of: firstly, performing time-frequency analysis on the micro-seismic signals by utilizing sparse generalized S-transform, then extracting a maximum energy point in a time-frequency domain, performing connected domain decomposition on a time-frequency spectrum, screening a time-frequency connected region containing main energy, and finally extracting the maximum energy point from the time-frequency domain; and irrelevant noise interference is suppressed. And on the basis, performing energy superposition and normalization processing on the main energy time-frequency block in a constraint frequency band, and combining an energy peak position and an energy transition characteristic in a time window to realize accurate judgment of the arrival time of the microseismic signal first arrival wave.

The invention relates to the technical field of seismic data processing, in particular to an angle gather random noise adaptive suppression method based on generalized S transformation. The method comprises the following steps: selecting a seismic signal in angle gather data to carry out generalized S transformation time window adjustment parameter optimization to obtain a time window parameter under a CM optimal criterion; calculating a corresponding generalized S transform time-frequency spectrum for angle gather data under the same elastic parameter and different angles by using the obtained time window parameter; performing superposition, normalization and threshold calculation on the obtained time-frequency spectrum to obtain a corresponding time-frequency domain adaptive filter; performing point multiplication on the time-frequency spectrum of each seismic signal and the time-frequency domain adaptive filter to obtain a filtered time-frequency spectrum; and inverse transformation is carried out on the filtered time-frequency spectrum of each seismic signal, and random noise adaptive suppression of angle gather data is completed. According to the method, the problem of parameter selection in a traditional model-driven denoising algorithm is avoided by utilizing an adaptive algorithm, manual intervention is reduced, the analysis cost is reduced, and the processing efficiency is improved.

The invention provides a coal measure strata tight sandstone gas earthquake identification method and device, and relates to the technical field of earthquake gas reservoir prediction, and the method comprises the steps: carrying out the coal seam multiple suppression processing of the seismic data of a target work area, and obtaining the preprocessed seismic data; generalized S transformation is carried out on the preprocessed seismic data to obtain time-frequency spectrum data; extracting a plurality of gas layer sensitive time-frequency attributes from the time-frequency spectrum data; and performing multi-attribute well control fusion processing on the plurality of gas reservoir sensitive time-frequency attributes to obtain a gas reservoir sensitive factor, and based on the gas reservoir sensitive factor, generating an earthquake identification map for identifying the tight sandstone gas of the coal measure strata. According to the method, the data signal-to-noise ratio is increased by pressing coal bed interlayer multiples, multi-angle gas bed sensitive time-frequency attributes are extracted through high-resolution time-frequency analysis, a visual recognition map is generated through well control fusion, the problem that tight sandstone gas weakness response is difficult to recognize due to coal bed strong shielding is systematically solved, and the method has the advantages of being high in practicability and the like. And the identification precision and reliability are improved.

The application provides a time-varying coherence estimation method of ground motion based on Bayesian collaborative optimization. The method constructs double target constraints covering physical consistency and spectrum graph distinguishability through generalized S transform and time-frequency bidirectional exponential smoothing method, cooperatively determines the optimal combination of generalized S transform kernel parameters and exponential smoothing operator by using a Bayesian optimization algorithm, and outputs an optimal parameter set and a corresponding time-varying coherence spectrum graph. Therefore, more real data reflecting the spatial variation characteristics of ground motion can be obtained, and the design requirements under the limit adverse state can be met.

The application discloses a kind of sparse generalized W transform gas reservoir characterization time-frequency analysis method, for the first time L1 norm is introduced to the mathematical relationship between generalized W transform and seismic signal for sparse constraint, and is solved using Bregman iteration algorithm, to obtain a kind of analysis result with higher time-frequency resolution.The method absorbs the advantage that generalized W transform highlights low-frequency information of seismic signal, avoids the problem of main frequency splitting, and provides a more sparse time-frequency representation for non-stationary seismic signals. When applied to the time-frequency analysis of actual seismic data, it can provide a high-precision seismic spectral decomposition result for gas reservoirs, thereby more accurately delineating the low-frequency abnormal area of gas reservoir.

The invention relates to the field of electrical equipment fault diagnosis and safety protection calculation, in particular to a series arc fault and reignition detection method based on time-frequency characteristics. The method comprises the following steps: collecting and preprocessing a line current signal; performing high-resolution time-frequency analysis on the signal by adopting generalized S-transform, and extracting time-frequency characteristics in a frequency band; constructing a two-dimensional feature vector; and global optimization is carried out on the support vector machine hyper-parameters by using a grey wolf optimization-particle swarm hybrid algorithm, a classification model is established, and accurate identification of three working conditions of a normal state, a fault arc and contact arc reignition is realized. According to the method, the classification model is established, the defects of a traditional method in the aspect of distinguishing instantaneous reignition and continuous fault arcs are overcome, the method has the advantages of being high in recognition accuracy, high in anti-interference capacity, clear in physical significance and the like, and a reliable technical scheme is provided for early fault early warning and intelligent protection of an electrical system.

The invention provides a fracture detection method based on sub-azimuth frequency variation inversion, and the method comprises the steps: building a suitable HTI medium shale equivalent model based on the geological data of shale through employing a self-consistent theory (SCA), a differential equivalent medium theory (DEM) and a Chapman fracture theory; a shale rock physical model is used for researching the influence of fracture density and gas saturation on the dispersion and attenuation of a shale reservoir, and sensitive parameters are optimized to construct a rock physical interpretation quantity version of a dispersion attribute; carrying out azimuth processing and ellipse fitting on the original OVT data to obtain a dominant azimuth of fracture development, and carrying out time-frequency analysis on pre-stack seismic data based on an inversion spectrum decomposition method of generalized S transformation to obtain near, medium and far offset data of different frequencies; and performing frequency variation inversion by using a further derived Ruger approximation formula to obtain a frequency dispersion attribute, and predicting the fracture density and the gas saturation through a rock physical quantity version. The method can be used for predicting the fracture density and the gas saturation of the shale, and the prediction precision is higher.

The embodiment of the invention provides a cable defect positioning method and device, equipment, a storage medium and a program product. The method comprises the following steps: generating a Gaussian envelope linear frequency modulation excitation signal through a signal generator; determining a Gaussian envelope linear frequency modulation excitation signal as an incident signal of the cable to be detected; injecting the incident signal into the to-be-detected cable to obtain real-time waveform information of the incident signal and real-time waveform information of the reflected signal of the to-be-detected cable; determining adaptive parameters; respectively carrying out discrete generalized transformation calculation on the real-time waveform information of the incident signal and the real-time waveform information of the reflected signal according to the self-adaptive parameters to obtain a time-frequency distribution result of the incident signal and a time-frequency distribution result of the reflected signal; and determining the defect position of the to-be-detected cable according to the time-frequency distribution result of the incident signal and the time-frequency distribution result of the reflected signal. According to the method, the purposes of high response speed and high positioning precision in the cable defect positioning process can be achieved.

This invention addresses the limitations of single signal sources in containing limited fault information, failing to comprehensively reflect the operating status of centrifugal fire pumps, and the low accuracy of fault diagnosis in centrifugal fire pumps due to the difficulty of dynamically and adaptively differentiating feature weights during fault diagnosis using convolutional neural networks (CNNs). To address these issues, a fault diagnosis method for centrifugal fire pumps based on multi-source information fusion and an improved EfficientNet-b0-TL model is proposed. Vibration signals from centrifugal fire pumps are collected from different angles and locations using multiple sensors, and the signals are cleaned, aligned, and denoised. A generalized S-transform is used to perform time-frequency transformation on the denoised signals, and time-frequency features are extracted and fused. A CBAM module is introduced to improve the EfficientNet-b0-TL model, constructing a fault diagnosis model. This model is then applied to actual centrifugal fire pump experimental data, and the results show that this method has high fault diagnosis accuracy.

The invention discloses a coherent MIMO radar multi-channel accumulation detection method. According to the method, firstly, single-channel signals are accumulated through generalized Radon-Fourier transform (GRFT), then a virtual direction vector is introduced to replace a traditional direction vector, inter-channel differences and target motion parameters are effectively decoupled, and therefore inter-channel envelope differences and phase differences can be estimated in a lower dimension. And based on the estimated difference information, constructing an envelope alignment function and a phase compensation function, and finally realizing coherent accumulation of the multi-channel signals. According to the method, the echo signals of the high-speed maneuvering target can be effectively accumulated under the condition of low signal-to-noise ratio, and the calculation complexity is remarkably reduced.

The application discloses a high-speed maneuvering target detection method based on deep learning and generalized Radon-Fourier transformation, and the method comprises the following steps: obtaining three-dimensional data through echo signal preprocessing; inputting the three-dimensional data into a target / noise classification network to obtain the probability of existence of a target in each distance unit; inputting the data of the distance unit in which the target may exist into a velocity acceleration classification network to output the search interval of the velocity and acceleration of the target; accumulating energy in the search interval by using generalized Radon-Fourier transformation; and finally obtaining a detection result through constant false alarm detection. The application combines the neural network and the generalized Radon-Fourier transformation, and can effectively balance the detection performance and the calculation amount. The application has a high accuracy in estimating the distance unit where the target is located and the motion parameters of the target; in addition, unnecessary search operations can be reduced according to the rough estimation result of the network, so that the calculation amount is greatly reduced.

The application discloses a ship scattering point three-freedom rotation micro-Doppler frequency extraction method, belongs to the technical field of radar signal processing and target identification, is used for micro-Doppler frequency extraction, and comprises the following steps: constructing a ship motion geometric model, and constructing a three-component amplitude modulation and frequency modulation micro-Doppler frequency model; performing generalized S transformation on a one-dimensional echo in the azimuth direction of a range cell, synchronously compressing the generalized S transformation in combination with a time-frequency track and an instantaneous frequency operator, and calculating a micro-Doppler frequency estimation value; constructing an adaptive ridge filter, decomposing the micro-Doppler frequency estimation value to obtain a micro-Doppler frequency estimation value subcomponent, and linearly superimposing the subcomponent to obtain an unoptimized micro-Doppler frequency estimation value. Through the construction of the three-freedom rotation micro-Doppler frequency model and the synchronous compression generalized S transformation and adaptive ridge filter method, high-precision and high-resolution extraction of the rotation micro-Doppler frequency of the ship scattering point is realized, and the problems of multi-component time-frequency cross and noise interference are effectively solved.

The invention discloses a stratum quality factor Q estimation method based on a centroid frequency movement method. The method has the main functions of improving the signal-to-noise ratio of post-stack data and performing reservoir prediction according to seismic attenuation characteristics. And the time-frequency domain is focused for analysis to obtain a more accurate amplitude spectrum result. According to the method, an improved frequency shift method based on generalized S transformation is applied to attenuation estimation of post-stack seismic data, and the frequency resolution is increased as much as possible on the basis of ensuring that the time resolution can identify a target layer. According to the requirements of actual data, by adjusting parameters lambda and p under generalized S transformation, a more appropriate frequency resolution and a more accurate amplitude spectrum result are obtained, the seismic data resolution is improved, and finer interlayer information is identified. From the perspective of extracting the post-stack profile Q value model, the improved frequency shift method based on generalized S transformation can establish a relatively accurate underground Q value model, and the Q value model can be applied to the post-stack profile through conventional inverse Q filtering processing.

The invention belongs to the field of power distribution network traveling wave head analysis, and particularly relates to a traveling wave head extraction method based on multi-scale time-frequency enhancement and low-rank sparse optimization. According to the method, starting from two aspects of time-frequency transformation precision improvement and feature enhancement denoising, high-resolution time-frequency distribution is obtained by adopting adaptive window width wide-sense S transformation, and the resolvability and robustness of a wave head mutation point are remarkably improved through multi-scale sparse-low rank joint optimization; therefore, stable and accurate traveling wave head detection can still be realized under the condition of low signal-to-noise ratio, and the fault positioning accuracy is finally improved.

The present application relates to the technical field of geophysical exploration, in particular to a pre-stack OVT domain low-order fault identification method based on azimuth and frequency domain control. The method comprises the following steps: performing pre-stack OVT domain gather anisotropy analysis; analyzing target area tectonic stress characteristics, performing OVT domain gather data azimuth optimization and division, and performing partial data stacking on OVT gathers; constructing data volumes controlled by different azimuths and frequency domains through improved generalized S transform spectral decomposition; extracting coherent attribute analysis; and drawing a target area fault distribution plan. The method can effectively improve the accuracy and reliability of low-order fault interpretation, meet the needs of fine fault system interpretation at different exploration and development stages, and provide reliable basis for oilfield exploration and development target optimization and drilling and completion scheme design.

The invention relates to a well constraint-based seismic data frequency division weighted fusion optimization method. The method comprises the following steps of: processing seismic data by using high-resolution generalized S transform to obtain broadband seismic data; processing the broadband seismic data to obtain a frequency division seismic data set; according to the sound wave and density data of the actual drilling well and the dominant frequency of the seismic channel beside the well, a synthetic seismic record is obtained through convolution calculation; a single well optimal frequency division weighted fusion coefficient is calculated according to the synthetic seismic record and the well-side seismic trace; processing the optimal frequency division weighted fusion coefficient of each well to obtain an optimal frequency division weighted fusion coefficient distribution matrix; and processing the frequency division seismic data set according to the optimal frequency division weighted fusion coefficient distribution matrix to obtain optimized seismic data. The method can effectively solve the problems that an existing seismic data fusion optimization method is insufficient in quantification, high in multiplicity of solutions and low in precision, and has an obvious effect on improving seismic data quality and improving oil and gas field description precision.

The invention discloses a fault type detection method for a planetary gearbox, and the method comprises the steps: collecting a vibration signal of the planetary gearbox, and obtaining a time-frequency characteristic graph through the optimal generalized S transformation; obtaining a time-frequency enhancement feature map through a time-frequency enhancement module, and inputting the time-frequency enhancement feature map into a convolutional neural network to extract a convolutional feature map; constructing a sample pair set in the convolutional feature map, and calculating a consistency score and measurement loss through a feature measurement module; and inputting the convolution feature graph into a fault classification module to obtain a fault classification probability, calculating classification cross entropy loss, fusing the classification cross entropy loss and measurement loss into total loss, and performing back propagation optimization on parameters of each module. According to the invention, fault features are enhanced through the time-frequency enhancement module, feature distribution is optimized through the feature measurement module, end-to-end diagnosis is realized in combination with the convolutional neural network, dynamic interference can be effectively inhibited, the accuracy and stability of fault diagnosis are improved, the method adapts to complex working conditions of large-scale equipment in the oil and gas industry, and safe production of well drilling is guaranteed.

The invention discloses a time-frequency domain cross-correlation reconstruction method and system for a rock breaking source of a drill jumbo, and relates to the technical field of seismic signal processing and intelligent detection, and the method comprises the steps: collecting the seismic signals of the rock breaking source in the drilling process of the drill jumbo, including a pilot signal and a side wall signal; respectively carrying out generalized S transformation on the pilot signal and the side wall signal to obtain respective time-frequency domain representation; performing sliding cross-correlation on the two time-frequency domain representations under set time delay to obtain a cross-correlation spectrum; and the cross-correlation spectrum is recovered to a time domain through inverse transformation of generalized S transformation, and an enhanced seismic signal is obtained. Based on generalized S transformation and time-frequency cross-correlation, underground reflection information is effectively extracted, and reconstruction of seismic source wave field features while drilling is realized.

The invention relates to a cable defect positioning method, device and equipment and a medium, and relates to the technical field of defect detection. The method comprises the following steps: transforming a reflection coefficient spectrum of a cable based on generalized S transformation estimation to obtain a time-frequency spectrum of the cable; performing spectrum peak screening and false peak elimination on the time-frequency spectrum to obtain a target signal of the cable; performing frequency domain summation on the target signal to obtain a positioning curve of the cable; and determining the defect position of the cable based on the positioning curve and the signal propagation speed. By adopting the method, accurate positioning of cable defects can be realized.

This invention discloses a three-degree-of-freedom rotational micro-Doppler frequency extraction method for ship scattering points, belonging to the field of radar signal processing and target recognition technology. It is used for micro-Doppler frequency extraction, including constructing a ship motion geometric model and a three-component amplitude-modulated and frequency-modulated micro-Doppler frequency model; performing a generalized S-transform on the azimuth one-dimensional echo of the range cell, and simultaneously compressing the generalized S-transform using time-frequency trajectories and instantaneous frequency operators to calculate the micro-Doppler frequency estimate; constructing an adaptive ridge filter to decompose the micro-Doppler frequency estimate into sub-components, and linearly superimposing the sub-components to obtain the unoptimized micro-Doppler frequency estimate. This invention achieves high-precision, high-resolution extraction of the rotational micro-Doppler frequency of ship scattering points by constructing a three-degree-of-freedom rotational micro-Doppler frequency model and employing a simultaneously compressed generalized S-transform and adaptive ridge filter method, effectively solving the problems of multi-component time-frequency crossover and noise interference.

The application provides a generalized S transform time-frequency analysis method based on dictionary learning and matching pursuit decomposition, comprising the following steps: step 1, inputting an original seismic signal set, and constructing a wavelet library based on K-SVD dictionary learning; step 2, combining the fast complex domain matching pursuit algorithm with the wavelet library constructed based on K-SVD dictionary learning to improve the speed and accuracy of seismic signal decomposition; step 3, obtaining the time-frequency spectrum of independent wavelets based on the generalized S transform, and superimposing the time-frequency spectrum of all independent wavelets to perform time-frequency joint analysis of the high-resolution time-frequency spectrum; step 4, studying the change of reservoir characteristics by using single-frequency attributes based on the high-resolution time-frequency spectrum; and step 5, eliminating the high-resolution time-frequency spectrum representing the change of reservoir characteristics after strong reflection. The generalized S transform time-frequency analysis method based on dictionary learning and matching pursuit decomposition forms a high-resolution time-frequency analysis technology according to the development of actual application requirements and the characteristics of non-stationary signals.

The invention discloses a rotating machinery fault diagnosis method based on second-order generalized S synchronous extraction transformation, and the method comprises the steps: collecting a vibration signal of rotating machinery equipment; parameters of generalized S transformation are selected by using a particle swarm algorithm, so that the time-frequency representation obtains the optimal time-frequency resolution; performing second-order synchronous extraction transformation based on generalized S transformation on the vibration signal x by using the parameters to obtain time-frequency representation; instantaneous rotating speed information is extracted from the time-frequency representation, and angular domain resampling is carried out on the vibration signal by using the rotating speed; and carrying out order analysis on the resampled signal to obtain a fault feature order graph of the rotating machine so as to judge the fault type of the rotating machine. Accurate extraction of the rotating speed can be realized under the conditions of strong variable rotating speed, low signal-to-noise ratio and lack of rotating speed information, and fault diagnosis of the rotating machinery is realized by utilizing order analysis.

The present application relates to a kind of cable defect positioning method, device, equipment and medium, it relates to defect detection technical field.The method includes: based on generalized S transformation estimation, the reflection coefficient spectrum of cable is transformed, the time-frequency spectrum of cable is obtained;Spectrum peak screening and pseudo-peak elimination are carried out to time-frequency spectrum, and the target signal of cable is obtained;Frequency domain summation is carried out to target signal, and the positioning curve of cable is obtained;Based on positioning curve and signal propagation velocity, the defect position of cable is determined.The precision positioning of cable defect can be realized by using the present method.

The invention discloses a variable-speed planetary gearbox fault diagnosis method, and relates to the technical field of mechanical fault diagnosis, and the method comprises the steps: processing a vibration signal of a variable-speed planetary gearbox through optimal generalized S transformation to obtain an optimal time-frequency diagram, and representing time-frequency characteristics; extracting enhanced features by combining a time-frequency enhanced attention mechanism with a plurality of receptive fields, aggregating the enhanced features through a space attention mechanism to form joint weighted features, and inputting the joint weighted features into a deep residual network to obtain final feature representation; and finally outputting a fault prediction value by constructing a domain adaptive network and adopting cross entropy loss and multi-kernel maximum mean difference domain adaptive loss combined training. The space attention weight is adjusted through a time-frequency attention mechanism, fault key feature representation is strengthened, and redundant information interference is reduced; deep features are extracted in combination with a deep residual network, and multi-core maximum mean difference is adopted to align source domain and target domain feature distribution, so that distribution difference is reduced, and domain adaptive diagnosis of the variable-speed planetary gearbox is realized.

This invention discloses a sparse generalized S-transform time-frequency analysis method with adaptive parameter optimization. This method employs a two-stage optimization strategy to optimize the core window parameters of the generalized S-transform and the sparse inversion solution parameters, respectively, and establishes a fitness function for evaluation to obtain the parameter combination that achieves the optimal time-frequency resolution. Finally, the optimized parameters are used to complete high-resolution time-frequency analysis of the signal. The advantages of this method lie in its adaptive parameter finding capability, reducing the subjectivity of manual parameter selection, and its significant advantages in time-frequency resolution and adaptability, providing time-frequency analysis technical support for subsequent seismic data processing and interpretation.

The invention discloses a seismic oscillation velocity pulse recognition method based on generalized S transform and a convolutional neural network, and the method comprises the steps: firstly carrying out the high-resolution time-frequency analysis of a one-dimensional seismic oscillation velocity time history through the generalized S transform, and converting the time history into a two-dimensional time-frequency spectrogram; inputting the time-frequency spectrogram into a pre-trained convolutional neural network, automatically extracting discrimination features of pulse type and non-pulse type seismic oscillation through deep learning, and outputting pulse existence and probability thereof; carrying out component decomposition on an original speed time history by adopting an ICEEMDAN method for the seismic oscillation record which is identified as a pulse type, separating out main pulse components, and reconstructing pulses; and finally, extracting characteristic parameters such as amplitude, period, quantity and the like from the reconstructed pulse signal to realize accurate identification and quantitative analysis of single pulse and multiple pulses. According to the invention, the precision and consistency of pulse recognition can be effectively improved, the degree of manual participation is reduced, and high robustness is maintained under the conditions of high noise and complex waveforms.

This invention discloses a method for predicting the life of rolling bearings based on optimal time-frequency spectrum and CNN-ALSTM network, comprising: Step 1, obtaining the vibration dataset X = {x1, x2, ..., x...} for the entire life cycle of the bearing. n Using the time-frequency energy evaluation index as the fitness function, the snake swarm optimization algorithm is used to optimize the adjustment factor p of the generalized S-transform to obtain the optimal time spectrum of each group of vibration data in the vibration dataset; Step 2, based on the optimal time spectrum obtained in Step 1 for each group of vibration data, the optimal time spectrum dataset S = {S1, S2, ..., S...} for the entire bearing life cycle is generated. n Step 3: Add a dense convolutional layer after the convolutional layers of the CNN network structure, and incorporate the self-attention mechanism into the first layer of the LSTM network to establish a CNN-ALSTM network; input the optimal time-spectrum dataset S into the CNN-ALSTM network for training to obtain the bearing's remaining service life prediction model. This invention can improve the prediction accuracy of rotating machinery bearing components and achieve the goal of accurately predicting the remaining service life of rotating machinery equipment.