Equipment signal separation method and device, computer program product and electronic equipment

Abstract

The invention discloses an equipment signal separation method and device, a computer program product and electronic equipment. The method relates to the field of signal monitoring or other related fields, and comprises the following steps: receiving M time domain pulse signals of a power machine room in a preset time period, and calculating the time domain feature of each time domain pulse signal to obtain M groups of time domain features, each time domain pulse signal comprising N sub-pulse signals; the M groups of time domain features are input into a sparse representation model, a sparse coefficient matrix is output, the sparse coefficient matrix and the M groups of time domain features are input into a clustering model, Y target clusters are output, and the sparse representation model is used for carrying out signal separation processing on the M time domain pulse signals; and determining Y target time domain pulse signals according to the Y target clusters. According to the invention, the problems of low processing precision and low processing efficiency during the separation operation of the pulse signal in the prior art are solved.

Description

Technical Field

[0001] The present application relates to the field of signal monitoring or other related fields, and in particular, to a method, apparatus, computer program product and electronic device for separating device signals. Background Art

[0002] With the continuous expansion of the scale and complexity of the power system, the need to ensure the safe and stable operation of the power system is becoming increasingly urgent. As one of the key equipment in the power system, switchgear equipment plays an important role in the power system, and its reliability and stability directly affect the safe operation of the entire power grid. However, since the switchgear equipment may be affected by process limitations, improper handling, and external factors such as lightning overvoltage, contamination, and weak insulation during the design, manufacturing, installation, and operation and maintenance, the switchgear equipment may have defects or hidden dangers to varying degrees in actual operation. These defects and hidden dangers may cause equipment failures or abnormal conditions, such as external insulation flashover breakdown, internal insulation flashover breakdown, phase-to-phase insulation flashover breakdown, and bushing flashover breakdown. These faults will not only cause equipment damage, but may also cause the safety and stability of the power grid to be destroyed due to the tripping of the switchgear equipment, thereby causing power grid safety and stability problems, causing large-scale power outages, and causing serious losses to users' electricity and economy.

[0003] In order to timely discover and diagnose the faults of switchgear equipment, multi-source signal recognition technology plays an important role in the monitoring and fault diagnosis of switchgear equipment. Traditional switchgear monitoring methods mainly rely on the analysis of time domain features, but in high-frequency current monitoring, they are often affected by environmental interference and multi-source discharge signals, making it difficult to distinguish between environmental interference and partial discharge signals. In order to effectively separate multi-source signals and accurately identify and classify each signal, traditional signal processing methods are often limited to feature extraction and simple classification methods based on time domain. For example, the processing method can be waveform analysis that judges the nature of the signal by observing the shape, amplitude and duration of the signal waveform; it can be a frequency domain analysis method that converts the signal to the frequency domain and analyzes the frequency component of the signal to identify the source of the signal; it can also be a time domain analysis that simultaneously obtains the time and frequency information of the signal to improve the accuracy of signal recognition. However, these methods often cannot distinguish different types of partial discharge signals well when the signal is complex and the noise interference is large, and they rely on manual experience and subjective judgment in the process of signal recognition and classification, resulting in poor consistency and repeatability of the results, slow processing speed, and inability to meet real-time requirements.

[0004] With regard to the problems of low processing accuracy and low processing efficiency when performing separation operations on pulse signals in related technologies, no effective solution has been proposed so far. Summary of the invention

[0005] The main purpose of the present application is to provide a method, apparatus, computer program product and electronic device for separating device signals, so as to solve the problems of low processing accuracy and low processing efficiency when performing separation operations on pulse signals in related technologies.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for separating device signals is provided. The method includes: receiving M time domain pulse signals in a power room within a preset time period, and calculating the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a time domain pulse signal after mixing multiple switch cabinet devices emitted by the power room, and each time domain pulse signal contains N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; inputting the M groups of time domain features into a sparse representation model, outputting a sparse coefficient matrix, and inputting the sparse coefficient matrix and the M groups of time domain features into a clustering model, and outputting Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer; determining Y target time domain pulse signals according to the Y target clusters, wherein the Y target time domain pulse signals refer to the signals obtained after separating the M time domain pulse signals.

[0007] Furthermore, receiving M time domain pulse signals from the power room within a preset time period includes: acquiring the high-frequency current time domain signal of the power room in real time, and sending a pulse capture signal to the acquisition device when the amplitude of the high-frequency current time domain signal is greater than the trigger threshold, wherein the acquisition device starts the time domain pulse capture operation when receiving the pulse capture signal; and acquiring N pulse durations when receiving M high-frequency current time domain signals fed back by the acquisition device, and capturing and processing the M high-frequency current time domain signals according to the N pulse durations to obtain M time domain pulse signals.

[0008] Further, each group of time domain features includes at least one of the following: signal mean, signal standard deviation, signal peak, pulse width, waveform transition rate and amplitude change rate, and calculating the time domain features of each time domain pulse signal includes at least one of the following: for a time domain pulse signal, obtaining the digit point of the time domain pulse signal in N pulse durations, determining the start index time and the end index time according to the digit point, calculating the difference between the end index time and the start index time, and obtaining the pulse width of the time domain pulse signal; for a time domain pulse signal, calculating the absolute value of the difference between two adjacent sub-pulse signals in the time domain pulse signal to obtain N-1 first difference absolute values, and calculating the average value of the N-1 first difference absolute values ​​to obtain the time domain pulse signal. The waveform transition rate of the pulse signal is calculated; for a time domain pulse signal, the absolute value of the difference between the ratio of two adjacent sub-pulse signals in the time domain pulse signal and a preset parameter is calculated to obtain N-1 second absolute values ​​of the difference, and the average value of the N-1 second absolute values ​​of the difference is calculated to obtain the amplitude change rate of the time domain pulse signal; for a time domain pulse signal, the mean and standard deviation of N sub-pulse signals associated with the time domain pulse signal are calculated to obtain the signal mean and signal standard deviation of the time domain pulse signal; for a time domain pulse signal, the signal amplitudes of N sub-pulse signals associated with the time domain pulse signal are obtained, the signal amplitude with the largest value is screened out from the N signal amplitudes, and the signal amplitude with the largest value is determined as the signal peak value of the time domain pulse signal.

[0009] Furthermore, M groups of time domain features are input into the sparse representation model, and outputting a sparse coefficient matrix includes: obtaining a dictionary matrix of M time domain pulse signals; constructing a signal vector using each time domain pulse signal and the time domain features associated with each pulse signal to obtain M signal vectors; calculating a preset reconstruction error using the dictionary matrix and the M signal vectors to obtain M preset reconstruction errors, wherein the preset reconstruction error refers to the reconstruction error between each time domain pulse signal and the sparse representation of each time domain pulse signal; determining M candidate sparse coefficient matrices based on the M preset reconstruction errors, and screening the M candidate sparse coefficient matrices to obtain a sparse coefficient matrix.

[0010] Furthermore, the M candidate sparse coefficient matrices are screened to obtain the sparse coefficient matrix, including: calculating the sparsity penalty item of the candidate sparse coefficient matrix corresponding to each preset reconstruction error to obtain M sparsity penalty items; obtaining the trade-off coefficient, calculating the product of each sparsity penalty item and the trade-off coefficient to obtain M trade-off products; calculating the sum of each trade-off product and each preset reconstruction error to obtain M preset sums, screening out the preset sum with the smallest value from the M preset sums, and determining the candidate sparse coefficient matrix associated with the preset sum with the smallest value as the sparse coefficient matrix.

[0011] Furthermore, the sparse coefficient matrix and M groups of time-domain features are input into the clustering model, and Y target clusters are output, including: obtaining a bandwidth parameter, wherein the bandwidth parameter is used to control the width of the kernel function in the clustering model; calculating the Gaussian kernel function of any two time-domain pulse signals in each time-domain pulse signal according to the bandwidth parameter, and obtaining M Gaussian kernel functions, wherein the Gaussian kernel function is used to characterize the similarity between every two time-domain pulse signals; calculating the mean of the M Gaussian kernel functions to obtain a density estimate, determining an initial value of the cluster center according to the density estimate, obtaining Y initial cluster centers, and determining Y target clusters based on the Y initial cluster centers.

[0012] Furthermore, an initial value of the cluster center is determined according to the density estimation value to obtain Y initial cluster centers, and Y target clusters are determined based on the Y initial cluster centers, including: for a time domain pulse signal, calculating the distance between the time domain pulse signal and the Y initial cluster centers to obtain a group of cluster labels, clustering the time domain pulse information according to the group of cluster labels to obtain Y cluster centers, and determining Y first updated cluster centers according to the Y cluster centers; for each first updated cluster center, calculating the mean of the time domain pulse signal contained in the first updated cluster center to obtain the center mean, updating the first updated cluster center according to the center mean to obtain the second updated cluster center; iteratively processing the Y second updated cluster centers, and when the Y second updated cluster centers do not change, determining the Y target clusters according to the Y second updated cluster centers.

[0013] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a device for separating equipment signals is provided. The device includes: a receiving unit, which is used to receive M time domain pulse signals in a power room within a preset time period, and calculate the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a time domain pulse signal of multiple switch cabinet devices emitted by the power room after mixing, and each time domain pulse signal contains N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; an input unit, which is used to input the M groups of time domain features into a sparse representation model, output a sparse coefficient matrix, and input the sparse coefficient matrix and the M groups of time domain features into a clustering model, and output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, and the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer; a determination unit, which is used to determine Y target time domain pulse signals according to the Y target clusters, wherein the Y target time domain pulse signals refer to the signals obtained after separating the M time domain pulse signals.

[0014] According to another aspect of an embodiment of the present invention, an electronic device is provided, comprising one or more processors and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein a method for separating device signals is executed when the computer-readable instructions are executed.

[0015] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, a method for separating device signals is performed.

[0016] Through the present application, the following steps are adopted: receiving M time domain pulse signals in a preset time period from a power room, and calculating the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes multiple switch cabinet devices, and each time domain pulse signal refers to a time domain pulse signal of multiple switch cabinet devices emitted by the power room, and each time domain pulse signal contains N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; inputting the M groups of time domain features into a sparse representation model, outputting a sparse coefficient matrix, and inputting the sparse coefficient matrix and the M groups of time domain features into a clustering model, and outputting Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, and the sparse coefficient matrix refers to the M groups of time domain features. The feature corresponds to the coefficient vector of sparse representation, and Y is a positive integer; Y target time-domain pulse signals are determined according to Y target clusters, wherein the Y target time-domain pulse signals refer to the signals obtained after separating the M time-domain pulse signals, which solves the problems of low processing accuracy and low processing efficiency when performing separation operations on pulse signals in related technologies. By receiving multiple time-domain pulse signals in a preset time period in an electric power room and calculating their time-domain features, the time-domain features are input into a sparse representation model to output a sparse coefficient matrix, and then the sparse coefficient matrix and the time-domain features are input into a clustering model to output multiple target clusters. Finally, multiple target time-domain pulse signals are determined based on the target clusters, thereby achieving the effect of quickly and accurately separating time-domain pulse signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 is a flow chart of a method for separating device signals provided in an embodiment of the present application;

[0019] Figure 2 is a schematic diagram of an optional device signal separation method provided according to an embodiment of the present application;

[0020] Figure 3is a schematic diagram of a device signal separation apparatus provided according to an embodiment of the present application;

[0021] Figure 4 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set between this system and the relevant user or organization. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0026] It should be noted that the collected information used in this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse to use it.

[0027] The present invention is described below in conjunction with preferred implementation steps. Figure 1 is a flow chart of a method for separating device signals according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0028] Step S101, receiving M time domain pulse signals from the power room within a preset time period, and calculating the time domain characteristics of each time domain pulse signal to obtain M groups of time domain characteristics, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a time domain pulse signal mixed from multiple switch cabinet devices emitted by the power room, each time domain pulse signal contains N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers.

[0029] Specifically, the power room refers to a room equipped with multiple switch cabinets. Due to the complexity of the signal and the large noise interference, it is impossible to distinguish the different types of partial discharge signals emitted by the power room. Therefore, in order to effectively sort out the multi-source signals output by the power room and accurately identify each signal, it is possible to receive the mixed time domain pulse signals emitted by multiple switch cabinets in the power room within a preset time period, and then calculate the time domain characteristics of each time domain pulse signal to obtain multiple groups of time domain characteristics. Finally, the time domain characteristics are used to perform signal separation operations, where the time domain characteristics may include signal mean, signal standard deviation, signal peak, pulse width, waveform jump rate, amplitude change rate, etc.

[0030] Step S102, input M groups of time domain features into the sparse representation model, output a sparse coefficient matrix, and input the sparse coefficient matrix and the M groups of time domain features into the clustering model, and output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer.

[0031] Specifically, after calculating the time domain features of each time domain pulse signal, the time domain features can be processed using a sparse representation method to make full use of the structural information of the signal, thereby improving the accuracy and distinctiveness of the sparse representation. At this time, the time domain features can be input into the sparse representation model, and the model outputs a sparse coefficient matrix based on techniques such as dictionary learning.

[0032] Furthermore, the sparse coefficient matrix and the time domain features are taken as inputs, and a clustering model is used to determine the potential patterns between the time domain pulse signals, and the model outputs multiple clusters, wherein the clustering model can adopt a model composed of a K-means clustering algorithm.

[0033] Step S103, determining Y target time-domain pulse signals according to the Y target clusters, wherein the Y target time-domain pulse signals refer to signals obtained by separating the M time-domain pulse signals.

[0034] Specifically, after the sparse representation model and the clustering model process the time domain features to obtain multiple clusters, the target time domain pulse signals can be determined based on these clusters, thereby achieving the effect of multi-source signal separation. The separated pulse signals can further be used to determine the switch cabinet equipment with faults in the power room. For example, after multiple target time domain pulse signals are determined based on the target clusters, the switch cabinet equipment with faults can be determined based on the determined target time domain pulse signals, thereby being able to quickly and accurately determine the faulty equipment, thereby reducing fault handling time and maintenance costs, and providing strong support for the reliability and stability of the power system.

[0035] The device signal separation method provided in the embodiment of the present application receives M time domain pulse signals in a preset time period in a power room, and calculates the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a time domain pulse signal of multiple switch cabinet devices emitted by the power room, each time domain pulse signal includes N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; the M groups of time domain features are input into a sparse representation model, a sparse coefficient matrix is ​​output, and the sparse coefficient matrix and the M groups of time domain features are input into a clustering model to output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, and the sparse coefficient matrix refers to M The time domain features of the group correspond to the coefficient vector of the sparse representation, and Y is a positive integer; Y target time domain pulse signals are determined according to the Y target clusters, wherein the Y target time domain pulse signals refer to the signals obtained after separating the M time domain pulse signals, which solves the problems of low processing accuracy and low processing efficiency when performing separation operations on pulse signals in the related technology, by receiving multiple time domain pulse signals in the power room within a preset time period, and calculating their time domain features, using the time domain features to input the sparse representation model, and then outputting a sparse coefficient matrix, and then inputting the sparse coefficient matrix and the time domain features into the clustering model, outputting multiple target clusters, and finally determining multiple target time domain pulse signals based on the target clusters, thereby achieving the effect of quickly and accurately separating the time domain pulse signals.

[0036] The time domain pulse signal is collected by a triggered collection method. Optionally, in the device signal separation method provided in the embodiment of the present application, receiving M time domain pulse signals in the power room within a preset time period includes: acquiring the high-frequency current time domain signal of the power room in real time, and sending a pulse capture signal to the collection device when the amplitude of the high-frequency current time domain signal is greater than the trigger threshold, wherein the collection device starts the time domain pulse capture operation when receiving the pulse capture signal; when receiving M high-frequency current time domain signals fed back by the collection device, acquiring N pulse durations, capturing and processing the M high-frequency current time domain signals according to the N pulse durations, and obtaining M time domain pulse signals.

[0037] Specifically, the acquisition device can be an acquisition card connected to a computer. The time domain pulse signal can be obtained by triggering acquisition. Before acquisition, it is necessary to first detect the high-frequency current time domain signal in the power room in real time. At the same time, a trigger threshold for triggering the acquisition card operation is set inside the monitoring device. When the amplitude of the current time domain signal reaches or exceeds this trigger threshold, the pulse capture signal is triggered and sent to the acquisition device.

[0038] After the acquisition device receives the pulse capture signal, it starts the time domain pulse capture operation, starts to collect high-frequency current time domain signals, and returns these signals to the computer. At this time, after receiving multiple high-frequency current time domain signals, the time domain signals are captured and processed according to the set pulse duration. For example, when the pulse duration is set to N, the time domain signal that satisfies t-t0<=N from the trigger time can be captured from the trigger time t0 according to the pulse duration as the time domain pulse signal. At this time, multiple time domain pulse signals can be obtained, and each time domain pulse signal can be x=[x1, x2, ..., x N ], where x N Indicates a sub-pulse signal. This embodiment acquires time-domain pulse signals by using an acquisition device and a triggered acquisition method, and uses these signals as original data sources to provide data support for equipment status monitoring and maintenance.

[0039] There are multiple types of time domain features. Optionally, in the device signal separation method provided in the embodiment of the present application, each group of time domain features includes at least one of the following: signal mean, signal standard deviation, signal peak, pulse width, waveform jump rate and amplitude change rate. Calculating the time domain features of each time domain pulse signal includes at least one of the following: for a time domain pulse signal, obtaining the digital position of the time domain pulse signal in N pulse durations, determining the start index time and the end index time according to the digital position, calculating the difference between the end index time and the start index time, and obtaining the pulse width of the time domain pulse signal; for a time domain pulse signal, calculating the absolute value of the difference between two adjacent sub-pulse signals in the time domain pulse signal, obtaining N-1 first difference absolute values, and calculating N-1 first difference absolute values. The waveform transition rate of the time domain pulse signal is obtained by calculating the average of the absolute values ​​of the values. For a time domain pulse signal, the absolute value of the difference between the ratio of two adjacent sub-pulse signals in the time domain pulse signal and a preset parameter is calculated to obtain N-1 second absolute values ​​of the difference, and the average of the N-1 second absolute values ​​of the difference is calculated to obtain the amplitude change rate of the time domain pulse signal. For a time domain pulse signal, the mean and standard deviation of the N sub-pulse signals associated with the time domain pulse signal are calculated to obtain the signal mean and signal standard deviation of the time domain pulse signal. For a time domain pulse signal, the signal amplitudes of the N sub-pulse signals associated with the time domain pulse signal are obtained, the signal amplitude with the largest value is screened out from the N signal amplitudes, and the signal amplitude with the largest value is determined as the signal peak value of the time domain pulse signal.

[0040] Specifically, after obtaining multiple time domain pulse signals, the time domain features can be determined according to the sub-pulse signals in the time domain pulse signals. First, the mean of a group of sub-pulse signals can be calculated, and then the average value reflecting the overall level of the time domain pulse signal can be obtained. The signal mean can describe the center position of the signal and can be calculated by the following formula:

[0041]

[0042] Where N is the number of sub-pulse signals, x i Represents the i-th sub-pulse signal.

[0043] In addition, the sub-pulse signal can also be used to calculate the signal standard deviation used to describe the volatility and stability of the signal, thereby determining the degree to which the signal value deviates from the mean. The signal standard deviation can be calculated using the following formula: Among them, A mean represents the mean, N is the number of sub-pulse signals, x i Represents the i-th sub-pulse signal.

[0044] When calculating the signal peak value used to describe the maximum amplitude of the signal, it can be obtained by capturing the prominent peak of the sub-pulse signal, that is, determined by the following formula: Apeak=max(x).

[0045] In addition, when calculating the pulse width in the time domain characteristics: the starting index time and the ending index time can be found by obtaining the digital points of the time domain pulse signal in the pulse duration, and then the pulse width of the time domain pulse signal can be obtained by using the difference between the ending index time and the starting index time. It should be noted that the pulse width can describe the duration of the signal pulse and reflect the time domain characteristics of partial discharge, which can be calculated by the following formula:

[0046] A width =t end -t start ;

[0047] Among them, t end is the end index of x, t start is the starting index of x.

[0048] When calculating the waveform transition rate in the time domain characteristics, the absolute value of the difference between every two sub-pulse signals in the time domain pulse signal can be calculated, and then the average of these absolute values ​​can be calculated to obtain the waveform transition rate. The waveform transition rate can capture the discontinuity and rapid changes of the signal. It measures the rate of change of the waveform of the signal in a short period of time and can be calculated by the following formula: Where N is the number of sub-pulse signals, x i Represents the i-th sub-pulse signal.

[0049] Finally, when calculating the amplitude change rate, it can be determined by using the absolute value of the difference between the ratio of two adjacent sub-pulse signals in the time domain pulse signal and the preset parameter, that is, it can be determined by calculating the ratio of the difference between two adjacent sub-pulse signals and the previous sub-pulse signal. This time domain feature can measure the rate of change of the signal amplitude over time, reflecting the rapid change of the signal amplitude, which can be calculated by the following formula: Where N is the number of sub-pulse signals, x i Indicates the ith sub-pulse signal. This embodiment calculates the time domain features by using the sub-pulse signals of each time domain pulse signal, thereby laying a foundation for subsequent signal separation.

[0050] In order to determine the sparse representation of the time domain pulse signal, optionally, in the device signal separation method provided in the embodiment of the present application, M groups of time domain features are input into the sparse representation model, and the output sparse coefficient matrix includes: obtaining a dictionary matrix of M time domain pulse signals; constructing a signal vector using each time domain pulse signal and the time domain features associated with each pulse signal to obtain M signal vectors; calculating a preset reconstruction error using the dictionary matrix and the M signal vectors to obtain M preset reconstruction errors, wherein the preset reconstruction error refers to the reconstruction error between each time domain pulse signal and the sparse representation of each time domain pulse signal; determining M candidate sparse coefficient matrices based on the M preset reconstruction errors, and screening the M candidate sparse coefficient matrices to obtain a sparse coefficient matrix.

[0051] Specifically, the dictionary matrix is ​​a data structure that stores keys and values ​​in a matrix in a one-to-one correspondence. Each key corresponds to a row or a column of the matrix, and the value corresponds to another row or another column of the matrix. This structure can be used to quickly find the value corresponding to the key, usually using a hash table or other fast search algorithm. When using the sparse representation model to sparsely represent the time domain pulse signal, first obtain the dictionary matrix D = [d1, d2, ..., d M ], the dictionary matrix can provide the atomic set of the signal, which plays a key role in signal reconstruction, where d i Represents the i-th column in the dictionary, i=1, 2, ..., M.

[0052] Furthermore, the time domain features of each time domain pulse signal can be used to construct a signal vector. By combining these time domain features, a signal vector is generated for each pulse signal. For example, the signal vector can be X i =[x i , A mean , A std , A peak , A width , A jumpRate , A changeRate ], where x i represents the time domain pulse signal, A mean , A std , A peak , A width , A jumpRate , A changeRate They represent the time domain features corresponding to the time domain pulse signal, A mean represents the signal mean, A std Indicates the signal standard deviation, A peak Indicates the signal peak value, A width Indicates pulse width, A jumpRate Indicates the waveform transition rate and A changeRate Indicates the rate of change of amplitude.

[0053] Furthermore, the dictionary matrix and the above signal vector can be used to calculate the preset reconstruction error Among them, the reconstruction error is obtained by comparing the signal vector of each time domain pulse signal with its sparse representation (i.e., the approximation obtained by linear combination of column vectors in the dictionary matrix). Finally, a sparse coefficient matrix can be determined based on the calculated preset reconstruction error, wherein each sparse coefficient matrix corresponds to a specific signal vector, contains weights or coefficients corresponding to each column in the dictionary matrix, and these coefficients indicate the contribution of each atom in reconstructing the original signal. After obtaining the above-mentioned multiple candidate sparse coefficient matrices, these candidate sparse coefficient matrices are screened to determine a better sparse coefficient matrix, and then the coefficient matrix can be used to perform a sparse representation of the signal. This embodiment constructs a reconstruction error by using a dictionary matrix and a signal vector, and then uses the reconstruction error to determine the sparse coefficient matrix, and finally lays the foundation for realizing signal clustering.

[0054] Optionally, in the device signal separation method provided in an embodiment of the present application, M candidate sparse coefficient matrices are screened to obtain a sparse coefficient matrix, including: calculating the sparsity penalty item of the candidate sparse coefficient matrix corresponding to each preset reconstruction error to obtain M sparse penalty items; obtaining a trade-off coefficient, calculating the product of each sparsity penalty item and the trade-off coefficient to obtain M trade-off products; calculating the sum of each trade-off product and each preset reconstruction error to obtain M preset sums, screening out the preset sum with the smallest value from the M preset sums, and determining the candidate sparse coefficient matrix associated with the preset sum with the smallest value as the sparse coefficient matrix.

[0055] Specifically, after the candidate sparse coefficient matrix C is calculated, the sparsity penalty term can be calculated based on the candidate sparse coefficient matrix, that is, the sparsity penalty term Ω(C) is obtained by calculating the norm of the candidate sparse coefficient matrix. After obtaining the trade-off coefficient γ used to adjust the importance between the sparsity penalty term and the reconstruction error, the product γ·Ω(C) of the sparsity penalty term and the trade-off coefficient is calculated to obtain the trade-off product, and finally the sum of the trade-off product and the preset reconstruction error is calculated, and the minimum sum is found, and then the sparse coefficient matrix is ​​determined based on the sum, that is, the minimum preset sum can be calculated by the following formula: minD, Among them, the signal vector X i =[x i , A mean , A std , A peak , A width , A jumpRate , A changeRate ], D = [d1, d2, ..., d M ] is the dictionary matrix, di represents the i-th column in the dictionary, i = 1, 2, ..., M, C = [c1, c2, ..., c M ] is the candidate sparse coefficient matrix, c i Represents the signal x i The corresponding sparse coefficient vector, i = 1, 2, ..., M, represents the preset reconstruction error between the ith signal and its sparse representation, γ is a trade-off parameter, and Ω(C) is a sparsity penalty term. This embodiment uses a candidate sparse coefficient matrix to screen out a sparse coefficient matrix, providing data support for subsequent signal separation using a clustering model.

[0056] In order to achieve effective separation of time domain pulse signals, optionally, in the device signal separation method provided in the embodiment of the present application, the sparse coefficient matrix and M groups of time domain features are input into the clustering model, and Y target clusters are output, including: obtaining a bandwidth parameter, wherein the bandwidth parameter is used to control the width of the kernel function in the clustering model; calculating the Gaussian kernel function of any two time domain pulse signals in each time domain pulse signal according to the bandwidth parameter, and obtaining M Gaussian kernel functions, wherein the Gaussian kernel function is used to characterize the similarity between every two time domain pulse signals; calculating the mean of the M Gaussian kernel functions to obtain a density estimate, determining an initial value of the cluster center according to the density estimate, obtaining Y initial cluster centers, and determining Y target clusters based on the Y initial cluster centers.

[0057] After the sparse coefficient matrix is ​​output according to the sparse representation model, the sparse coefficient matrix and the time domain features can be input into the clustering model constructed by the clustering algorithm, and the clusters after clustering can be determined by the model. Specifically, the cluster center can be initialized first, and the bandwidth parameter h for controlling the width of the kernel function can be obtained at this time. Then, the Gaussian kernel function of any two time domain pulse signals in each time domain pulse signal is calculated based on the bandwidth parameter. The Gaussian kernel function K(x i , x j ):

[0058]

[0059] Where h is the bandwidth parameter, x i and x j Respectively represent any two time domain pulse signals in each time domain pulse signal. Further, the density estimation value is determined by using the mean of the Gaussian kernel function, and the density estimation value p(x) can be calculated by the following formula:

[0060]

[0061] Where M is the total number of pulse time domain signals, K(x i , x j) is a Gaussian kernel function. Finally, the pulse time domain signal can be pulse-distributed according to the density estimation value, thereby obtaining multiple initial clustering centers, and finally determining the corresponding target cluster based on the initial clustering centers. This embodiment can effectively divide the time domain pulse signal into multiple meaningful clusters by calculating the target cluster using a sparse coefficient matrix, laying a foundation for further analysis and research.

[0062] Optionally, in the device signal separation method provided in the embodiment of the present application, the initial value of the cluster center is determined according to the density estimation value to obtain Y initial cluster centers, and Y target clusters are determined based on the Y initial cluster centers, including: for a time domain pulse signal, calculating the distance between the time domain pulse signal and the Y initial cluster centers to obtain a set of cluster labels, clustering the time domain pulse information according to the set of cluster labels to obtain Y cluster centers, and determining Y first updated cluster centers based on the Y cluster centers; for each first updated cluster center, calculating the mean of the time domain pulse signal contained in the first updated cluster center to obtain the center mean, updating the first updated cluster center according to the center mean to obtain the second updated cluster center; iteratively processing the Y second updated cluster centers, and when the Y second updated cluster centers do not change, determining the Y target clusters based on the Y second updated cluster centers.

[0063] Specifically, after calculating the density estimate value according to the Gaussian kernel function, the density estimate value is used to determine the initial cluster center, and then the distance between each time domain pulse signal and the initial cluster center is calculated, and each time domain pulse signal is assigned to the cluster with the closest distance, that is, the cluster label label(x i ):

[0064] label(x i ) = argmin||x i -μ k || 2 ;

[0065] Among them, μ k represents the kth initial cluster center, x i Represents a time domain pulse signal. After determining the cluster center according to the cluster label and further determining the first updated cluster center, the mean of the time domain pulse signal contained in the first updated cluster center can be calculated, and then the cluster center can be updated. The mean can be calculated by the following formula:

[0066]

[0067] Among them, C kRepresents the kth first updated cluster center. After repeating the above steps to implement the second updated cluster center for iterative processing, it is determined whether the second updated cluster center will change, or whether the iteration has reached a preset number of times. If the second updated cluster center does not change, multiple target clusters can be determined based on the second updated cluster center, thereby achieving the goal of multi-source signal separation.

[0068] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0069] The present application also provides a method for separating device signals. Figure 2 is a schematic diagram of an optional device signal separation method provided according to an embodiment of the present application, such as Figure 2 As shown, the method includes:

[0070] In order to realize the signal separation of the switch cabinet equipment, the high-frequency current time domain signal of the power room can be first obtained in real time, and the amplitude of the high-frequency current time domain signal can be detected. When the amplitude of the high-frequency current time domain signal is greater than the trigger threshold, the time domain pulse signal is collected by trigger acquisition, that is, the high-frequency current time domain signal is captured and processed according to the pulse duration to obtain multiple time domain pulse signals.

[0071] Furthermore, the corresponding time domain features can be calculated by the sub-pulse signals in the time domain pulse signal, so as to obtain time domain features such as signal mean, signal standard deviation, signal peak, pulse width, waveform jump rate and amplitude change rate, and then the calculated time domain features are calculated based on the improved sparse representation method to obtain the sparse coefficient of each time domain pulse signal. After the sparse coefficient is calculated, the sparse coefficient and the time domain feature are used together to construct a clustering model, and the model outputs multiple clusters, and finally the separation of multi-source time domain pulse signals is realized based on the clusters.

[0072] This embodiment receives multiple time domain pulse signals from the power room within a preset time period, calculates their time domain features, uses the time domain features to input a sparse representation model, and then outputs a sparse coefficient matrix. The sparse coefficient matrix and the time domain features are then input into a clustering model, and multiple target clusters are output. Finally, multiple target time domain pulse signals are determined based on the target clusters, thereby achieving the effect of quickly and accurately separating time domain pulse signals.

[0073] The embodiment of the present application also provides a device signal separation device. It should be noted that the device signal separation device of the embodiment of the present application can be used to execute the device signal separation method provided in the embodiment of the present application. The device signal separation device provided in the embodiment of the present application is introduced below.

[0074] Figure 3 is a schematic diagram of a device signal separation apparatus provided according to an embodiment of the present application, such as Figure 3 As shown, the device includes: a receiving unit 30, an input unit 31, and a determining unit 32.

[0075] The receiving unit 30 is used to receive M time domain pulse signals from the power room within a preset time period, and calculate the time domain characteristics of each time domain pulse signal to obtain M groups of time domain characteristics, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a mixed time domain pulse signal of multiple switch cabinet devices emitted by the power room, and each time domain pulse signal includes N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers;

[0076] An input unit 31 is used to input M groups of time domain features into a sparse representation model, output a sparse coefficient matrix, and input the sparse coefficient matrix and the M groups of time domain features into a clustering model, and output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer;

[0077] The determination unit 32 is used to determine Y target time-domain pulse signals according to the Y target clusters, wherein the Y target time-domain pulse signals refer to signals obtained by separating the M time-domain pulse signals.

[0078] The device signal separation device provided in the embodiment of the present application receives M time domain pulse signals in the power room within a preset time period through the receiving unit 30, and calculates the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes multiple switch cabinet devices, each time domain pulse signal refers to a time domain pulse signal of multiple switch cabinet devices emitted by the power room, each time domain pulse signal contains N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; the input unit 31 inputs the M groups of time domain features into the sparse representation model, outputs a sparse coefficient matrix, and inputs the sparse coefficient matrix and the M groups of time domain features into the clustering model, and outputs Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, and the sparse coefficient matrix It refers to the coefficient vector of sparse representation corresponding to M groups of time domain features, and Y is a positive integer; the determination unit 32 determines Y target time domain pulse signals according to Y target clusters, wherein the Y target time domain pulse signals refer to the signals obtained after separating the M time domain pulse signals, which solves the problems of low processing accuracy and low processing efficiency when performing separation operations on pulse signals in related technologies, by receiving multiple time domain pulse signals from the power room within a preset time period, and calculating their time domain features, using the time domain features to input the sparse representation model, and then outputting a sparse coefficient matrix, and then inputting the sparse coefficient matrix and the time domain features into the clustering model, outputting multiple target clusters, and finally determining multiple target time domain pulse signals based on the target clusters, thereby achieving the effect of quickly and accurately separating the time domain pulse signals.

[0079] Optionally, in the device for separating equipment signals provided in the embodiment of the present application, the receiving unit 30 includes: a first acquisition module, used to acquire the high-frequency current time domain signal of the power room in real time, and send a pulse capture signal to the acquisition device when the amplitude of the high-frequency current time domain signal is greater than the trigger threshold, wherein the acquisition device starts the time domain pulse capture operation when receiving the pulse capture signal; a second acquisition module, used to acquire N pulse durations when receiving M high-frequency current time domain signals fed back by the acquisition device, and capture and process the M high-frequency current time domain signals according to the N pulse durations to obtain M time domain pulse signals.

[0080] Optionally, in the device signal separation apparatus provided in the embodiment of the present application, the receiving unit 30 includes: a third acquisition module, which is used to obtain the digital position of the time domain pulse signal in N pulse durations for a time domain pulse signal, determine the start index time and the end index time according to the digital position, calculate the difference between the end index time and the start index time, and obtain the pulse width of the time domain pulse signal; a first calculation module, which is used to calculate the absolute value of the difference between two adjacent sub-pulse signals in the time domain pulse signal for a time domain pulse signal, obtain N-1 first difference absolute values, and calculate the average value of the N-1 first difference absolute values ​​to obtain the waveform transition rate of the time domain pulse signal; a second calculation module, which is used to calculate the absolute value of the difference between two adjacent sub-pulse signals in the time domain pulse signal for a time domain pulse signal, obtain N-1 first difference absolute values, and calculate the average value of the N-1 first difference absolute values ​​to obtain the waveform transition rate of the time domain pulse signal. signal, calculate the absolute value of the difference between the ratio of two adjacent sub-pulse signals in the time domain pulse signal and a preset parameter, obtain N-1 second absolute values ​​of the difference, calculate the average value of the N-1 second absolute values ​​of the difference, and obtain the amplitude change rate of the time domain pulse signal; a third calculation module is used to calculate the mean and standard deviation of N sub-pulse signals associated with the time domain pulse signal for a time domain pulse signal, and obtain the signal mean and signal standard deviation of the time domain pulse signal; a fourth acquisition module is used to obtain the signal amplitudes of N sub-pulse signals associated with the time domain pulse signal for a time domain pulse signal, screen out the signal amplitude with the largest value from the N signal amplitudes, and determine the signal amplitude with the largest value as the signal peak value of the time domain pulse signal.

[0081] Optionally, in the device signal separation apparatus provided in the embodiment of the present application, the input unit 31 includes: a fifth acquisition module, used to acquire a dictionary matrix of M time domain pulse signals; a construction module, used to construct a signal vector using each time domain pulse signal and the time domain features associated with each pulse signal, to obtain M signal vectors; a fourth calculation module, used to calculate a preset reconstruction error using the dictionary matrix and the M signal vectors, to obtain M preset reconstruction errors, wherein the preset reconstruction error refers to the reconstruction error between each time domain pulse signal and the sparse representation of each time domain pulse signal; a determination module, used to determine M candidate sparse coefficient matrices based on the M preset reconstruction errors, and screen the M candidate sparse coefficient matrices to obtain a sparse coefficient matrix.

[0082] Optionally, in the device signal separation apparatus provided in the embodiment of the present application, the determination unit 31 includes: a fifth calculation module, used to calculate the sparsity penalty item of the candidate sparse coefficient matrix corresponding to each preset reconstruction error, and obtain M sparsity penalty items; a sixth acquisition module, used to obtain the trade-off coefficient, calculate the product of each sparsity penalty item and the trade-off coefficient, and obtain M trade-off products; a sixth calculation module, used to calculate the sum of each trade-off product and each preset reconstruction error, and obtain M preset sums, and filter out the preset sum with the smallest value from the M preset sums, and determine the candidate sparse coefficient matrix associated with the preset sum with the smallest value as the sparse coefficient matrix.

[0083] Optionally, in the device signal separation apparatus provided in the embodiment of the present application, the input unit 31 includes: a seventh acquisition module, used to acquire bandwidth parameters, wherein the bandwidth parameters are used to control the width of the kernel function in the clustering model; a seventh calculation module, used to calculate the Gaussian kernel functions of any two time domain pulse signals in each time domain pulse signal according to the bandwidth parameters, and obtain M Gaussian kernel functions, wherein the Gaussian kernel function is used to characterize the similarity between every two time domain pulse signals; an eighth calculation module, used to calculate the mean of the M Gaussian kernel functions, obtain a density estimation value, determine the initial value of the cluster center according to the density estimation value, obtain Y initial cluster centers, and determine Y target clusters based on the Y initial cluster centers.

[0084] Optionally, in the device signal separation apparatus provided in the embodiment of the present application, the input unit 31 includes: a ninth calculation module, which is used to calculate the distance between a time domain pulse signal and Y initial cluster centers for a time domain pulse signal to obtain a set of cluster labels, cluster the time domain pulse information according to the set of cluster labels to obtain Y cluster centers, and determine Y first updated cluster centers according to the Y cluster centers; a tenth calculation module, which is used to calculate the mean of the time domain pulse signal contained in the first updated cluster center for each first updated cluster center to obtain the center mean, update the first updated cluster center according to the center mean to obtain the second updated cluster center; an iteration module, which is used to iteratively process the Y second updated cluster centers, and when the Y second updated cluster centers do not change, determine the Y target clusters according to the Y second updated cluster centers.

[0085] The above-mentioned device signal separation apparatus includes a processor and a memory. The above-mentioned receiving unit 30, input unit 31, determination unit 32, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0086] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problem of low processing accuracy and low processing efficiency in the separation operation of pulse signals in the related art can be solved by adjusting kernel parameters.

[0087] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0088] An embodiment of the present invention provides a computer storage medium, which is used to store a program. When the program is running, the device where the computer storage medium is located is controlled to execute a method for separating device signals.

[0089] Figure 4 is a schematic diagram of an electronic device provided according to an embodiment of the present application, such as Figure 4 As shown, an embodiment of the present invention provides an electronic device, the electronic device 40 includes a processor, a memory, and a program stored in the memory and executable on the processor, the processor is used to execute computer-readable instructions, wherein the computer-readable instructions execute a method for separating device signals when they are executed. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0090] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of a method for separating device signals in various embodiments of the present application.

[0091] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0092] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0095] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0096] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0097] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0098] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0099] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for separating device signals, characterized in that: include: Receive M time domain pulse signals from a power room within a preset time period, and calculate the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes a plurality of switch cabinet devices, each time domain pulse signal refers to a mixed time domain pulse signal of the plurality of switch cabinet devices emitted by the power room, and each time domain pulse signal includes N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; Input the M groups of time domain features into a sparse representation model, output a sparse coefficient matrix, and input the sparse coefficient matrix and the M groups of time domain features into a clustering model, output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer; Y target time-domain pulse signals are determined according to the Y target clusters, wherein the Y target time-domain pulse signals are signals obtained by separating the M time-domain pulse signals.

2. The method according to claim 1, characterized in that The M time domain pulse signals received from the power room within a preset time period include: Acquire the high-frequency current time domain signal of the power room in real time, and send a pulse capture signal to the acquisition device when the amplitude of the high-frequency current time domain signal is greater than the trigger threshold, wherein the acquisition device starts the time domain pulse capture operation when receiving the pulse capture signal; When M high-frequency current time domain signals fed back by the acquisition device are received, N pulse durations are obtained, and the M high-frequency current time domain signals are captured and processed according to the N pulse durations to obtain the M time domain pulse signals.

3. The method according to claim 1, characterized in that: Each set of time domain features includes at least one of the following: signal mean, signal standard deviation, signal peak, pulse width, waveform transition rate, and amplitude change rate. The time domain features of each time domain pulse signal are calculated to include at least one of the following: For a time domain pulse signal, obtain the digit position of the time domain pulse signal in N pulse durations, determine the start index time and the end index time according to the digit position, calculate the difference between the end index time and the start index time, and obtain the pulse width of the time domain pulse signal; For a time domain pulse signal, calculate the absolute value of the difference between two adjacent sub-pulse signals in the time domain pulse signal to obtain N-1 first difference absolute values, and calculate the average value of the N-1 first difference absolute values ​​to obtain the waveform transition rate of the time domain pulse signal; For a time domain pulse signal, calculate the absolute value of the difference between the ratio of two adjacent sub-pulse signals in the time domain pulse signal and a preset parameter to obtain N-1 second absolute values ​​of the difference, calculate the average value of the N-1 second absolute values ​​of the difference, and obtain the amplitude change rate of the time domain pulse signal; For a time domain pulse signal, calculating the mean and standard deviation of N sub-pulse signals associated with the time domain pulse signal to obtain a signal mean and a signal standard deviation of the time domain pulse signal; For a time domain pulse signal, the signal amplitudes of N sub-pulse signals associated with the time domain pulse signal are obtained, the signal amplitude with the largest value is screened out from the N signal amplitudes, and the signal amplitude with the largest value is determined as the signal peak value of the time domain pulse signal.

4. The method according to claim 1, characterized in that: The M groups of time domain features are input into the sparse representation model, and the output sparse coefficient matrix includes: Obtaining a dictionary matrix of the M time-domain pulse signals; A signal vector is constructed using each time domain pulse signal and the time domain features associated with each pulse signal to obtain M signal vectors; Calculating a preset reconstruction error using the dictionary matrix and the M signal vectors to obtain M preset reconstruction errors, wherein the preset reconstruction error refers to a reconstruction error between each time domain pulse signal and a sparse representation of each time domain pulse signal; M candidate sparse coefficient matrices are determined according to the M preset reconstruction errors, and the M candidate sparse coefficient matrices are screened to obtain the sparse coefficient matrix.

5. The method according to claim 4, characterized in that The M candidate sparse coefficient matrices are screened to obtain the sparse coefficient matrix including: Calculate the sparsity penalty term of the candidate sparse coefficient matrix corresponding to each preset reconstruction error to obtain M sparsity penalty terms; Obtain a trade-off coefficient, calculate the product of each sparsity penalty term and the trade-off coefficient, and obtain M trade-off products; Calculate the sum of each trade-off product and each preset reconstruction error to obtain M preset sums, select the preset sum with the smallest value from the M preset sums, and determine the candidate sparse coefficient matrix associated with the preset sum with the smallest value as the sparse coefficient matrix.

6. The method according to claim 1, characterized in that The sparse coefficient matrix and the M groups of time domain features are input into the clustering model, and the output Y target clusters include: Obtaining a bandwidth parameter, wherein the bandwidth parameter is used to control the width of the kernel function in the clustering model; According to the bandwidth parameter, the Gaussian kernel function of any two time domain pulse signals in each time domain pulse signal is calculated to obtain M Gaussian kernel functions, wherein the Gaussian kernel function is used to characterize the similarity between every two time domain pulse signals; The means of the M Gaussian kernel functions are calculated to obtain a density estimation value, an initial value of a cluster center is determined according to the density estimation value to obtain Y initial cluster centers, and the Y target clusters are determined based on the Y initial cluster centers.

7. The method according to claim 6, characterized in that Determining an initial value of a cluster center according to the density estimation value to obtain Y initial cluster centers, and determining the Y target clusters based on the Y initial cluster centers includes: For a time domain pulse signal, calculate the distance between the time domain pulse signal and the Y initial cluster centers to obtain a set of cluster labels, cluster the time domain pulse information according to the set of cluster labels to obtain Y cluster centers, and determine Y first updated cluster centers according to the Y cluster centers; For each first updated cluster center, calculating the mean of the time domain pulse signal contained in the first updated cluster center to obtain a center mean, and updating the first updated cluster center according to the center mean to obtain a second updated cluster center; Iterative processing is performed on the Y second updated cluster centers, and when the Y second updated cluster centers do not change, the Y target clusters are determined according to the Y second updated cluster centers.

8. A device for separating device signals, characterized in that: include: A receiving unit, used to receive M time domain pulse signals from a power room within a preset time period, and calculate the time domain features of each time domain pulse signal to obtain M groups of time domain features, wherein the power room includes a plurality of switch cabinet devices, each time domain pulse signal refers to a mixed time domain pulse signal of the plurality of switch cabinet devices emitted by the power room, and each time domain pulse signal includes N sub-pulse signals, M is greater than or equal to 2, and M and N are positive integers; An input unit, used to input the M groups of time domain features into a sparse representation model, output a sparse coefficient matrix, and input the sparse coefficient matrix and the M groups of time domain features into a clustering model, and output Y target clusters, wherein the sparse representation model is used to perform signal separation processing on the M time domain pulse signals, the sparse coefficient matrix refers to the coefficient vector of the sparse representation corresponding to the M groups of time domain features, and Y is a positive integer; A determination unit is used to determine Y target time-domain pulse signals according to the Y target clusters, wherein the Y target time-domain pulse signals refer to signals obtained after separating the M time-domain pulse signals.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for separating device signals according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the device signal separation method described in any one of claims 1 to 7.

Images