Switch cabinet fault detection method, device, equipment and medium

Abstract

The invention discloses a switch cabinet fault detection method, device and equipment and a medium. The method comprises the following steps: acquiring an air sound signal and a structure vibration signal of a switch cabinet to be detected; preprocessing the air sound signal and the structure vibration signal to obtain multi-channel time frequency data; determining a time-frequency coherence coefficient according to the multi-channel time-frequency data, and determining a target mask according to the time-frequency coherence coefficient and a coherence threshold; determining an air sound component signal and a structure sound component signal according to the target mask and the multi-channel time frequency data; and determining the fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal. According to the technical scheme, the air sound component and the structure sound component are obtained based on multi-channel time frequency data separation, feature aliasing and mutual interference between channels are reduced, and the fault sound source type is analyzed and determined based on the separated component signals, so that identification of the fault propagation path in the switch cabinet is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a switch cabinet fault detection method, device, equipment and medium. BACKGROUND

[0002] The switch cabinet is an important device in the power system for controlling, protecting and isolating electrical equipment. During its operation, events such as partial discharge, arc breakdown, mechanical component loosening or fault impact may occur. These faults produce different forms of signals when they occur, including pressure waves (air sound) propagating through the air and elastic waves (structure vibration) propagating along the switch cabinet structure.

[0003] In the prior art, only a single type of sensor is usually used for monitoring, such as using only acoustic sensors to collect air sound signals or using only acceleration sensors to collect structure vibration signals. This single-channel acquisition method is prone to signal aliasing when facing complex working conditions: some faults produce both sound energy in the air path and vibration energy in the structure path, and the two types of signals may overlap or couple in the time domain and frequency domain, making it difficult to accurately distinguish the propagation medium and the real source of the fault based on single-channel signals. SUMMARY

[0004] The present application provides a switch cabinet fault detection method, device, equipment and medium, which separates air sound components and structure sound components based on multi-channel time-frequency data, reduces feature aliasing and mutual interference between channels, and analyzes and determines the fault sound source type based on the separated component signals, to realize the identification of the internal fault propagation path of the switch cabinet.

[0005] According to an aspect of the present application, a switch cabinet fault detection method is provided, comprising:

[0006] obtaining air sound signals and structure vibration signals of a switch cabinet to be detected;

[0007] preprocessing the air sound signals and structure vibration signals to obtain multi-channel time-frequency data;

[0008] determining a time-frequency coherence coefficient according to the multi-channel time-frequency data, and determining a target mask according to the time-frequency coherence coefficient and a coherence threshold;

[0009] determining air sound component signals and structure sound component signals according to the target mask and the multi-channel time-frequency data;

[0010] determining a fault sound source type of the switch cabinet to be detected according to the air sound component signals and the structure sound component signals.

[0011] According to another aspect of the present application, a switch cabinet fault detection device is provided, comprising:

[0012] an acquisition module configured to acquire an air-borne signal and a structure-borne signal of a switch cabinet to be detected;

[0013] a preprocessing module configured to preprocess the air-borne signal and the structure-borne signal to obtain multi-channel time-frequency data;

[0014] a mask determination module configured to determine a time-frequency coherence coefficient according to the multi-channel time-frequency data, and determine a target mask according to the time-frequency coherence coefficient and a coherence threshold;

[0015] a component signal determination module configured to determine an air-borne component signal and a structure-borne component signal according to the target mask and the multi-channel time-frequency data;

[0016] a type determination module configured to determine a fault sound source type of the switch cabinet to be detected according to the air-borne component signal and the structure-borne component signal.

[0017] According to another aspect of the present application, an electronic device is provided, which comprises:

[0018] at least one processor; and

[0019] a memory connected to the at least one processor in communication; wherein,

[0020] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the switch cabinet fault detection method according to any one of the embodiments of the present application.

[0021] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to execute the switch cabinet fault detection method according to any one of the embodiments of the present application when executed by the processor.

[0022] The technical scheme of the embodiment of the present application obtains the air sound signal and the structure vibration signal of the switch cabinet to be detected, pre-processes the air sound signal and the structure vibration signal to obtain multi-channel time-frequency data, determines a time-frequency coherence coefficient according to the multi-channel time-frequency data, determines a target mask according to the time-frequency coherence coefficient and a coherence threshold, determines an air sound component signal and a structure sound component signal according to the target mask and the multi-channel time-frequency data, and determines the fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal. The technical scheme separates the air sound component and the structure sound component based on the multi-channel time-frequency data, reduces the feature aliasing and mutual interference between channels, and determines the fault sound source type based on the separated component signals, so as to realize the identification of the internal fault propagation path of the switch cabinet.

[0023] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0025] Figure 1 is a flow chart of a switch cabinet fault detection method provided according to the first embodiment of the present application;

[0026] Figure 2 is a flow chart of a switch cabinet fault detection method provided according to the second embodiment of the present application;

[0027] Figure 3 is a structural schematic diagram of a switch cabinet fault detection device provided according to the third embodiment of the present application;

[0028] Figure 4 is a structural schematic diagram of an electronic device provided according to the fourth embodiment of the present application. DETAILED DESCRIPTION

[0029] In the following, the technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative effort should belong to the protection scope of the present application.

[0030] It should be noted that the terms "first", "second" and "target" and the like in the description, claims, and drawings of the present application are intended to distinguish similar objects, and are not necessarily used to describe a particular order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0031] Embodiment one

[0032] Figure 1 It is a flow chart of a switch cabinet fault detection method according to the embodiment one of the present application. The embodiment can be applicable to the case of detecting the fault of the switch cabinet. The method can be executed by a switch cabinet fault detection device. The switch cabinet fault detection device can be realized in the form of hardware and / or software. The switch cabinet fault detection device can be configured in an electronic device with data processing capability. As shown in the figure, the method comprises: Figure 1

[0033] S110, obtaining the air sound signal and the structure vibration signal of the switch cabinet to be detected.

[0034] In the embodiment, the switch cabinet to be detected can be an electric power switch cabinet device that needs to be detected for fault. The air sound signal can be an air sound signal propagating in the environment of the switch cabinet. In the embodiment, the air vibration can be excited by the internal electrical or mechanical fault of the switch cabinet, and the pressure wave signal in the form of sound wave is propagated in the air medium. The structure vibration signal can be a structure vibration signal excited by the internal mechanical movement or electrical fault impact force of the switch cabinet, and the elastic wave signal in the form of elastic wave is propagated in the solid medium such as the metal cabinet body and the element shell of the switch cabinet.

[0035] ​In the embodiment, the air acoustic signals of the switch cabinet to be detected can be collected by the acoustic sensor, and the structural vibration signals of the switch cabinet to be monitored can be synchronously obtained by the structural vibration sensor. Specifically, in the embodiment, the acoustic sensor and the structural vibration sensor can be respectively arranged on the shell or the internal panel of the switch cabinet. The acoustic sensor is preferably a condenser microphone or a piezoelectric microphone, which is used to collect the air acoustic signals propagating in the environment of the switch cabinet. The structural vibration sensor can be an accelerometer or a piezoelectric vibration pickup, which is used to monitor the structural vibration response of the shell, the bus support or the mechanism base of the switch cabinet. In order to ensure the corresponding relationship of the collected signals in the time dimension, the two types of sensors are synchronously triggered and sampled by the same data acquisition card, and the data sampling rate is generally set to 48 kHz-96 kHz, so as to fully cover the energy band of the typical discharge sound, mechanical impact sound and local vibration signal in the operation of the switch cabinet. The collected air acoustic signals and structural vibration signals are respectively stored as time sequence data.

[0036] In S120, the air acoustic signals and the structural vibration signals are preprocessed to obtain multi-channel time-frequency data.

[0037] The preprocessing can be time synchronization, amplitude normalization and band-pass filtering, etc. The multi-channel time-frequency data can be a multi-channel independent representation time-frequency domain complex data set formed by synchronously aligning the preprocessed air acoustic signals and structural vibration signals after short-time Fourier transform.

[0038] In the embodiment, the air acoustic signals and the structural vibration signals can be preprocessed by time synchronization, amplitude normalization and band-pass filtering, etc. to obtain preprocessed data. The multi-channel independent representation time-frequency domain complex data set formed by synchronously aligning the preprocessed data after short-time Fourier transform, i.e. the multi-channel time-frequency data, can be obtained.

[0039] In the embodiment, the air acoustic signals and the structural vibration signals can be preprocessed to obtain multi-channel time-frequency data, including: respectively performing time synchronization, amplitude normalization and band-pass filtering on the air acoustic signals and the structural vibration signals to obtain processed air acoustic signals and processed structural vibration signals; and performing short-time Fourier transform on the processed air acoustic signals and the processed structural vibration signals to obtain the multi-channel time-frequency data.

[0040] The time synchronization can be a processing operation of completely aligning the sampling time stamps of the air sound signal and the structure vibration signal. The amplitude normalization can be a processing operation of mapping the amplitudes of the air sound signal and the structure vibration signal to a uniform numerical interval. The band-pass filtering can be a processing operation of screening effective frequency components in the air sound and the structure vibration signal and filtering out irrelevant frequency band noise. The processed air sound signal can refer to an air sound signal obtained after time synchronization, amplitude normalization and band-pass filtering processing operations. The processed structure vibration signal can refer to a structure vibration signal obtained after time synchronization, amplitude normalization and band-pass filtering processing operations. The Short-Time Fourier Transform (STFT) can be a processing operation of dividing a time domain signal into multiple segments, and performing Fourier transform on each segment separately, and finally obtaining a two-dimensional distribution of the frequency components of the signal varying with time.

[0041] In the embodiment, due to the original signal, the sampling is not synchronized, the amplitudes are inconsistent, and the background noise is affected, resulting in distortion of subsequent time-frequency analysis. Moreover, the response time delay and sensitivity of the acoustic sensor and the vibration sensor are not synchronized, and the sampling system clock error will introduce millisecond-level offset. At the same time, the output level of different sensors and the transmission path gain difference are significant, combined with electromagnetic environmental noise interference, causing distortion of coherent analysis and spectral characteristics. Therefore, in order to eliminate the offset introduced by the sensor characteristic difference and the sampling system, in the embodiment, first, time synchronization calibration is performed on the air sound signal and the structure vibration signal, which can be realized by pulse synchronization signal or cross-correlation peak alignment, and then the signal amplitude is normalized. The normalization coefficient can be the root mean square or the maximum amplitude of each channel signal. Then, the synchronized signal is subjected to band-pass filtering processing, wherein the filtering range can be set to 1 kHz-30 kHz, or can be set according to actual needs, so as to remove the high-frequency noise caused by the direct current drift and electromagnetic interference, to obtain the processed air sound signal and the processed structure vibration signal.

[0042] In the embodiment, after the time synchronization, amplitude normalization and band-pass filtering processing operations, the air sound signal and the structure vibration signal are converted into signal data with equal amplitude characteristics and time sequence alignment, and then the processed air sound signal and the processed structure vibration signal are subjected to short-time Fourier transform processing respectively, to obtain two-dimensional corresponding time-frequency domain complex spectrum matrices, and then the two-dimensional corresponding time-frequency domain complex spectrum matrices are stacked by dimensions to construct a synchronous and aligned multi-channel time-frequency data.

[0043] In the embodiment, through such a setting, the data consistency is improved, the comparability of the air sound and the structure vibration signal on the time axis and the amplitude is ensured, the amplitude deviation caused by the sensitivity and path loss of different sensors is eliminated through normalization processing, and the direct current drift and high-frequency interference components are also removed, improving the data quality.

[0044] S130, determining a time-frequency coherence coefficient according to the multi-channel time-frequency data, and determining a target mask according to the time-frequency coherence coefficient and a coherence threshold.

[0045] The time-frequency coherence coefficient can be a quantitative index representing the degree of linear correlation between the air-borne signal and the structure-borne signal at each time-frequency point. In this embodiment, the time-frequency coherence coefficient can be represented by a matrix. The target mask can be an adaptive mask for classifying time-frequency components based on the time-frequency coherence coefficient. In this embodiment, the target mask can be a time-frequency domain binary matrix determined by comparing the time-frequency coherence coefficient with the coherence threshold. The target mask in this embodiment can include a structure-borne mask and an air-borne mask. The coherence threshold can be a threshold value preset for the time-frequency coherence coefficient. For example, the value of the coherence threshold can be set to 0.7-0.9. In this embodiment, the coherence threshold can be dynamically corrected according to a preset statistical threshold or historical operation data.

[0046] In this embodiment, the multi-channel time-frequency data can be subjected to short-time Fourier transform processing to obtain complex frequency spectrum matrix data of the air-borne signal and the structure-borne signal. Then, the cross-power spectral density and the self-power spectral density are determined based on the complex frequency spectrum matrix data of the air-borne signal and the structure-borne signal, respectively. Then, the time-frequency coherence coefficient is obtained by operating the cross-power spectral density and the self-power spectral density. Then, based on the difference relationship between the time-frequency coherence coefficient and the coherence threshold, the structure-borne mask and the air-borne mask are determined.

[0047] S140, determining an air-borne component signal and a structure-borne component signal according to the target mask and the multi-channel time-frequency data.

[0048] The air-borne component signal can be an independent acoustic signal component with air as the propagation medium in the switch cabinet. The structure-borne component signal can be a vibration-sound coupled signal component highly coherent with the structure vibration of the switch cabinet, excited by equipment fault or normal operation. It can be understood that the air-borne component signal in this embodiment mainly corresponds to environmental background noise or air disturbance signal unrelated to equipment failure. The structure-borne component signal directly reflects the mechanical or electrical state of the switch cabinet.

[0049] In this embodiment, the target mask can be applied to the multi-channel time-frequency data for time-frequency domain screening, and then the time-domain component signals are reconstructed by inverse transformation, so as to extract the corresponding air sound component signals and structure sound component signals. Specifically, in this embodiment, the structure sound mask in the target mask can be applied to the frequency spectrum matrix of the structure vibration signal, and the air sound mask can be applied to the frequency spectrum matrix of the air sound signal, so as to screen out the time-frequency components of the structure sound and the air sound respectively, obtain the screened frequency spectrum matrix, and then convert the frequency spectrum matrix back to the time-domain signal by using the inverse short-time Fourier transform, and further extract the corresponding air sound component signals and structure sound component signals.

[0050] S150, determining the fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal.

[0051] The fault sound source type can refer to a specific fault category that causes the abnormal sound vibration signal of the switch cabinet. The fault sound source type in this embodiment can include an air sound type and a structure sound type. In this embodiment, the energy ratio of the air sound component signal and the structure sound component signal in the time domain can be analyzed with the corresponding ratio threshold value, and then the fault sound source type of the switch cabinet to be detected can be determined.

[0052] The technical scheme of the embodiment of the application comprises the following steps: obtaining the air sound signal and the structure vibration signal of the switch cabinet to be detected; preprocessing the air sound signal and the structure vibration signal to obtain multi-channel time-frequency data; determining a time-frequency coherence coefficient according to the multi-channel time-frequency data, and determining a target mask according to the time-frequency coherence coefficient and a coherence threshold value; determining the air sound component signal and the structure sound component signal according to the target mask and the multi-channel time-frequency data; and determining the fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal. According to the technical scheme, the air sound component and the structure sound component are separated based on the multi-channel time-frequency data, the feature aliasing and mutual interference between channels are reduced, and the fault sound source type is analyzed and determined based on the separated component signals, so as to realize the identification of the internal fault propagation path of the switch cabinet.

[0053] Embodiment two

[0054] Figure 2 is a flowchart of a switch cabinet fault detection method according to the embodiment two of the application. The embodiment is optimized based on the above-mentioned embodiment. The specific optimization is that the time-frequency coherence coefficient is determined according to the multi-channel time-frequency data, which comprises the following steps: determining first complex frequency spectrum data and second complex frequency spectrum data based on the multi-channel time-frequency data; determining mutual power spectral density and self-power spectral density according to the first complex frequency spectrum data and the second complex frequency spectrum data; and determining the time-frequency coherence coefficient according to the mutual power spectral density and the self-power spectral density. As shown in Figure 2 the method comprises the following steps:

[0055] S210, acquire the air sound signal and the structure vibration signal of the switch cabinet to be detected.

[0056] S220, pre-process the air sound signal and the structure vibration signal to obtain multi-channel time-frequency data.

[0057] S230, determine first complex frequency spectrum data and second complex frequency spectrum data based on the multi-channel time-frequency data.

[0058] The first complex frequency spectrum data can be a complex frequency spectrum matrix corresponding to the air sound signal. The second complex frequency spectrum data can be a complex frequency spectrum matrix corresponding to the structure vibration signal. In this embodiment, short-time Fourier transform processing can be performed on the multi-channel time-frequency data to obtain complex frequency spectrum representations of the air sound signal and the structure vibration signal, i.e., the first complex frequency spectrum data and the second complex frequency spectrum data.

[0059] In this embodiment, due to the short-time mutation characteristics (such as discharge / impact) of the acoustic fault signal, if the window function is too wide, time blurring will occur, and if the window function is too narrow, spectral accuracy will be lost. In addition, spectral leakage and boundary effects will cause energy to migrate across frequencies, affecting coherence results. To specifically solve this problem, this embodiment can perform windowed framing processing on the air sound signal and the structure vibration signal contained in the multi-channel time-frequency data to calculate the amplitude-frequency distribution with a fixed frame length, wherein the windowing function is a Hanning window or a Kaiser window, and the frame shift is not greater than 50% of the frame length. The windowed air sound signal and the structure vibration signal are obtained. Then, the windowed air sound signal and the structure vibration signal are subjected to fast Fourier transform to obtain complex frequency spectrum coefficients, i.e., the first complex frequency spectrum matrix and the second complex frequency spectrum matrix.

[0060] In this embodiment, Hanning window or Kaiser window weighting framing can be used, and the frame shift can be set to not greater than 50% of the frame length, which can ensure spectral smoothing and reasonable delay control, thereby improving time-frequency resolution and calculation stability to reduce edge effects.

[0061] S240, determine mutual power spectral density and self-power spectral density according to the first complex frequency spectrum data and the second complex frequency spectrum data.

[0062] The mutual power spectral density can be a complex matrix representing the correlation of the energy and phase of two signals at each time-frequency point, and is used to reflect the time-frequency domain coupling relationship between the air sound signal and the structure vibration signal. The self-power spectral density can be a real number matrix representing the energy intensity distribution of a single signal at each time-frequency point, and is used to reflect the frequency component and energy size of the signal itself.

[0063] The complex spectrum data of the air sound signal and the complex spectrum data of the structure vibration signal can be subjected to complex conjugate multiplication operation to form a cross spectrum matrix in the embodiment, and then time average or exponential weighted average is performed on each frequency point of the cross spectrum matrix to obtain a smoothed cross power spectrum density. In the embodiment, the complex spectrum of the air sound signal and the complex spectrum of the structure vibration signal can be subjected to self-multiplication operation respectively to obtain a corresponding self-spectrum matrix, and the same time average or exponential weighted average is performed on the self-spectrum matrix to eliminate transient energy mutation to obtain a self-power spectrum density.

[0064] The complex conjugate operation on the complex spectrum matrix and the time average or exponential weighted average can effectively suppress transient energy fluctuation and avoid the situation that the calculation result is oscillated or distorted due to the fact that the cross spectrum calculation relies on multi-frame average to offset noise, but the energy peak of transient impact destroys the stationary hypothesis.

[0065] S250, determining a time-frequency coherence coefficient according to the cross power spectrum density and the self-power spectrum density.

[0066] In the embodiment, the modulus of the cross power spectrum density can be divided by the square root of the product of the modulus of the self-power spectrum density of the air sound signal and the self-power spectrum density of the structure vibration signal, and finally a time-frequency coherence coefficient matrix between 0 and 1 is generated at each frame and each frequency point. When the denominator is less than a preset minimum threshold value, the preset minimum threshold value is used to replace the calculation to avoid numerical overflow. In the embodiment, the threshold replacement can be used when the denominator is too small to ensure that the result is limited to the effective interval of 0-1, thereby ensuring that the coherence matrix value is stable and the value range is clear; and the problem of numerical divergence of the calculation result due to the fact that the self-power spectrum is close to zero when the energy of a certain frequency point signal is too low or is overwhelmed by noise and the denominator is too small.

[0067] S260, determining a target mask according to the time-frequency coherence coefficient and a coherence threshold.

[0068] In the embodiment, the target mask includes a structure sound mask and an air sound mask; and correspondingly, determining the target mask according to the time-frequency coherence coefficient and the coherence threshold includes: determining a difference relationship between the time-frequency coherence coefficient and the coherence threshold, and determining the structure sound mask and the air sound mask according to the difference relationship.

[0069] The difference relationship can be obtained by subtracting the coherence threshold from each element contained in the time-frequency coherence coefficient. The structure sound mask can be used to mark the time-frequency interval in which the air sound and the structure vibration are highly coherent in the multi-channel time-frequency data. The air sound mask can be used to mark the time-frequency interval in which the air sound and the structure vibration have no significant coherence in the multi-channel time-frequency data.

[0070] In this embodiment, due to the coherence characteristics of the sound source distribution fluctuate with the ambient noise, the current state, and the looseness of the switching mechanism, the fixed criterion cannot dynamically respond to the translation and expansion of the coherence characteristics. Therefore, in this embodiment, the difference relationship between the time-frequency coherence coefficient and the coherence threshold value can be determined, and then the structural sound mask and the air sound mask are determined based on the difference relationship between the time-frequency coherence coefficient and the coherence threshold value. Specifically, in this embodiment, a weight function can be established according to the difference relationship between the time-frequency coherence coefficient and the coherence threshold value and normalized to obtain two types of mask complementary matrices, that is, the structural sound mask and the air sound mask, and then the adaptive separation of structural sound and air sound is realized, and the complementary mask is formed.

[0071] In a specific implementation, due to the continuous distribution of time-frequency energy on the frequency spectrum, the binary mask will forcibly classify the frequency bands near the critical value, causing energy aliasing or missing. Therefore, in this embodiment, the time-frequency coherence coefficient and the coherence threshold value of each frame and each frequency point can also be compared point by point. When the time-frequency coherence coefficient is higher than the coherence threshold value, the corresponding frequency point is determined as a structural sound feature frequency point, and its mask weight is assigned to the structural sound mask; when the time-frequency coherence coefficient is lower than the coherence threshold value, the corresponding frequency point is determined as an air sound feature frequency point, and its mask weight is assigned to the air sound mask. The mask weight can be linearly normalized according to the difference between the time-frequency coherence coefficient and the coherence threshold value to obtain a continuously changing weight value between 0 and 1, thereby forming a complementary distribution of the structural sound mask and the air sound mask, and thus obtaining a continuously changing mask weight to improve the separation accuracy.

[0072] In this embodiment, by such a setting, the corresponding structural sound mask and air sound mask can be obtained by separation according to the difference relationship between the time-frequency coherence coefficient and the coherence threshold value, and the adaptive separation of structural sound and air sound is realized.

[0073] Further, in this embodiment, the coherence threshold value can be dynamically corrected according to the preset statistical threshold or historical operation data, but due to the mechanical looseness, arc intensity, or load current fluctuation in the long-term operation of the switch cabinet, the energy coupling characteristics will change, causing the original threshold value to be invalid. Therefore, in this embodiment, the update trigger condition of the coherence threshold value can be determined according to the time sequence change of the average coherence degree, wherein if the change amplitude of the average coherence degree exceeds the preset offset threshold value, the coherence threshold value is adaptively updated in a moving average manner, so as to trigger the moving average correction when the offset exceeds the threshold value based on the time sequence change of the average coherence degree, and realize online update. The preset offset threshold value can be set according to actual needs.

[0074] In addition, in the embodiment, the step of determining the update trigger condition of the coherence threshold according to the time sequence change of the average coherence is to solve the problem that the sliding average window is too short and changes dramatically with noise, and is too long and responds slowly. In this regard, the average time window statistics can be performed on the average value of the time-frequency coherence coefficient to form the average coherence time sequence curve. When the average coherence change amplitude between adjacent time windows exceeds the preset offset threshold, the coherence threshold update process is triggered. The current coherence threshold is modified by using a sliding average method, wherein the new coherence threshold is obtained by weighting the current threshold and the historical threshold according to a preset smoothing coefficient; if the continuous update amplitude is lower than the offset threshold for multiple times, the coherence threshold is kept unchanged to maintain dynamic stability, so that the coherence threshold is kept stable and dynamic change, and oscillation is prevented.

[0075] In the embodiment, when the fault is in the early stage or the energy is weak (such as initial partial discharge or slight mechanical friction), the amplitudes of the air sound and the structure vibration signal are close to the background noise level. At this time, the mask weight based on the time-frequency coherence calculation is easily disturbed by instantaneous noise peaks, environmental vibration changes and other factors, and large fluctuations occur. Such fluctuations can cause distortion of the energy of the separated signal in the mask application stage, and then affect the accuracy of the subsequent energy ratio analysis and fault source discrimination. Therefore, after the adaptive mask is generated according to the time-frequency coherence coefficient, the smoothing coefficient can be determined according to the local energy variance in the target frequency band. When the local energy variance is greater than the preset variance threshold, the smoothing coefficient is increased, and the mask weight is processed by using the smoothing coefficient to perform sliding weighting in the time domain, so that the defect that the mask weight is easily disturbed by noise under the condition of low signal-to-noise ratio and instantaneous fluctuation is overcome, and the stability and consistency of the separated signal are ensured.

[0076] In addition, when the air sound and the structure vibration signal are separated in the process of monitoring the switch cabinet fault, the complementary principle of the adaptive target mask is generally used, that is, the mask value of one signal is high at a certain time-frequency point, and the mask value of the other signal is low at the time-frequency point, so that effective separation is achieved. However, in the high frequency band, the energy of the two signals often decays sharply at the same time (especially when far away from the sound source), at this time, the coherence calculation result tends to be unstable, the mask complementary relationship will fail, and the mask blank area or double low value area appears, resulting in the lack of characteristics in the frequency band, damaging the integrity of the separated signal, and especially affecting the early partial discharge fault which is sensitive to high frequency characteristics. In this regard, the high frequency energy decay area in the target frequency band can be detected, the mask weight in the high frequency band is reconstructed by using the mask weight distribution of adjacent frequency bands for interpolation or extrapolation, and the mask weight in the high frequency band is compensated according to the typical high frequency characteristic template in the preset fault mode library. After reconstruction, the complementary constraint is applied to the structure sound mask and the air sound mask to ensure the integrity of the high frequency band characteristics of the separated signal.

[0077] Specifically, the system first performs energy analysis on the time-frequency diagram to detect the energy attenuation region of the high frequency band in the target frequency band when generating the adaptive mask of the air-borne signal and the structure-borne signal. When detecting that the mask weight value in the region is double low, the system uses the mask weight value distribution of the adjacent middle frequency band of the high frequency band, and adopts a two-dimensional interpolation method to reconstruct the mask weight value of the high frequency attenuation region. At the same time, the system calls a typical high frequency characteristic template matched with the current fault type from a preset fault mode library, superimposes the characteristic mode in the template into the reconstructed mask, and realizes high frequency characteristic compensation. After the mask weight value reconstruction and compensation are completed, the system applies complementary constraint to the masks of the two signals to ensure that the total value of the mask at the same time-frequency point is kept within a set range, and improves the continuity and stability of the separated signals. Thus, the continuity and complementarity of the high frequency mask are restored, the separated high frequency characteristics are more complete, which is helpful for accurate detection of early local discharge and other high frequency characteristic significant fault types, and reduces the misjudgment rate caused by the loss of high frequency characteristics.

[0078] S270, determining an air-borne component signal and a structure-borne component signal according to the target mask and the multi-channel time-frequency data.

[0079] In the embodiment, optionally, determining an air-borne component signal and a structure-borne component signal according to the target mask and the multi-channel time-frequency data includes: performing mask weighting operation based on the multi-channel time-frequency data and the structure-borne mask and the air-borne mask respectively to obtain structure spectrum data and air spectrum data; determining a structure-borne component signal according to the structure-borne mask and the structure spectrum data; and determining an air-borne component signal according to the air-borne mask and the air spectrum data.

[0080] The structure spectrum data can be a structure spectrum matrix of the structure-borne signal obtained by filtering after the multi-channel time-frequency data is multiplied by the structure-borne mask element by element. The air spectrum data is an air spectrum matrix of the air-borne signal obtained by filtering after the air-borne complex spectrum in the multi-channel time-frequency data is multiplied by the air-borne mask element by element.

[0081] In this embodiment, after the mask generation is completed, the complex spectrum multiplication operation causes amplitude deviation and phase distortion, and if the mask weighting lacks amplitude correction or phase recovery, the reconstructed signal of the inverse transform may not be of the same scale and have insufficient correlation. Therefore, in this embodiment, the complex spectrum matrix of the structural vibration signal and the air sound signal contained in the multi-channel time-frequency data can be subjected to mask weighting operation respectively to obtain structural spectrum data and air spectrum data subjected to energy suppression and enhancement processing, and then the weighted structural spectrum data and air spectrum data are subjected to inverse short-time Fourier transform to reconstruct the corresponding time-domain component signals. After reconstruction, amplitude normalization and phase alignment processing are performed to ensure the energy contrast and timing consistency of the air sound component signal and the structural sound component signal. Specifically, in this embodiment, the structural sound mask can be applied to the spectral matrix of the structural vibration signal, the air sound mask can be applied to the air spectrum matrix of the air sound signal, and point-by-point multiplication operation can be performed to achieve energy weighting of the target frequency component, and then the weighted spectral matrix is converted back to the time-domain signal through inverse short-time Fourier transform, thereby obtaining the purified structural sound component signal and air sound component signal.

[0082] In this embodiment, by such a setting, the air sound component signal and the structural sound component signal can be obtained by processing the target mask and the multi-channel time-frequency data, thereby realizing accurate separation of the fault feature signal and the environmental interference signal, and greatly improving the reliability, sensitivity and robustness of the switch cabinet state monitoring and fault diagnosis.

[0083] In addition, to ensure the energy comparability and time alignment between signals, amplitude normalization and phase correction processing can be further performed. This processing can significantly suppress non-target channel interference and improve the purity of the separated signals, providing high-quality input for subsequent energy ratio discrimination.

[0084] Optionally, since the relationship between the mask and the fault type is nonlinear and non-Gaussian, a regular threshold cannot fit all modes. Therefore, the adaptive mask can be optimized by using a residual network before being applied to the multi-channel time-frequency data, to update the adaptive mask. Specifically, the time-frequency feature matrix of the original structural sound mask and the original air sound mask can be used as the network input of the residual network, and the original structural sound mask and the original air sound mask can be updated by weighting according to the optimized mask weight output by the residual network, wherein the updated mask is obtained by weighting the original mask according to the optimized mask output by the residual network and the original mask according to a preset fusion coefficient. The optimized adaptive mask is applied to subsequent multi-channel time-frequency data processing to improve the separation accuracy of air sound and structural sound. Thus, the optimization capability of the adaptive mask is improved, and the separation accuracy of complex sound sources is enhanced.

[0085] In practical applications, the sound source features monitored by the switch cabinet show obvious differences in different frequency and time scales, and a single scale cannot balance the transient local and global trends. Therefore, when a single resolution time-frequency feature is input into a residual network, high-frequency details are easily lost, low-frequency energy is leaked, and the separation accuracy is reduced. To this end, the time-frequency feature matrices of the original structural sound mask and the original air sound mask can be processed by multi-scale decomposition before being input into the residual network to generate feature sub-matrices of different resolutions. Then, the residual network receives the feature sub-matrices of different resolutions, calculates the corresponding optimized mask weights, and combines the optimized mask weights of different resolutions according to a preset multi-scale fusion strategy to obtain the final optimized mask weight. Through multi-scale decomposition and fusion, the global structure and local details of the time-frequency feature are retained, the stability of air sound and structural sound separation is improved under different working conditions, and high-frequency detail loss and low-frequency energy leakage are reduced.

[0086] However, due to the large structural differences of different types of switch cabinets, the feature spectrum of fault modes (such as partial discharge, mechanical jamming, and insulation breakdown) is also different. The real fault signal samples collected in the field are often scarce, especially the early and slight fault data. This lack of data can lead to insufficient feature expression of the residual network for certain working conditions, reduced model generalization ability, and weakened effect of the optimized mask in unobserved working conditions, thereby affecting the quality of the separated signal and the accuracy of fault discrimination. To this end, real fault signal samples can be collected, and virtual augmented samples can be generated by processing the real fault signal samples through spectral perturbation. Then, a mixed training set composed of real fault samples, virtual augmented samples, and simulation samples synthesized based on the sound and structure coupling model is input into the residual network to improve the mask optimization generalization ability of the residual network under different devices and different fault modes.

[0087] Specifically, before optimizing the adaptive mask using the residual network, the system can first collect real fault signal samples from the running sites of multiple switch cabinets. To effectively increase the number and diversity of training data, the system inputs the samples into a spectrum perturbation module, generates virtual augmented samples with similar characteristics but differentiated details to the original signals by controlled perturbation of frequency, phase, and specific energy interval. At the same time, the system synthesizes air sound and structure sound signal pairs based on the sound and structure coupling physical simulation model by setting different fault source parameters (including position, strength, and spectral form), forming artificial simulation samples. Finally, the real fault samples, virtual augmented samples, and simulation samples form a mixed training set input into the residual network for training. This improves the accuracy and stability of mask optimization under multiple working conditions and multiple fault categories, effectively enhances the separation and identification performance of the system under early fault and weak signal conditions, and reduces false positives and false negatives caused by model overfitting.

[0088] In addition, since the network optimization result is susceptible to noise in the weak signal area monitored for the switch cabinet fault, and needs to be strengthened in the strong signal area. That is, if a fixed fusion coefficient is used, over-correction or under-correction is likely to occur in the case of uneven energy distribution, causing the mask to be distorted in part of the frequency band. To this end, the fusion coefficient can be dynamically adjusted according to the target frequency band energy concentration before the optimized mask weight of the network output and the original mask weight are fused and calculated, wherein the target frequency band energy concentration is the ratio of the energy value of the target frequency band to the total energy value of the full frequency domain, the target frequency band is the fault characteristic frequency band determined according to the time-frequency analysis result of the air sound or the structure sound, the energy value of the target frequency band is obtained by summing the amplitude square in the time-frequency feature matrix corresponding to the frequency band, and the total energy value of the full frequency domain is obtained by summing the amplitude square in the time-frequency feature matrix corresponding to the full frequency domain. When the energy concentration of the target frequency band is higher than the preset concentration threshold, the weight of the optimized mask weight is increased; when the energy concentration is lower than the concentration threshold, the weight of the original mask weight is increased; thereby the dynamic fusion coefficient based on the energy concentration emphasizes the effect of the optimized mask in the strong signal area, and retains the stability of the original mask in the weak signal area, thereby improving the separation precision.

[0089] S280, determining a fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal.

[0090] In this embodiment, optionally, the fault sound source type of the switch cabinet to be detected is determined according to the air sound component signal and the structure sound component signal, including: respectively determining a first time domain energy and a second time domain energy of the air sound component signal and the structure sound component signal; determining energy ratio data according to the first time domain energy and the second time domain energy, and determining the fault sound source type of the switch cabinet to be detected according to the energy ratio data and a ratio threshold.

[0091] The first time domain energy can be the time domain energy corresponding to the air sound component signal. In this embodiment, the first time domain energy can be considered as the total energy value of the air sound component signal in the time domain, which is used to represent the energy intensity of the environmental interference signal. The second time domain energy can be the time domain energy corresponding to the structure sound component signal. In this embodiment, the second time domain energy can be considered as the total energy value of the structure sound component signal in the time domain, which is used to represent the energy intensity of the switch cabinet body vibration sound coupling characteristic signal. The energy ratio data can be the ratio of the time domain energy of the air sound component signal to the time domain energy of the structure sound component signal. The ratio threshold can be pre-set according to actual needs. The ratio threshold in this embodiment can include a first ratio threshold and a second ratio threshold.

[0092] The energy of the extracted air-borne signal and structure-borne signal in the time domain can be calculated respectively in this embodiment, for example, the root mean square energy value of each channel, so as to obtain the first time domain energy and the second time domain energy, and then the ratio of the time domain energy of the two is determined according to the first time domain energy and the second time domain energy, and by comparing the ratio of the time domain energy with the pre-set ratio threshold value, the specific fault sound source type of the switch cabinet to be detected can be determined according to the comparison result.

[0093] In this embodiment, through such a setting, the fault sound source type can be automatically determined based on the energy ratio analysis of the separated component signals, so as to realize the identification and classification of the internal fault propagation path of the switch cabinet.

[0094] In this embodiment, the ratio threshold value includes a first ratio threshold value and a second ratio threshold value, and the fault sound source type of the switch cabinet to be detected is determined according to the energy ratio data and the ratio threshold value, including: in the case that the energy ratio data is greater than the first ratio threshold value, determining the fault sound source type of the switch cabinet to be detected as an air-borne type; and in the case that the energy ratio data is less than the second ratio threshold value, determining the fault sound source type of the switch cabinet to be detected as a structure-borne type.

[0095] The first ratio threshold value can be a critical threshold value for determining the sound source category as air-borne dominant, and the second ratio threshold value is a critical threshold value for determining the sound source category as structure-borne dominant. The first ratio threshold value and the second ratio threshold value of this embodiment can be determined through a self-calibration experiment of typical characteristics of structure-borne sound and air-borne sound. For example, the numerical value of the first ratio threshold value can be set in the range of 5-10, and the numerical value of the second ratio threshold value can be set in the range of 0.3-0.5, and the specific value can also be set according to actual needs.

[0096] In this embodiment, when the energy ratio data is greater than the pre-set first ratio threshold value, it indicates that the fault sound source is mainly contributed by the air propagation path, and it indicates that the fault sound source category is air-borne dominant at this time, so the fault sound source type is determined as an air-borne type, for example, it can include arc discharge, surface discharge and other fault types. When the energy ratio data is less than the second ratio threshold value, it indicates that the fault sound source category is structure-borne dominant at this time, so the fault sound source type is a structure-borne type, for example, mechanical impact, mechanism jamming or loosening fault. If the energy ratio data is between the first ratio threshold value and the second ratio threshold value, it can be considered that there is a multi-path coupled sound source or further classification processing is performed.

[0097] In this embodiment, through such a setting, through the energy ratio determination mechanism, different sound source types can be quickly distinguished without manual judgment, and automatic classification and identification of the switch cabinet abnormal event can be realized.

[0098] In the embodiment, by synchronously collecting the air sound signal and the structure vibration signal of the switch cabinet under the same time reference, and uniformly pre-processing and multi-channel time-frequency alignment, the coherent relationship of the two types of signals on different frequency components can be quantitatively described in the time-frequency domain, the separation mask suitable for the actual working condition change is dynamically generated according to the coherent characteristics, and then the air sound component and the structure sound component are effectively separated, the feature aliasing and mutual interference between channels are reduced, and then the fault sound source type is automatically determined based on the energy ratio analysis of the separated component signals, so as to realize the identification and classification of the fault propagation path in the switch cabinet.

[0099] Embodiment three

[0100] Figure 3 is a structural schematic diagram of a switch cabinet fault detection device provided according to the embodiment three of the present application. As shown in the figure, the device comprises: Figure 3

[0101] The acquisition module 310 is configured to acquire the air sound signal and the structure vibration signal of the switch cabinet to be detected.

[0102] The preprocessing module 320 is configured to pre-process the air sound signal and the structure vibration signal to obtain multi-channel time-frequency data.

[0103] The mask determination module 330 is configured to determine the time-frequency coherence coefficient according to the multi-channel time-frequency data, and determine the target mask according to the time-frequency coherence coefficient and the coherence threshold.

[0104] The component signal determination module 340 is configured to determine the air sound component signal and the structure sound component signal according to the target mask and the multi-channel time-frequency data.

[0105] The type determination module 350 is configured to determine the fault sound source type of the switch cabinet to be detected according to the air sound component signal and the structure sound component signal.

[0106] Optionally, the preprocessing module 320 is specifically configured to:

[0107] respectively perform time synchronization, amplitude normalization and band-pass filtering processing operations on the air sound signal and the structure vibration signal to obtain the processed air sound signal and the processed structure vibration signal;

[0108] perform short-time Fourier transform processing on the processed air sound signal and the processed structure vibration signal to obtain the multi-channel time-frequency data.

[0109] Optionally, the mask determination module 330 is specifically configured to:

[0110] determine the first complex frequency spectrum data and the second complex frequency spectrum data based on the multi-channel time-frequency data;

[0111] ​determine the cross-power spectral density and the self-power spectral density according to the first complex frequency spectrum data and the second complex frequency spectrum data;

[0112] determine the time-frequency coherence coefficient according to the cross-power spectral density and the self-power spectral density.

[0113] Optionally, the target mask includes a structural sound mask and an air sound mask.

[0114] Correspondingly, the mask determining module 330 is specifically used for:

[0115] determining a difference relationship of the time-frequency coherence coefficient and the coherence threshold value, and determining the structural sound mask and the air sound mask according to the difference relationship.

[0116] Optionally, the component signal determining module 340 is specifically used for:

[0117] performing mask weighting operation on the multi-channel time-frequency data based on the structural sound mask and the air sound mask respectively, to obtain structural spectrum data and air spectrum data;

[0118] determining a structural sound component signal according to the structural sound mask and the structural spectrum data;

[0119] determining an air sound component signal according to the air sound mask and the air spectrum data.

[0120] Optionally, the type determining module 350 includes:

[0121] a time domain energy determining unit, configured to determine first time domain energy and second time domain energy of the air sound component signal and the structural sound component signal respectively;

[0122] a fault type determining unit, configured to determine energy ratio data according to the first time domain energy and the second time domain energy, and determine the fault sound source type of the switch cabinet to be detected according to the energy ratio data and a ratio threshold value.

[0123] Optionally, the ratio threshold value includes a first ratio threshold value and a second ratio threshold value.

[0124] The fault type determining unit is specifically used for:

[0125] in a case where the energy ratio data is greater than the first ratio threshold value, determining that the fault sound source type of the switch cabinet to be detected is an air sound type;

[0126] in a case where the energy ratio data is less than the second ratio threshold value, determining that the fault sound source type of the switch cabinet to be detected is a structural sound type.

[0127] The switch cabinet fault detection device provided in the embodiment of the application can execute the switch cabinet fault detection method provided in any embodiment of the application, and has the corresponding function modules and beneficial effects of the execution method.

[0128] Embodiment Four

[0129] Figure 4 is a structural diagram of an electronic device provided according to Embodiment Four of the present application. The electronic device is intended to represent a variety of forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0130] As shown in Figure 4 , the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., connected in communication with the at least one processor 11, wherein the memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0131] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, a speaker, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0132] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the switchgear fault detection method.

[0133] In some embodiments, the switchgear fault detection method can be implemented as a computer program tangibly embodied in a computer readable storage medium, e.g., storage unit 18. In some embodiments, parts or all of the computer program can be loaded and / or installed onto electronic device 10 via, e.g., ROM 12 and / or communication unit 19. When the computer program is loaded onto RAM 13 and executed by processor 11, one or more steps of the above-described switchgear fault detection method can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the switchgear fault detection method by other any suitable means, e.g., with the aid of firmware.

[0134] The various implementations of the system and techniques described above can be realized in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0135] Computer programs used to implement the processes of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed, can implement the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0136] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0137] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0138] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), blockchain network, and the Internet.

[0139] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0140] It should be understood that the various forms of flow shown above can be reordered, added to, or have steps deleted. For example, the steps described in the present application can be performed in parallel, in series, or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, and this is not limited herein.

[0141] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting faults in a switchgear, characterized in that, include: Acquire the airborne acoustic signal and structural vibration signal of the switchgear to be tested; The airborne sound signal and the structural vibration signal are preprocessed to obtain multi-channel time-frequency data; The time-frequency coherence coefficient is determined based on the multi-channel time-frequency data, and the target mask is determined based on the time-frequency coherence coefficient and the coherence threshold. The airborne acoustic component signal and the structure acoustic component signal are determined based on the target mask and the multi-channel time-frequency data. The fault sound source type of the switchgear under test is determined based on the airborne sound component signal and the structured sound component signal.

2. The method according to claim 1, characterized in that, The airborne acoustic signal and structural vibration signal are preprocessed to obtain multi-channel time-frequency data, including: The airborne sound signal and the structural vibration signal are respectively subjected to time synchronization, amplitude normalization and bandpass filtering to obtain the processed airborne sound signal and the processed structural vibration signal. The processed airborne sound signal and the processed structural vibration signal are subjected to short-time Fourier transform processing to construct multi-channel time-frequency data.

3. The method according to claim 1, characterized in that, Determining the time-frequency coherence coefficient based on the multi-channel time-frequency data includes: The first complex spectrum data and the second complex spectrum data are determined based on the multi-channel time-frequency data; The cross power spectral density and the self power spectral density are determined based on the first complex spectrum data and the second complex spectrum data. The time-frequency coherence coefficient is determined based on the cross-power spectral density and the self-power spectral density.

4. The method according to claim 1, characterized in that, The target mask includes a structured acoustic mask and an airborne acoustic mask; Accordingly, determining the target mask based on the time-frequency coherence coefficient and coherence threshold includes: The difference between the time-frequency coherence coefficient and the coherence threshold is determined, and the structured acoustic mask and the airborne acoustic mask are determined based on the difference.

5. The method according to claim 4, characterized in that, Determining the airborne acoustic component signal and the structure-borne acoustic component signal based on the target mask and the multi-channel time-frequency data includes: Based on the multi-channel time-frequency data and mask weighting operations performed with the structure acoustic mask and the air acoustic mask respectively, structure spectrum data and air spectrum data are obtained. The structural acoustic component signal is determined based on the structural acoustic mask and the structural spectral data; The airborne acoustic component signal is determined based on the airborne acoustic mask and the airborne spectrum data.

6. The method according to claim 1, characterized in that, The fault sound source type of the switchgear under test is determined based on the airborne sound component signal and the structure-borne sound component signal, including: The first time-domain energy and the second time-domain energy of the airborne acoustic component signal and the structure acoustic component signal are determined respectively. The energy ratio data is determined based on the first time domain energy and the second time domain energy, and the fault sound source type of the switch cabinet to be tested is determined based on the energy ratio data and the ratio threshold.

7. The method according to claim 6, characterized in that, The ratio threshold includes a first ratio threshold and a second ratio threshold; The fault sound source type of the switchgear under test is determined based on the energy ratio data and ratio threshold, including: If the energy ratio data is greater than the first ratio threshold, the fault sound source type of the switch cabinet to be tested is determined to be airborne sound. If the energy ratio data is less than the second ratio threshold, the fault sound source type of the switch cabinet to be tested is determined to be structural sound.

8. A switchgear fault detection device, characterized in that, include: The acquisition module is used to acquire the airborne acoustic signal and structural vibration signal of the switchgear under test; The preprocessing module is used to preprocess the airborne acoustic signal and the structural vibration signal to obtain multi-channel time-frequency data; The mask determination module is used to determine the time-frequency coherence coefficient based on the multi-channel time-frequency data, and to determine the target mask based on the time-frequency coherence coefficient and the coherence threshold. The component signal determination module is used to determine the airborne acoustic component signal and the structure acoustic component signal based on the target mask and the multi-channel time-frequency data. The type determination module is used to determine the fault sound source type of the switch cabinet under test based on the airborne sound component signal and the structural sound component signal.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the switch cabinet fault detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the switch cabinet fault detection method according to any one of claims 1-7.

Images