An intelligent fault diagnosis method and system for explosion-proof distribution box
Through the combination of Fourier transform and signal processing technology, time-domain and frequency-domain signal processing channels are built, and the problem of explosion-proof distribution box fault diagnosis relies on manual inspection, achieving efficient and accurate fault diagnosis and real-time monitoring.
Patent Information
- Application Number
- CN202411297778.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-18
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2044-09-18
AI Technical Summary
The fault diagnosis of existing explosion-proof distribution boxes relies on manual inspection and empirical judgment, and is easily affected and cannot achieve real-time monitoring and accurate early warning, resulting in low diagnostic accuracy and efficiency.
Fourier transform and signal processing technology are adopted to build a time domain signal processing channel and a frequency domain signal processing channel through the fusion analysis of time domain signals and frequency domain signals, and combine an abnormality recognition model to realize automated fault diagnosis of voltage waveforms.
It improves the accuracy and efficiency of fault diagnosis of explosion-proof distribution boxes, realizes real-time monitoring and accurate early warning of faults, and enhances the diagnosis ability of new fault types.
Smart Images
Figure CN118861940B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of explosion-proof distribution box technology, and specifically to an intelligent fault diagnosis method and system for explosion-proof distribution boxes. Background Art
[0002] With the continuous development of industrial technology, explosion-proof distribution boxes are important equipment to ensure the safety of industrial production. Their safety and stability directly affect the normal operation of the production line. In the actual operation process, explosion-proof distribution boxes may fail due to various reasons, such as aging of electrical components, loose wiring, and excessively high ambient temperature. If these faults are not discovered and handled in time, they may cause serious safety accidents. Traditional explosion-proof distribution box fault diagnosis mainly relies on manual inspection and experience judgment, requiring maintenance personnel to have rich professional knowledge and practical experience. Through the appearance inspection of the distribution box, the testing of electrical components, and the analysis of operation data, the cause and location of the fault can be judged. Manual inspection and judgment are easily affected by human factors, resulting in low accuracy and efficiency of fault diagnosis, and it is impossible to achieve real-time monitoring and early warning of faults. It is difficult to adapt to new fault types and diagnostic needs, which increases the risk of production lines to a certain extent.
[0003] Therefore, in the current explosion-proof distribution box fault diagnosis related technologies, there are technical problems that rely on manual inspections and experience-based judgments, which are easily affected and cannot achieve real-time monitoring and accurate early warning of faults, resulting in low accuracy and efficiency in explosion-proof distribution box fault diagnosis. The technical effect of improving the accuracy and efficiency of distribution box fault diagnosis has been achieved. Summary of the invention
[0004] The present application provides an intelligent fault diagnosis method and system for explosion-proof distribution boxes, and adopts technical means such as Fourier transform, construction of signal processing channels, and fusion processing to solve the technical problems of the existing explosion-proof distribution box fault diagnosis, which relies on manual inspection and experience judgment and is easily affected, and cannot achieve real-time monitoring and accurate early warning of faults, resulting in low accuracy and efficiency of explosion-proof distribution box fault diagnosis, thereby achieving the technical effect of improving the accuracy and efficiency of distribution box fault diagnosis.
[0005] The present application provides an intelligent fault diagnosis method for an explosion-proof distribution box, the method comprising: obtaining a voltage waveform of the explosion-proof distribution box, the voltage waveform being collected by an oscilloscope; performing signal conversion on the voltage waveform, the signal conversion referring to converting a time domain signal of the voltage waveform into a frequency domain signal by Fourier transform; establishing a corresponding relationship between the time domain signal and the frequency domain signal; constructing a time domain signal processing channel and a frequency domain signal processing channel, the time domain signal processing channel being used for performing abnormality identification and analysis on the time domain signal, and the frequency domain signal processing channel being used for performing abnormality identification and analysis on the frequency domain signal; establishing a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel, and based on the corresponding relationship between the time domain signal and the frequency domain signal, fusing and outputting a temporal abnormality analysis result of the time domain signal processing channel and a frequency abnormality analysis result of the frequency domain signal processing channel to obtain a fault diagnosis result.
[0006] In a possible implementation, the time domain signal processing channel and the frequency domain signal processing channel are constructed, and the following processing is performed: the signal processing channel composition is configured, including a signal screening module, a signal processing module, and an abnormality recognition model, wherein the signal screening module is used to screen the voltage waveform signal according to a preset target, the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency feature analysis processing on the time domain signal, the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectral density analysis processing on the frequency domain signal, the abnormality recognition model is used to perform abnormality recognition analysis on the signal imported by the signal processing module, and the abnormality recognition module is an operational model that has been converged through supervised training with a training data set.
[0007] In a possible implementation, the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency feature analysis processing on the time domain signal, and performs the following processing: decomposing according to different time scales according to the time mapping relationship of the time domain signal to obtain multiple decomposition scale signals; identifying the frequency features of the decomposition scale signals to obtain the characteristics of the signals at each scale; based on the time mapping relationship between different time scales, calculating the similarity of the characteristics of the signals at each scale to obtain the feature similarity; and performing abnormal time positioning and frequency feature labeling according to the feature similarity.
[0008] In a possible implementation, based on the time mapping relationship between different time scales, similarity calculation is performed on the signal features of each scale to obtain feature similarity, and the following processing is also performed: based on the time mapping relationship, the signals of the different time scales are aligned; based on the alignment relationship of frequency features, the distance between each pair of aligned frequency features is calculated, and the distances are accumulated to construct a distance matrix; based on the distance matrix, a path search is performed according to path constraints to obtain the path with the minimum distance to generate the feature similarity, wherein the path constraints include a search starting point and a search direction.
[0009] In a possible implementation, abnormal time location and frequency feature labeling are performed according to the feature similarity, and the following processing is also performed: frequency features whose feature similarity does not meet the threshold are screened to locate the different time scales; minimum time scale location decomposition is performed based on the different time scales to determine the abnormal time location; frequency feature labeling is performed according to the abnormal time location.
[0010] In a possible implementation, the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectrum density analysis on the frequency domain signal, and also performs the following processing: establish a fault characteristic frequency, use the fault characteristic frequency to traverse and compare the frequency distribution of the frequency domain signal, and mark the fault characteristic frequency; identify abnormal peaks based on the frequency distribution, and mark abnormal peaks of the characteristic frequency; perform power spectrum density analysis on the frequency domain signal to obtain the distribution relationship between power and frequency.
[0011] In a possible implementation, based on the correspondence between the time domain signal and the frequency domain signal, the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel are fused and output to obtain a fault diagnosis result, and the following processing is also performed: based on the correspondence between the time domain signal and the frequency domain signal, a correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result is established; according to the temporal anomaly analysis result and the frequency anomaly analysis result, a fault type and a fault target are respectively obtained; it is determined whether the fault type and the fault target satisfy the correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result; if satisfied, the fault type and the fault target are used as the fault diagnosis result; if not satisfied, the topological structure of the fault target is obtained, the voltage influence relationship of the fault target is determined, and a fusion coefficient is configured; based on the fusion coefficient, the temporal anomaly analysis result and the frequency anomaly analysis result are fused to obtain the fault diagnosis result.
[0012] In a possible implementation, the temporal anomaly analysis result and the frequency anomaly analysis result are fused based on the fusion coefficient to obtain the fault diagnosis result, and the following processing is also performed: a correlation analysis is performed based on the fault types of the temporal anomaly analysis result and the frequency anomaly analysis result to determine the cause fault type and the result fault type, wherein the cause fault type refers to the fault type that causes the result fault type; a priority coefficient is configured based on the correlation, wherein the priority coefficient of the result fault type is higher than the priority coefficient of the cause fault type; a fault target and a fault type are searched based on the fusion coefficient and the priority coefficient, and the anomaly analysis result with the highest evaluation result is determined as the fault diagnosis result.
[0013] The present application also provides an intelligent fault diagnosis system for an explosion-proof distribution box, comprising:
[0014] A voltage waveform acquisition module, which is used to acquire the voltage waveform of the explosion-proof distribution box, and the voltage waveform is collected by an oscilloscope;
[0015] A waveform signal conversion module, the waveform signal conversion module is used to perform signal conversion on the voltage waveform, the signal conversion refers to converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform;
[0016] A corresponding relationship establishing module, the corresponding relationship establishing module is used to establish a corresponding relationship between the time domain signal and the frequency domain signal;
[0017] A signal processing channel construction module, wherein the signal processing channel construction module is used to construct a time domain signal processing channel and a frequency domain signal processing channel, wherein the time domain signal processing channel is used to perform anomaly identification and analysis on the time domain signal, and the frequency domain signal processing channel is used to perform anomaly identification and analysis on the frequency domain signal;
[0018] An analysis result fusion output module is used to establish a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel. Based on the correspondence between the time domain signal and the frequency domain signal, the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel are fused and output to obtain a fault diagnosis result.
[0019] The intelligent fault diagnosis method and system for explosion-proof distribution boxes proposed in this application is intended to obtain the voltage waveform of the explosion-proof distribution box; perform signal conversion on the voltage waveform, which means converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform; establish a correspondence between the time domain signal and the frequency domain signal; construct a time domain signal processing channel and a frequency domain signal processing channel; establish a fusion connection layer of the processing channel, fuse and output the temporal anomaly analysis results of the time domain signal processing channel and the frequency anomaly analysis results of the frequency domain signal processing channel, and obtain the fault diagnosis result. This solves the technical problem that the existing fault diagnosis of explosion-proof distribution boxes relies on manual inspections and experience judgments, is easily affected, cannot realize real-time monitoring and accurate early warning of faults, and leads to low accuracy and efficiency of fault diagnosis of explosion-proof distribution boxes, and achieves the technical effect of improving the accuracy and efficiency of fault diagnosis of distribution boxes. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.
[0021] Figure 1 A schematic diagram of a flow chart of an intelligent fault diagnosis method for an explosion-proof distribution box provided in an embodiment of the present application;
[0022] Figure 2 A schematic diagram of the structure of an intelligent fault diagnosis system for an explosion-proof distribution box provided in an embodiment of the present application.
[0023] Explanation of reference numerals: voltage waveform acquisition module 10 , waveform signal conversion module 20 , corresponding relationship establishment module 30 , signal processing channel construction module 40 , analysis result fusion output module 50 . DETAILED DESCRIPTION
[0024] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.
[0025] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.
[0026] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server 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 modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.
[0027] The present application embodiment provides an intelligent fault diagnosis method for an explosion-proof distribution box, such as Figure 1 As shown, the method includes:
[0028] Step S100, obtain the voltage waveform of the explosion-proof distribution box, and the voltage waveform is collected by an oscilloscope. Use the oscilloscope to measure and record the waveform of the voltage output by the explosion-proof distribution box that changes with time, wherein the oscilloscope can display the dynamic changes of the voltage signal in real time, and intuitively show the stability, fluctuation and abnormal signals of the distribution box voltage. Specifically, correctly connect the probe of the oscilloscope to the output end of the explosion-proof distribution box to ensure that the connection is safe and stable, set the sampling rate, trigger conditions and other parameters of the oscilloscope to accurately capture and record the details of the voltage waveform, and the voltage waveform data collected by the oscilloscope can be further analyzed and processed, for example, observe the amplitude, frequency, phase and other characteristics of the waveform, judge whether the voltage is stable, whether there are interference signals such as harmonics or noise, which helps to timely discover potential faults or problems of the explosion-proof distribution box.
[0029] Step S200, performing signal conversion on the voltage waveform, the signal conversion refers to converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform. The time domain signal of the voltage waveform (i.e., the waveform of the voltage changing with time) is converted into a frequency domain signal by Fourier transform, wherein Fourier transform is a mathematical method widely used in signal processing, image processing and other fields, which can convert a signal from the time domain to the frequency domain. In the analysis of the voltage waveform, Fourier transform can decompose the voltage waveform into a combination of sine waves or cosine waves of different frequencies, so that the various frequency components of the voltage waveform can be observed and analyzed in the frequency domain, so as to understand the spectrum characteristics of the voltage waveform. Specifically, through Fourier transform, a spectrum diagram of the voltage waveform is obtained, which shows the amplitude and phase information of each frequency component in the voltage waveform, so as to facilitate the analysis of the stability of the voltage waveform and the existence of interference signals. For example, it can be observed whether there are abnormal frequency components, or whether the amplitude of certain frequency components exceeds the normal range, so as to judge whether there is a fault in the explosion-proof distribution box.
[0030] Step S300, establish the correspondence between the time domain signal and the frequency domain signal. In the signal processing process, the intrinsic connection and conversion law between the time domain signal and the frequency domain signal are clarified and established. Specifically, the time domain signal describes the change of the signal over time, which directly reflects the fluctuation and change of the signal on the time axis; the frequency domain signal is obtained by converting the time domain signal through Fourier transform, which describes the distribution of the signal at different frequencies, that is, the spectrum characteristics of the signal; establishing the correspondence between the time domain signal and the frequency domain signal is to clarify the connection between the two signal representation methods, that is, the characteristics of the frequency domain signal can be inferred by observing and analyzing the time domain signal; conversely, the characteristics of the time domain signal can be inferred by observing and analyzing the frequency domain signal. For example, in the voltage waveform analysis of the explosion-proof distribution box, if a periodic fluctuation is observed in the time domain, then in the frequency domain, this periodic fluctuation will correspond to one or more discrete frequency components. By establishing the correspondence between the two, the characteristics of the voltage waveform can be more deeply understood, such as the harmonic components and noise levels contained therein.
[0031] Step S400, constructing a time domain signal processing channel and a frequency domain signal processing channel, wherein the time domain signal processing channel is used to perform anomaly identification and analysis on the time domain signal, and the frequency domain signal processing channel is used to perform anomaly identification and analysis on the frequency domain signal. The time domain signal processing channel and the frequency domain signal processing channel are constructed based on the neural network model in order to perform abnormality identification and analysis in the time domain and frequency domain respectively at different stages of signal processing. The two channels are responsible for processing and analyzing different forms of signal data to provide more comprehensive and accurate fault diagnosis information. Specifically, the time domain signal processing channel is mainly used to perform abnormality identification and analysis on the original time domain signal. The signal will be filtered, feature extracted, etc. to identify and locate abnormalities or fault characteristics in the time domain signal. For example, for the voltage waveform of the explosion-proof distribution box, time domain processing can detect abnormal conditions such as sudden changes in voltage and periodic fluctuations; the frequency domain signal processing channel performs abnormality identification and analysis on the frequency domain signal processed by frequency domain analysis methods such as Fourier transform. The signal will be decomposed into components of different frequencies, and these components will be further analyzed and identified. Frequency domain processing can discover fault characteristics hidden in complex signals, such as specific frequency components, frequency offsets or harmonic distortion. By constructing time domain signal processing channels and frequency domain signal processing channels and comprehensively utilizing the information in the time domain and frequency domain, a more comprehensive and in-depth analysis of the operating status of the explosion-proof distribution box is carried out. The signal change information in the time dimension provided by the time domain channel and the distribution and characteristics of the signal in the frequency dimension provided by the frequency domain channel complement each other, jointly improving the accuracy and reliability of fault diagnosis.
[0032] In a possible implementation, step S400 further includes step S410, configuring a signal processing channel composition, including a signal screening module, a signal processing module, and an abnormality recognition model, wherein the signal screening module is used to screen the voltage waveform signal according to a preset target, the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency feature analysis and processing on the time domain signal, the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectral density analysis and processing on the frequency domain signal, the abnormality recognition model is used to perform abnormality recognition analysis on the signal imported by the signal processing module, and the abnormality recognition module is an operational model that has been converged through supervised training with a training data set. The signal processing channel is composed of a signal screening module, a signal processing module and an abnormality recognition model. Specifically, the signal screening module is used to screen the voltage waveform signal according to the preset target. The preset target may include the amplitude range, frequency range or other specific characteristics of the signal. Through screening, the signals that do not meet the requirements are excluded to ensure the quality and effectiveness of the subsequent processed signals, and the screened signals are sent to the corresponding time domain signal processing module and frequency domain signal processing module; the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency characteristic analysis and processing on the time domain signal. Signal time positioning can determine the specific time point when the abnormal signal appears, and frequency characteristic analysis can To reveal key information such as periodic components and harmonic distortion in the signal; the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectrum density analysis on the frequency domain signal. The characteristic frequency analysis can identify the dominant frequency components in the signal, and the power spectrum density analysis can provide the energy distribution of the signal at different frequencies, further revealing the spectral characteristics of the signal; the anomaly recognition model is the core of the entire processing flow, responsible for performing anomaly recognition analysis on the signal imported by the signal processing module, and outputting the corresponding anomaly recognition results. The anomaly recognition model is an operational model that has been trained and converged through supervised training with a training data set, and can automatically learn and recognize anomalies in the signal.
[0033] In a possible implementation, step S410 further includes step S411, decomposing the time domain signal according to different time scales according to the time mapping relationship of the time domain signal to obtain multiple decomposition scale signals. The original time domain signal is decomposed at multiple scales to extract the characteristics of the signal at different time scales, so as to more comprehensively analyze the changes and anomalies of the signal. Specifically, the time mapping relationship refers to the law of signal change over time. The original signal is decomposed into multiple sub-signals with different frequencies and time scales, i.e., multiple decomposition scale signals, by using wavelet transform, empirical mode decomposition (EMD), etc. Each decomposition scale signal reflects the characteristics of the original signal at a certain time scale. For example, in the voltage waveform analysis of the explosion-proof distribution box, the decomposition signals of different time scales may reveal the fluctuation mode, periodic changes and possible abnormal events of the voltage in different time periods. It also includes step S412, identifying the frequency characteristics of the decomposition scale signal to obtain the characteristics of each scale signal. The frequency characteristics of each decomposed scale signal obtained by multi-scale decomposition are analyzed and extracted, such as main frequency, bandwidth, peak frequency, etc. In the voltage waveform analysis of the explosion-proof distribution box, the frequency characteristics may correspond to different equipment states. For example, some frequency components indicate the normal operation state of the equipment, while other frequency components reflect potential faults or abnormalities. It also includes step S413, based on the time mapping relationship between different time scales, the similarity of the signal characteristics of each scale is calculated to obtain feature similarity. The time mapping relationship is used to compare and analyze the similarity of signal characteristics at different time scales. Specifically, the signal characteristics of each scale may include frequency characteristics, energy characteristics, waveform characteristics, etc., and then the similarity or difference between the signal characteristics of different time scales is quantitatively calculated by Euclidean distance, cosine similarity, etc. to obtain feature similarity. It also includes step S414, according to the feature similarity, abnormal time positioning and frequency feature labeling are performed. Using the feature similarity results calculated previously, the possible anomalies in the signal are temporally located, and the relevant frequency features are marked. The anomaly time location is determined by analyzing the changes in feature similarity. When the feature similarity shows a significant decrease or abnormal fluctuation, it means that an abnormal change has occurred in the signal within the time period. By comparing the feature similarities of different time scales, the time point or time period when the anomaly occurred can be accurately located; frequency feature marking is to identify and mark the frequency components in the abnormal signal. By analyzing the signal frequency characteristics of the abnormal time period, the abnormal frequency components can be determined and marked.
[0034] In a possible implementation, step S413 further includes step S4131, aligning the signals of different time scales based on the time mapping relationship. According to the time correspondence of the signals at different time scales, the signals of each scale are aligned and unified in time to ensure that the time information matches during comparison and analysis. It also includes step S4132, based on the alignment relationship of the frequency features, calculating the distance between each pair of aligned frequency features, and accumulating the distances to construct a distance matrix. The alignment relationship of frequency features refers to finding the corresponding frequency components in signals of different time scales, ensuring one-to-one correspondence in the frequency domain, using metrics such as Euclidean distance and Manhattan distance to calculate the distance between each pair of aligned frequency features, and quantifying the degree of difference between two frequency features. After calculating the distance between all aligned frequency feature pairs, these distance values are accumulated through the DTW algorithm to construct a distance matrix. The matrix dimension is related to the number of frequency features, and each element represents the distance between the corresponding frequency feature pairs. The DTW (Dynamic Time Warping) algorithm is a dynamic programming algorithm that calculates the similarity of two time series, especially sequences of different lengths. The similarity between two time series is calculated by extending and shortening the time series. It also includes step S4133, based on the distance matrix, path search is performed according to the path constraint conditions, and the minimum distance path is obtained to generate the feature similarity, wherein the path constraint conditions include the search starting point and the search direction. The distance matrix can be regarded as a graph structure, where nodes represent frequency features and edges represent the distances between frequency features. The purpose of path search is to find a path from the search starting point to the end point in this graph structure so that the sum of the distances of all edges on this path is minimized. The search starting point and search direction are part of the path constraints, which define the starting position and direction of the search. Specifically, the Dijkstra algorithm can be used to traverse the nodes and edges in the graph structure, and gradually find the optimal path according to the distance matrix and constraints, that is, consider all nodes and edges on the path, and calculate and compare the total distances of different paths according to the distances and constraints. Finally, a path with the minimum distance is found, that is, the nodes and edges passed from the search starting point to the end point so that the sum of the distances of all edges on the path is minimized. This minimum distance path represents the optimal similarity relationship between frequency features, that is, feature similarity. Among them, the Dijkstra algorithm is an algorithm for calculating the shortest path in a weighted graph. The core is to realize path search by gradually determining the shortest path from the starting point to each vertex.
[0035] In a possible implementation, step S414 further includes step S4141, screening the frequency features whose feature similarity does not meet the threshold, and locating the different time scales. In the feature similarity that has been calculated, find the frequency features whose similarity value is lower than the preset threshold, and further determine the different time scales where these frequency features are located, wherein the feature similarity threshold is a pre-set critical value for distinguishing which frequency feature similarities are high enough (i.e., meet the requirements) and which are not high enough (i.e., do not meet the requirements). It also includes step S4142, performing minimum time scale positioning decomposition based on the different time scales, and determining the abnormal time location. After the frequency features of different time scales have been identified, by further decomposing these features, the minimum time scale is found, and the time when the abnormality occurs is determined on this scale. It also includes step S4143, marking the frequency features according to the abnormal time location. After determining the time location of the abnormality, the frequency features at these time points or within the time period are marked to clearly show the changes in the frequency features at the abnormal time points.
[0036] In a possible implementation, step S410 further includes step S415, establishing a fault characteristic frequency, using the fault characteristic frequency to traverse and compare the frequency distribution of the frequency domain signal, and marking the fault characteristic frequency. Establishing a frequency feature associated with a specific fault mode, as a fault characteristic frequency, may be a specific value, or a change pattern within a certain frequency range, using the fault characteristic frequency to traverse and compare the frequency distribution of the frequency domain signal, once the frequency components matching the fault characteristic frequency are found in the frequency domain signal, these frequencies are marked, and the characteristic frequencies related to the fault can be effectively identified from the complex frequency domain signal. It also includes step S416, according to the frequency distribution, abnormal peaks are identified, and abnormal peaks of characteristic frequencies are marked. From the given frequency distribution data, those peaks that significantly deviate from the normal range are identified, and these abnormal peaks are marked. It also includes step S417, performing power spectral density analysis on the frequency domain signal to obtain the distribution relationship between power and frequency. By performing power spectral density analysis on the frequency domain signal, the distribution relationship between power and frequency can be obtained, that is, the energy distribution of the signal at different frequencies can be understood.
[0037] Step S500, establish a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel, and based on the correspondence between the time domain signal and the frequency domain signal, fuse and output the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel to obtain a fault diagnosis result. After establishing the time domain and frequency domain signal processing channels, a fusion connection layer needs to be established to connect the two channels. The fusion connection layer is responsible for receiving the abnormal analysis results from the two channels and fusing them according to the correspondence between the time domain and the frequency domain. Specifically, the time domain signal and the frequency domain signal are essentially the embodiment of the same signal in different representation methods, so there is a one-to-one correspondence between them, ensuring that the abnormal analysis results of the time domain and the frequency domain can accurately correspond to the different characteristics of the same signal. In the time domain channel, the main focus is on the change of the signal over time, and possible temporal anomalies, such as mutations, delays, etc., are analyzed as the temporal anomaly analysis results; in the frequency domain channel, through methods such as spectrum analysis, attention is paid to the distribution of the signal at different frequencies, and possible frequency anomalies are identified, such as the absence or enhancement of specific frequency components, as the frequency anomaly analysis results; the temporal anomaly analysis results and the frequency anomaly analysis results are fused and output to obtain the fault diagnosis information that integrates the time domain and frequency domain anomaly analysis results, that is, the fault diagnosis result, which more comprehensively reflects the characteristics of the signal, thereby improving the accuracy and reliability of fault diagnosis.
[0038] In a possible implementation, step S500 further includes step S510, establishing a correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result based on the correspondence between the time domain signal and the frequency domain signal. According to the correspondence between the time domain signal and the frequency domain signal, the temporal anomaly analysis result observed in the time domain is matched and corresponded with the corresponding frequency anomaly analysis result in the frequency domain. It also includes step S520, respectively obtaining the fault type and fault target according to the temporal anomaly analysis result and the frequency anomaly analysis result. According to the temporal anomaly analysis result and the frequency anomaly analysis result, by comparing the known fault mode or fault database, the possible fault type (for example, mechanical looseness, imbalance, misalignment) and fault target (for example, specific components such as bearings, gears, motors, or a module or unit) of the explosion-proof distribution box are obtained. It also includes step S530, judging whether the fault type and fault target meet the correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result. Verify whether the determined fault type and fault target are consistent with the temporal anomaly analysis results and frequency anomaly analysis results obtained by analysis. Specifically, check and verify whether the determined fault type and fault target match the anomaly analysis results of the time domain signal and the frequency domain signal. For example, if the temporal anomaly analysis results show that there is a mutation in the signal, then the determined fault type is a fault mode that can cause a mutation in the signal, such as a sudden breakage of a component or a sensor failure. It also includes step S540. When satisfied, the fault type and fault target are used as the fault diagnosis result. If the fault type and fault target meet the anomaly analysis results, that is, the diagnosis results of the two dimensions are consistent, the fault type and fault target are output as the fault diagnosis result. It also includes step S550. When not satisfied, the topological structure of the fault target is obtained, the voltage influence relationship of the fault target is determined, and the fusion coefficient is configured. When the fault type and fault target do not satisfy the correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result, the topological structure of the fault target is obtained, the voltage influence relationship of the fault target is determined, and the fusion coefficient is configured. Specifically, the position of the fault target in the network or system and other components connected to it are determined, and the influence of the fault target on the system voltage distribution is analyzed. By configuring the fusion coefficient, the information from different sources is effectively integrated to improve the accuracy and reliability of fault diagnosis. The fusion coefficient is pre-set according to various factors (such as the credibility of the data source, the accuracy of the analysis method, the characteristics of the fault target, etc.) to reflect the relative importance of different analysis results in comprehensive diagnosis. It also includes step S560, which fuses the temporal anomaly analysis result and the frequency anomaly analysis result based on the fusion coefficient to obtain the fault diagnosis result. The temporal anomaly analysis result and the frequency anomaly analysis result are weighted according to their respective fusion coefficients, so as to obtain a fault diagnosis result that integrates the temporal anomaly analysis result and the frequency anomaly analysis result, and outputs it as the fault diagnosis result.
[0039] In a possible implementation, step S560 further includes step S561, performing correlation analysis based on the fault types of the temporal anomaly analysis results and the frequency anomaly analysis results, determining the cause fault type and the result fault type, wherein the cause fault type refers to the fault type that is the cause of the result fault type. Among them, correlation refers to the cause-effect relationship of the fault type, comparing the fault types identified in the temporal anomaly analysis results and the frequency anomaly analysis results, analyzing the causal chain relationship between the fault types, determining which fault types may be the cause of other fault types, and further determining the cause fault type and the result fault type. Specifically, the cause fault type is usually a fault that occurs before the result fault type, which may be caused by factors such as damage to equipment components, improper operation or changes in external conditions, while the result fault type is the direct or indirect consequence of the cause fault. It also includes step S562, configuring a priority coefficient based on the correlation, wherein the priority coefficient of the result fault type is higher than the priority coefficient of the cause fault type. According to the causal relationship between fault types, different priorities are assigned to different types of faults, and the priority coefficient of the result fault type is set to be higher than the priority coefficient of the cause fault type. As a result, the fault more directly affects the operation, performance or safety of the equipment, and has a more significant negative impact on the stable operation of the system and the user experience. It also includes step S563, searching for fault targets and fault types according to the fusion coefficient and priority coefficient, and determining the abnormal analysis result with the highest evaluation result as the fault diagnosis result. Combined with the acquired fusion coefficient and priority coefficient, search and evaluate potential fault targets and fault types, and finally determine the most likely fault diagnosis result. Specifically, the temporal abnormal analysis results and frequency abnormal analysis results are weightedly fused by the fusion coefficient, and the potential fault targets and fault types are searched based on the priority coefficient. Then, a comprehensive evaluation is performed in combination with the influence relationship of the fault type, the component circuit, etc., including analyzing the relationship between the fault characteristics and the system behavior, comparing the possibility of different fault types, etc., and taking the abnormal analysis result with the highest evaluation result as the fault diagnosis result, that is, the fault diagnosis result with the highest accuracy.
[0040] In the above, refer to Figure 1 The present invention describes in detail an intelligent fault diagnosis method for an explosion-proof distribution box according to an embodiment of the present invention. Figure 2 An intelligent fault diagnosis system for an explosion-proof distribution box according to an embodiment of the present invention is described.
[0041] According to an embodiment of the present invention, an intelligent fault diagnosis system for an explosion-proof distribution box is used to solve the technical problem that the existing explosion-proof distribution box fault diagnosis relies on manual inspection and experience judgment, is easily affected, cannot achieve real-time monitoring and accurate early warning of faults, and leads to low accuracy and efficiency of explosion-proof distribution box fault diagnosis, thereby achieving the technical effect of improving the accuracy and efficiency of distribution box fault diagnosis. An intelligent fault diagnosis system for an explosion-proof distribution box includes: a voltage waveform acquisition module 10, a waveform signal conversion module 20, a corresponding relationship establishment module 30, a signal processing channel construction module 40, and an analysis result fusion output module 50.
[0042] A voltage waveform acquisition module 10, which is used to acquire the voltage waveform of the explosion-proof distribution box, and the voltage waveform is collected by an oscilloscope;
[0043] A waveform signal conversion module 20, the waveform signal conversion module 20 is used to perform signal conversion on the voltage waveform, the signal conversion refers to converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform;
[0044] A corresponding relationship establishing module 30, wherein the corresponding relationship establishing module 30 is used to establish a corresponding relationship between the time domain signal and the frequency domain signal;
[0045] A signal processing channel construction module 40, wherein the signal processing channel construction module 40 is used to construct a time domain signal processing channel and a frequency domain signal processing channel, wherein the time domain signal processing channel is used to perform anomaly identification and analysis on the time domain signal, and the frequency domain signal processing channel is used to perform anomaly identification and analysis on the frequency domain signal;
[0046] The analysis result fusion output module 50 is used to establish a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel, and based on the correspondence between the time domain signal and the frequency domain signal, the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel are fused and output to obtain a fault diagnosis result.
[0047] The specific configuration of the signal processing channel construction module 40 will be described in detail below. The signal processing channel construction module 40 further includes: configuring the signal processing channel composition, including a signal screening module, a signal processing module, and an abnormality recognition model, wherein the signal screening module is used to screen the voltage waveform signal according to a preset target, the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency feature analysis and processing on the time domain signal, the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectrum density analysis and processing on the frequency domain signal, the abnormality recognition model is used to perform abnormality recognition analysis on the signal imported by the signal processing module, and the abnormality recognition module is an operation model that has been trained and converged with a supervised training data set.
[0048] The specific configuration of the signal processing channel construction module 40 will be described in detail below. The signal processing channel construction module 40 may further include: decomposing the time domain signal according to different time scales according to the time mapping relationship of the time domain signal to obtain multiple decomposition scale signals; identifying the frequency characteristics of the decomposition scale signals to obtain the characteristics of each scale signal; calculating the similarity of the characteristics of each scale signal based on the time mapping relationship between different time scales to obtain the characteristic similarity; and locating the abnormal time and marking the frequency characteristics according to the characteristic similarity.
[0049] The specific configuration of the signal processing channel construction module 40 will be described in detail below. The signal processing channel construction module 40 may further include: aligning the signals of different time scales based on the time mapping relationship; calculating the distance between each pair of aligned frequency features based on the alignment relationship of the frequency features, and accumulating the distances to construct a distance matrix; based on the distance matrix, performing a path search according to the path constraint conditions, obtaining a path with the minimum distance to generate the feature similarity, wherein the path constraint conditions include a search starting point and a search direction.
[0050] The specific configuration of the signal processing channel construction module 40 will be described in detail below. The signal processing channel construction module 40 further includes: screening the frequency features whose feature similarity does not meet the threshold, locating the different time scales; performing minimum time scale location decomposition based on the different time scales, determining the abnormal time location; and marking the frequency features according to the abnormal time location.
[0051] The specific configuration of the signal processing channel construction module 40 will be described in detail below. The signal processing channel construction module 40 further includes: establishing a fault characteristic frequency, using the fault characteristic frequency to traverse and compare the frequency distribution of the frequency domain signal, and marking the fault characteristic frequency; according to the frequency distribution, performing abnormal peak identification, marking the abnormal peak of the characteristic frequency; performing power spectrum density analysis on the frequency domain signal to obtain the distribution relationship between power and frequency.
[0052] The specific configuration of the analysis result fusion output module 50 will be described in detail below. The analysis result fusion output module 50 may further include: establishing a correspondence between the temporal abnormality analysis result and the frequency abnormality analysis result based on the correspondence between the time domain signal and the frequency domain signal; obtaining the fault type and the fault target respectively according to the temporal abnormality analysis result and the frequency abnormality analysis result; judging whether the fault type and the fault target satisfy the correspondence between the temporal abnormality analysis result and the frequency abnormality analysis result; when satisfied, taking the fault type and the fault target as the fault diagnosis result; when not satisfied, obtaining the topological structure of the fault target, determining the voltage influence relationship of the fault target, and configuring the fusion coefficient; fusing the temporal abnormality analysis result and the frequency abnormality analysis result based on the fusion coefficient to obtain the fault diagnosis result.
[0053] The specific configuration of the analysis result fusion output module 50 will be described in detail below. The analysis result fusion output module 50 further includes: performing correlation analysis according to the fault types of the temporal abnormality analysis results and the frequency abnormality analysis results, determining the cause fault type and the result fault type, wherein the cause fault type refers to the fault type that is the cause of the result fault type; configuring a priority coefficient based on the correlation, wherein the priority coefficient of the result fault type is higher than the priority coefficient of the cause fault type; searching for fault targets and fault types according to the fusion coefficient and the priority coefficient, and determining the abnormality analysis result with the highest evaluation result as the fault diagnosis result.
[0054] An intelligent fault diagnosis system for an explosion-proof distribution box provided in an embodiment of the present invention can execute an intelligent fault diagnosis method for an explosion-proof distribution box provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.
[0055] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.
[0056] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.
Claims
1. An intelligent fault diagnosis method for an explosion-proof distribution box, characterized in that: The method comprises: Obtaining a voltage waveform of an explosion-proof distribution box, wherein the voltage waveform is collected by an oscilloscope; Performing signal conversion on the voltage waveform, wherein the signal conversion refers to converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform; Establishing a corresponding relationship between the time domain signal and the frequency domain signal; Constructing a time domain signal processing channel and a frequency domain signal processing channel, wherein the time domain signal processing channel is used to perform anomaly identification and analysis on the time domain signal, and the frequency domain signal processing channel is used to perform anomaly identification and analysis on the frequency domain signal; Establishing a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel, based on the corresponding relationship between the time domain signal and the frequency domain signal, fusing and outputting the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel to obtain a fault diagnosis result, wherein the time domain signal processing channel and the frequency domain signal processing channel are connected through the fusion connection layer, and are responsible for receiving the anomaly analysis results from the two channels; The constructing of the time domain signal processing channel and the frequency domain signal processing channel comprises: The signal processing channel is configured to include a signal screening module, a signal processing module, and an abnormality recognition model, wherein the signal screening module is used to screen the voltage waveform signal according to a preset target, the signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency feature analysis and processing on the time domain signal, the signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectrum density analysis and processing on the frequency domain signal, the abnormality recognition model is used to perform abnormality recognition and analysis on the signal imported by the signal processing module, and the abnormality recognition module is an operation model that has been trained and converged with a supervised training data set; Based on the correspondence between the time domain signal and the frequency domain signal, the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel are fused and output to obtain a fault diagnosis result, including: Based on the correspondence between the time domain signal and the frequency domain signal, a correspondence between the temporal anomaly analysis result and the frequency anomaly analysis result is established; According to the temporal anomaly analysis result and the frequency anomaly analysis result, respectively obtain the fault type and the fault target; Determine whether the fault type and fault target satisfy the corresponding relationship between the temporal abnormality analysis result and the frequency abnormality analysis result; When the conditions are met, the fault type and the fault target are used as the fault diagnosis result; When the condition is not satisfied, the topological structure of the fault target is obtained, the voltage influence relationship of the fault target is determined, and the fusion coefficient is configured; The temporal anomaly analysis result and the frequency anomaly analysis result are fused based on the fusion coefficient to obtain the fault diagnosis result.
2. The intelligent fault diagnosis method for an explosion-proof distribution box according to claim 1, characterized in that: The signal processing module of the time domain signal processing channel is used to perform signal time positioning and frequency characteristic analysis processing on the time domain signal, including: Decomposing the time domain signal according to different time scales according to the time mapping relationship of the time domain signal to obtain multiple decomposition scale signals; Identify the frequency characteristics of the decomposed scale signal to obtain the characteristics of each scale signal; Based on the time mapping relationship between different time scales, similarity calculation is performed on the signal features of each scale to obtain feature similarity; The abnormal time location and frequency feature annotation are performed according to the feature similarity.
3. The intelligent fault diagnosis method for an explosion-proof distribution box according to claim 2, characterized in that: The similarity calculation of the signal features of each scale based on the time mapping relationship between different time scales to obtain feature similarity includes: Based on the time mapping relationship, aligning the signals of different time scales; Based on the alignment relationship of frequency features, the distance between each pair of aligned frequency features is calculated, and the distances are accumulated to construct a distance matrix; Based on the distance matrix, a path search is performed according to path constraints to obtain a path with a minimum distance to generate the feature similarity, wherein the path constraints include a search starting point and a search direction.
4. The intelligent fault diagnosis method for an explosion-proof distribution box according to claim 3, characterized in that: The abnormal time location and frequency feature labeling are performed according to the feature similarity, including: Filter the frequency features whose feature similarity does not meet the threshold value, and locate the different time scales; Perform minimum time scale positioning decomposition based on the different time scales to determine the abnormal time positioning; Frequency feature annotation is performed according to the abnormal time location.
5. The intelligent fault diagnosis method for an explosion-proof distribution box according to claim 1, characterized in that: The signal processing module of the frequency domain signal processing channel is used to perform characteristic frequency and power spectrum density analysis on the frequency domain signal, including: Establishing a fault characteristic frequency, using the fault characteristic frequency to traverse and compare the frequency distribution of the frequency domain signal, and marking the fault characteristic frequency; According to the frequency distribution, abnormal peaks are identified and abnormal peaks of characteristic frequencies are marked; Perform power spectrum density analysis on the frequency domain signal to obtain the distribution relationship between power and frequency.
6. The intelligent fault diagnosis method for an explosion-proof distribution box according to claim 1, characterized in that: The temporal anomaly analysis result and the frequency anomaly analysis result are fused based on the fusion coefficient to obtain the fault diagnosis result, including: Perform correlation analysis on the fault types of the time anomaly analysis results and the frequency anomaly analysis results to determine the cause fault type and the result fault type, wherein the cause fault type refers to the fault type that is the cause of the result fault type; Configure a priority coefficient based on the correlation, wherein the priority coefficient of the result fault type is higher than the priority coefficient of the cause fault type; The fault target and fault type are searched according to the fusion coefficient and the priority coefficient, and the abnormal analysis result with the highest evaluation result is determined as the fault diagnosis result.
7. An intelligent fault diagnosis system for explosion-proof distribution boxes, characterized in that: The system is used to implement the intelligent fault diagnosis method for an explosion-proof distribution box according to any one of claims 1 to 6, and the system comprises: A voltage waveform acquisition module, which is used to acquire the voltage waveform of the explosion-proof distribution box, and the voltage waveform is collected by an oscilloscope; A waveform signal conversion module, the waveform signal conversion module is used to perform signal conversion on the voltage waveform, the signal conversion refers to converting the time domain signal of the voltage waveform into a frequency domain signal through Fourier transform; A corresponding relationship establishing module, the corresponding relationship establishing module is used to establish a corresponding relationship between the time domain signal and the frequency domain signal; A signal processing channel construction module, wherein the signal processing channel construction module is used to construct a time domain signal processing channel and a frequency domain signal processing channel, wherein the time domain signal processing channel is used to perform anomaly identification and analysis on the time domain signal, and the frequency domain signal processing channel is used to perform anomaly identification and analysis on the frequency domain signal; An analysis result fusion output module is used to establish a fusion connection layer of the time domain signal processing channel and the frequency domain signal processing channel, and based on the corresponding relationship between the time domain signal and the frequency domain signal, the temporal anomaly analysis result of the time domain signal processing channel and the frequency anomaly analysis result of the frequency domain signal processing channel are fused and output to obtain a fault diagnosis result, wherein the time domain signal processing channel and the frequency domain signal processing channel are connected by the fusion connection layer, which is responsible for receiving the anomaly analysis results from the two channels.