Fault positioning method and system based on deep learning and wavelet transform

Abstract

The invention discloses a fault positioning method and system based on deep learning and wavelet transform, and the method comprises the steps: carrying out the feature extraction of a to-be-detected signal, obtaining the feature information and fault features, obtaining a first preset wavelet function, a first preset decomposition layer number, a second preset wavelet function, and a second preset decomposition layer number through the feature information and fault features, and carrying out the detection of a fault. Obtaining a first deep learning model through the first preset wavelet function and the second preset wavelet function, obtaining a second deep learning model through the first preset decomposition layer number and the second preset decomposition layer number, and finally comparing the first deep learning model with the second deep learning model to obtain a fault positioning learning model. And solving the fault positioning learning model to obtain fault point information of the preset detection area. According to the method, the wavelet function and the number of decomposition layers can be considered at the same time, the positioning precision is effectively improved, unreasonable selection of the wavelet function and the number of decomposition layers can be effectively reduced, and the data processing time is effectively shortened.

Description

Technical Field

[0001] The present invention relates to a fault location method, and in particular to a fault location method and system based on deep learning and wavelet transform. Background Art

[0002] In low-voltage power supply lines and household appliances, fault arcs may be caused due to reasons such as wire insulation aging, damage or loose electrical connections. Moreover, when a series arc fault occurs, its fault current is less than the normal current, making it impossible for conventional line protection devices (such as residual current protection circuit breakers, fuses) to accurately judge the arc fault.

[0003] Currently, a fault arc detector is usually used to detect fault arcs. The fault arc detector determines the fault arc based on the time-frequency domain characteristics analysis of the current signal and the set threshold. However, the voltage and current during the operation of some interfering loads (such as dimmers, switch-mode power supply devices, motor-type load devices, startup and shutdown of devices, etc.) are very similar to those of series fault arcs. Coupled with factors such as environmental noise, load condition changes or network interference, it is easy to cause misoperation of the fault arc detection detector, resulting in a low accuracy of fault arc detection.

[0004] Wavelet transform is a time-frequency analysis method suitable for processing non-stationary signals and is widely used in fault diagnosis and location. Wavelet transform can effectively extract the instantaneous characteristics in the signal, thereby helping to identify and locate faults.

[0005] Wavelet transform will combine with a deep learning model for fault location. Through the training of the deep learning model and the characteristics obtained by wavelet transform, the location of the fault can be obtained more accurately. However, during the current process of fault location.

[0006] Although this method can perform fault location, it still has the following defects:

[0007] 1. The fault location model constructed only considering the wavelet function cannot fully meet the location conditions, resulting in a low location accuracy.

[0008] 2. Due to the inadequate processing of the wavelet function, a lot of signal processing time is often wasted, resulting in a reduced efficiency of fault location.

[0009] Disclosing the information of this background art section is only intended to increase the understanding of the overall background of the present application and should not be regarded as an admission or any form of implication that this information constitutes the prior art already known to those of ordinary skill in the art. Summary of the Invention

[0010] The object of the present invention is to overcome the shortcoming in the prior art that only considering wavelet functions results in low fault location accuracy, and to provide a fault location method and system based on deep learning and wavelet transform that only considering wavelet functions leads to low fault location accuracy.

[0011] To achieve the above object, the technical solution of the present invention is:

[0012] A fault location method based on deep learning and wavelet transform, the fault location method comprising:

[0013] S1. Obtain the signal to be detected;

[0014] S2. Extract features from the signal to be detected to obtain feature information and fault features;

[0015] S3. Obtain first processing information from the feature information and the fault features;

[0016] S4. Based on the first processing information, extract features from the signal to be detected to obtain the feature information to be processed;

[0017] S5. Based on the first processing information, obtain a first deep learning model and a second deep learning model;

[0018] S6. Based on the first deep learning model and the second deep learning model, construct a fault location learning model;

[0019] S7. Input the feature information to be processed into the fault location learning model for solution to obtain the fault point information of the preset detection area.

[0020] In the S2, the feature information includes the characteristics of the signal to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault features include the characteristics of the fault signal and the frequency characteristics of the fault signal of the fault signal;

[0021] In the S3, the first processing information includes a first preset wavelet function, a first preset decomposition level, a second preset wavelet function and a second preset decomposition level. The first preset wavelet function is obtained from the characteristics of the signal to be detected and the characteristics of the fault signal, the first preset decomposition level is obtained from the frequency characteristics to be detected and the frequency characteristics of the fault signal, the second preset wavelet function is obtained from the first preset wavelet function, and the second preset decomposition level is obtained from the first preset decomposition level.

[0022] In the S3, obtain the function type of the first preset wavelet function. When extracting features from the signal for the first time, randomly obtain a preset wavelet function corresponding to any function type as the nearest preset wavelet function. When the number of feature extraction times is greater than or equal to two, obtain the preset wavelet function for the previous feature extraction of the signal as the nearest preset wavelet function;

[0023] Obtain the function type of the first preset wavelet function. When the function type is one kind, use the preset wavelet function corresponding to this function type as the first preset wavelet function and the second preset wavelet function;

[0024] When the function type is multiple kinds and there is a preset wavelet function in the function types that is the same as the nearest preset wavelet function, use the nearest preset wavelet function as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function with a function type different from the first preset wavelet function as the second preset wavelet function;

[0025] When the function type is multiple kinds and there is no preset wavelet function in the function types that is the same as the nearest preset wavelet function, then randomly obtain a preset wavelet function corresponding to any function type as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function with a function type different from the first preset wavelet function as the second preset wavelet function;

[0026] The second preset decomposition layer number is a randomly selected preset decomposition layer number different from the first preset decomposition layer number among the alternative preset decomposition layer numbers.

[0027] In step S5, obtain a first deep learning model based on the first preset wavelet function and the second preset wavelet function, and obtain a second deep learning model based on the first preset decomposition layer number and the second preset decomposition layer number.

[0028] In step S6, when there is a same deep learning model between the first deep learning model and the second deep learning model, use the same deep learning model among them as the fault location learning model;

[0029] When the first deep learning model and the second deep learning model are both different, use the second deep learning model as the fault location learning model.

[0030] In step S7, after obtaining the fault location learning model, analyze the fault point information of the preset detection area obtained, obtain the actual fault point of the detection area, and obtain the fault point matching degree based on the actual fault point and the fault point information;

[0031] When the fault point matching degree does not meet the preset matching degree, obtain correction information based on the actual fault point and the fault point information, and correct the fault location learning model through the correction information until the fault point matching degree meets the preset matching degree.

[0032] The fault location method further includes S8, and S8 includes:

[0033] After obtaining the fault location learning model, historical fault signals and historical fault feature information corresponding to the historical fault signals are acquired. The historical fault feature information is analyzed through the fault location learning model to obtain historical fault location information. The historical actual location information corresponding to the historical fault signals is acquired, and historical fault location correction parameters are obtained based on the historical actual location information and the historical fault location information. The fault location learning model is corrected based on the fault location correction parameters.

[0034] A fault location system based on deep learning and wavelet transform, the system includes a signal acquisition module, an information extraction module, a feature processing module, a feature extraction module, a deep learning model construction module, a fault location learning model construction module, and a fault location module;

[0035] The signal acquisition module is used to acquire the signal to be detected;

[0036] The information extraction module is used to extract features from the signal to be detected to obtain feature information and fault features;

[0037] The feature processing module is used to obtain first processed information through the feature information and the fault features;

[0038] The feature extraction module is used to extract features from the signal to be detected based on the first processed information to obtain the feature information to be processed;

[0039] The deep learning model construction module is used to obtain a first deep learning model and a second deep learning model based on the first processed information;

[0040] The fault location learning model construction module is used to construct a fault location learning model based on the first deep learning model and the second deep learning model;

[0041] The fault location module is used to input the feature information to be processed into the fault location learning model for solution to obtain the fault point information of the preset detection area.

[0042] In the information extraction module, the feature information includes the characteristics of the signal to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault features include the characteristics of the fault signal and the frequency characteristics of the fault signal of the fault signal;

[0043] In the feature processing module, the first processed information includes a first preset wavelet function, a first preset decomposition level, a second preset wavelet function, and a second preset decomposition level. The first preset wavelet function is obtained through the characteristics of the signal to be detected and the characteristics of the fault signal, the first preset decomposition level is obtained through the frequency characteristics to be detected and the frequency characteristics of the fault signal, the second preset wavelet function is obtained through the first preset wavelet function, and the second preset decomposition level is obtained through the first preset decomposition level;

[0044] In the feature extraction module, obtain the function type of the first preset wavelet function. When performing feature extraction on the signal for the first time, randomly obtain a preset wavelet function corresponding to any function type as the most recent preset wavelet function. When the number of feature extraction times is greater than or equal to two, obtain the preset wavelet function used for the previous feature extraction of the signal as the most recent preset wavelet function; obtain the function type of the first preset wavelet function. When the function type is one, use the preset wavelet function corresponding to this function type as the first preset wavelet function and the second preset wavelet function; when the function type is multiple and there is a preset wavelet function in the function types that is the same as the most recent preset wavelet function, use the most recent preset wavelet function as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function of any function type different from the first preset wavelet function as the second preset wavelet function; when the function type is multiple and there is no preset wavelet function in the function types that is the same as the most recent preset wavelet function, then randomly obtain a preset wavelet function corresponding to any function type as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function of any function type different from the first preset wavelet function as the second preset wavelet function; the second preset decomposition layer number is a randomly selected preset decomposition layer number different from the first preset decomposition layer number among the alternative preset decomposition layer numbers;

[0045] In the deep learning model construction module, obtain the first deep learning model based on the first preset wavelet function and the second preset wavelet function, and obtain the second deep learning model based on the first preset decomposition layer number and the second preset decomposition layer number;

[0046] In the fault location learning model construction module, when there is the same deep learning model between the first deep learning model and the second deep learning model, use the same deep learning model among them as the fault location learning model;

[0047] When the first deep learning model and the second deep learning model are both different, use the second deep learning model as the fault location learning model;

[0048] In the fault location module, after obtaining the fault location learning model, analyze the fault point information of the preset detection area obtained, obtain the actual fault point of the detection area, and obtain the fault point matching degree based on the actual fault point and the fault point information;

[0049] When the fault point matching degree does not meet the preset matching degree, obtain correction information based on the actual fault point and the fault point information, and correct the fault location learning model through the correction information until the fault point matching degree meets the preset matching degree.

[0050] The fault location learning model construction module further includes:

[0051] After obtaining the fault location learning model, obtain historical fault signals and historical fault feature information corresponding to the historical fault signals, parse the historical fault feature information through the fault location learning model to obtain historical fault location information, obtain historical actual position information corresponding to the historical fault signals, obtain historical fault location correction parameters based on the historical actual position information and the historical fault location information, and correct the fault location learning model based on the fault location correction parameters.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] 1. In the fault location method based on deep learning and wavelet transform of the present invention, by extracting features from the signal to be detected, feature information and fault features are obtained. Based on the feature information and fault features, the first preset wavelet function and the first preset decomposition level are obtained. Then, the second preset wavelet function and the second preset decomposition level are obtained through the first preset wavelet function and the first preset decomposition level. The first deep learning model is obtained through the first preset wavelet function and the second preset wavelet function, and the second deep learning model is obtained through the first preset decomposition level and the second preset decomposition level. Finally, by comparing the first deep learning model and the second deep learning model, the fault location learning model is obtained, and the fault point information of the preset detection area is obtained by solving the fault location learning model, so that the fault location learning model can consider both the wavelet function and the decomposition level at the same time, avoiding the situation that the fault location model constructed only considering the wavelet function cannot fully meet the location conditions. Therefore, this design can consider both the wavelet function and the decomposition level at the same time, effectively improving the location accuracy.

[0054] 2. In the fault location method based on deep learning and wavelet transform of the present invention, by selecting different wavelet functions and different decomposition levels for different signals, the waste of data processing time caused by the unreasonable selection of wavelet functions and decomposition levels can be effectively reduced. Therefore, this design can effectively reduce the waste of data processing time caused by the unreasonable selection of wavelet functions and decomposition levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a flowchart of the method of the present invention.

[0056] Figure 2 is a structural diagram of the system of the present invention.

[0057] Figure 3 is a structural diagram of the device in Embodiment 3. DETAILED DESCRIPTION OF THE INVENTION

[0058] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0059] Example 1:

[0060] Refer to Figure 1 , a fault location method based on deep learning and wavelet transform, the fault location method includes:

[0061] S1. Obtain the signal to be detected;

[0062] S2. Extract features from the signal to be detected to obtain feature information and fault features;

[0063] S3. Obtain the first processing information through the feature information and fault features;

[0064] S4. Based on the first processing information, extract features from the signal to be detected to obtain the feature information to be processed;

[0065] S5. Obtain the first deep learning model and the second deep learning model based on the first processing information;

[0066] S6. Construct a fault location learning model based on the first deep learning model and the second deep learning model;

[0067] S7. Input the feature information to be processed into the fault location learning model for solution to obtain the fault point information in the preset detection area.

[0068] In S2, the feature information includes the characteristics of the signal to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault features include the characteristics of the fault signal and the frequency characteristics of the fault signal;

[0069] In S3, the first processing information includes a first preset wavelet function, a first preset decomposition level, a second preset wavelet function, and a second preset decomposition level. The first preset wavelet function is obtained through the characteristics of the signal to be detected and the characteristics of the fault signal. The first preset decomposition level is obtained through the frequency characteristics to be detected of the signal to be detected and the frequency characteristics of the fault signal. The second preset wavelet function is obtained through the first preset wavelet function, and the second preset decomposition level is obtained through the first preset decomposition level.

[0070] In S3, obtain the function type of the first preset wavelet function. When extracting features from the signal for the first time, randomly obtain the preset wavelet function corresponding to any function type as the nearest preset wavelet function. When the number of feature extraction times is greater than or equal to two, obtain the preset wavelet function for the previous feature extraction of the signal as the nearest preset wavelet function;

[0071] Obtain the function type of the first preset wavelet function. When the function type is one kind, use the preset wavelet function corresponding to this function type as the first preset wavelet function and the second preset wavelet function;

[0072] When the function type is multiple kinds and there is a preset wavelet function in the function types that is the same as the nearest preset wavelet function, use the nearest preset wavelet function as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function with a function type different from the first preset wavelet function as the second preset wavelet function;

[0073] When the function type is multiple kinds and there is no preset wavelet function in the function types that is the same as the nearest preset wavelet function, then randomly obtain a preset wavelet function corresponding to any function type as the first preset wavelet function, and at the same time randomly obtain a preset wavelet function with a function type different from the first preset wavelet function as the second preset wavelet function;

[0074] The second preset decomposition level is a randomly selected preset decomposition level different from the first preset decomposition level among the alternative preset decomposition levels.

[0075] In S5, obtain the first deep learning model based on the first preset wavelet function and the second preset wavelet function, and obtain the second deep learning model based on the first preset decomposition level and the second preset decomposition level.

[0076] In S6, when there is the same deep learning model between the first deep learning model and the second deep learning model, use the same deep learning model among them as the fault location learning model;

[0077] When the first deep learning model and the second deep learning model are both different, use the second deep learning model as the fault location learning model.

[0078] In S7, after obtaining the fault location learning model, analyze the fault point information of the obtained preset detection area to obtain the actual fault point of the detection area, and obtain the fault point matching degree based on the actual fault point and the fault point information;

[0079] When the fault point matching degree does not meet the preset matching degree, obtain the correction information based on the actual fault point and the fault point information, and correct the fault location learning model through the correction information until the fault point matching degree meets the preset matching degree.

[0080] The fault location method further includes S8, and S8 includes:

[0081] After obtaining the fault location learning model, historical fault signals and historical fault feature information corresponding to the historical fault signals are acquired. The historical fault feature information is analyzed through the fault location learning model to obtain historical fault location information. The historical actual position information corresponding to the historical fault signals is acquired, and historical fault location correction parameters are obtained based on the historical actual position information and the historical fault location information. The fault location learning model is corrected based on the fault location correction parameters.

[0082] Embodiment 2:

[0083] See Figure 2 , a fault location system based on deep learning and wavelet transform. The system includes a signal acquisition module, an information extraction module, a feature processing module, a feature extraction module, a deep learning model construction module, a fault location learning model construction module, and a fault location module;

[0084] The signal acquisition module is used to acquire the signal to be detected;

[0085] The information extraction module is used to extract features from the signal to be detected to obtain feature information and fault features;

[0086] The feature processing module is used to obtain first processed information through the feature information and the fault features;

[0087] The feature extraction module is used to extract features from the signal to be detected based on the first processed information to obtain the feature information to be processed;

[0088] The deep learning model construction module is used to obtain a first deep learning model and a second deep learning model based on the first processed information;

[0089] The fault location learning model construction module is used to construct a fault location learning model based on the first deep learning model and the second deep learning model;

[0090] The fault location module is used to input the feature information to be processed into the fault location learning model for solution to obtain the fault point information of the preset detection area.

[0091] In the information extraction module, the feature information includes the characteristics of the signal to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault features include the characteristics of the fault signal and the frequency characteristics of the fault signal;

[0092] In the feature processing module, the first processing information includes a first preset wavelet function, a first preset decomposition level, a second preset wavelet function, and a second preset decomposition level. The first preset wavelet function is obtained based on the characteristics of the signal to be detected and the characteristics of the fault signal. The first preset decomposition level is obtained based on the frequency characteristics of the signal to be detected and the frequency characteristics of the fault signal. The second preset wavelet function is obtained from the first preset wavelet function, and the second preset decomposition level is obtained from the first preset decomposition level.

[0093] In the feature extraction module, the function type of the first preset wavelet function is obtained. When the signal is feature-extracted for the first time, a preset wavelet function corresponding to any function type is randomly obtained as the most recent preset wavelet function. When the number of feature extraction times is greater than or equal to two, the preset wavelet function used for the previous feature extraction of the signal is obtained as the most recent preset wavelet function. The function type of the first preset wavelet function is obtained. When the function type is one, the preset wavelet function corresponding to this function type is used as the first preset wavelet function and the second preset wavelet function. When the function type is multiple and there is a preset wavelet function in the function types that is the same as the most recent preset wavelet function, the most recent preset wavelet function is used as the first preset wavelet function, and at the same time, a preset wavelet function of any function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function. When the function type is multiple and there is no preset wavelet function in the function types that is the same as the most recent preset wavelet function, then a preset wavelet function corresponding to any function type is randomly obtained as the first preset wavelet function, and at the same time, a preset wavelet function of any function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function. The second preset decomposition level is a random preset decomposition level different from the first preset decomposition level among the alternative preset decomposition levels.

[0094] In the deep learning model construction module, a first deep learning model is obtained based on the first preset wavelet function and the second preset wavelet function, and a second deep learning model is obtained based on the first preset decomposition level and the second preset decomposition level.

[0095] In the fault location learning model construction module, when there is the same deep learning model between the first deep learning model and the second deep learning model, the same deep learning model is used as the fault location learning model.

[0096] When the first deep learning model and the second deep learning model are both different, the second deep learning model is used as the fault location learning model.

[0097] In the fault location module, after obtaining the fault location learning model, analyze the fault point information of the preset detection area obtained, obtain the actual fault point of the detection area, and obtain the fault point matching degree based on the actual fault point and the fault point information;

[0098] When the fault point matching degree does not meet the preset matching degree, obtain correction information based on the actual fault point and the fault point information, and correct the fault location learning model through the correction information until the fault point matching degree meets the preset matching degree.

[0099] The fault location learning model construction module further includes:

[0100] After obtaining the fault location learning model, obtain historical fault signals and historical fault feature information corresponding to the historical fault signals, parse the historical fault feature information through the fault location learning model to obtain historical fault location information, obtain historical actual position information corresponding to the historical fault signals, obtain historical fault location correction parameters based on the historical actual position information and the historical fault location information, and correct the fault location learning model based on the fault location correction parameters.

[0101] Embodiment 3:

[0102] Refer to Figure 3 , a fault location device based on deep learning and wavelet transform, the device includes a processor and a memory;

[0103] The memory is used to store computer program code and transmit the computer program code to the processor;

[0104] The processor is used to execute the fault location method based on deep learning and wavelet transform described in Embodiment 1 according to the instructions in the computer program code.

[0105] A computer medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the fault location method based on deep learning and wavelet transform described in Embodiment 1.

[0106] The above is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. Any equivalent modification or change made by those of ordinary skill in the art according to the content disclosed by the present invention should be included in the protection scope recorded in the claims.

Claims

1. A fault location method based on deep learning and wavelet transform, characterized in that: The fault location method comprises: S1. Obtain the signal to be detected; S2. Extract features of the signal to be detected to obtain feature information and fault features; S3, obtaining first processing information through characteristic information and fault characteristics; S4. Extracting features of the signal to be detected based on the first processed information to obtain feature information to be processed; S5. Acquire a first deep learning model and a second deep learning model based on the first processing information; S6. Building a fault location learning model based on the first deep learning model and the second deep learning model; S7. Input the feature information to be processed into the fault location learning model for solution to obtain the fault point information of the preset detection area.

2. The fault location method based on deep learning and wavelet transform according to claim 1, characterized in that: In S2, the characteristic information includes the signal characteristics to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault characteristics include the fault signal characteristics and the fault frequency characteristics of the fault signal; In S3, the first processing information includes a first preset wavelet function, a first preset decomposition number, a second preset wavelet function and a second preset decomposition number, the first preset wavelet function is obtained through the characteristics of the signal to be detected and the fault signal characteristics, the first preset decomposition number is obtained through the frequency characteristics to be detected and the fault frequency characteristics, the second preset wavelet function is obtained through the first preset wavelet function, and the second preset decomposition number is obtained through the first preset decomposition number.

3. The fault location method based on deep learning and wavelet transform according to claim 2, characterized in that: In S3, the function type of the first preset wavelet function is obtained. When the signal is subjected to feature extraction for the first time, a preset wavelet function corresponding to any function type is randomly obtained as the most recent preset wavelet function. When the number of feature extractions is greater than or equal to two, the preset wavelet function used for the last feature extraction of the signal is obtained as the most recent preset wavelet function. When the function type is one, the preset wavelet function corresponding to the function type is used as the first preset wavelet function and the second preset wavelet function; When there are multiple types of functions, and there is a preset wavelet function that is the same as the most recent preset wavelet function in the function types, the most recent preset wavelet function is used as the first preset wavelet function, and at the same time, any preset wavelet function of a function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function; When there are multiple types of functions, and there is no preset wavelet function in the function types that is the same as the most recent preset wavelet function, a preset wavelet function corresponding to any function type is randomly obtained as the first preset wavelet function, and a preset wavelet function of any function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function; The second preset decomposition number is a random preset decomposition number among the alternative preset decomposition numbers and is different from the first preset decomposition number.

4. The fault location method based on deep learning and wavelet transform according to claim 3 is characterized in that: In S5, a first deep learning model is obtained based on the first preset wavelet function and the second preset wavelet function, and a second deep learning model is obtained based on the first preset number of decomposition layers and the second preset number of decomposition layers.

5. The fault location method based on deep learning and wavelet transform according to claim 4, characterized in that: In S6, when the first deep learning model and the second deep learning model have the same deep learning model, the same deep learning model is used as the fault location learning model; When the first deep learning model and the second deep learning model are different, the second deep learning model is used as the fault location learning model.

6. The fault location method based on deep learning and wavelet transform according to claim 5, characterized in that: In S7, after the fault location learning model is obtained, the fault point information of the preset detection area is analyzed to obtain the actual fault point of the detection area, and the fault point matching degree is obtained based on the actual fault point and the fault point information; When the fault point matching degree does not meet the preset matching degree, correction information is obtained based on the actual fault point and the fault point information, and the fault location learning model is corrected by the correction information until the fault point matching degree meets the preset matching degree.

7. The fault location method based on deep learning and wavelet transform according to claim 6, characterized in that: The fault location method further includes S8, wherein S8 includes: After the fault location learning model is obtained, the historical fault signal and the historical fault feature information corresponding to the historical fault signal are obtained, the historical fault feature information is parsed through the fault location learning model to obtain the historical fault location information, the historical actual position information corresponding to the historical fault signal is obtained, the historical fault location correction parameters are obtained based on the historical actual position information and the historical fault location information, and the fault location learning model is corrected based on the fault location correction parameters.

8. A fault location system based on deep learning and wavelet transform, characterized in that: The system includes a signal acquisition module, an information extraction module, a feature processing module, a feature extraction module, a deep learning model construction module, a fault location learning model construction module and a fault location module; The signal acquisition module is used to acquire the signal to be detected; The information extraction module is used to extract features from the signal to be detected to obtain feature information and fault features; The feature processing module is used to obtain first processing information through feature information and fault features; The feature extraction module is used to extract features of the signal to be detected based on the first processing information to obtain feature information to be processed; The deep learning model construction module is used to obtain a first deep learning model and a second deep learning model based on the first processing information; The fault location learning model construction module is used to construct a fault location learning model based on the first deep learning model and the second deep learning model; The fault location module is used to input the feature information to be processed into the fault location learning model for solution, and obtain the fault point information of the preset detection area; In the information extraction module, the characteristic information includes the signal characteristics to be detected and the frequency characteristics to be detected of the signal to be detected, and the fault characteristics include the fault signal characteristics and the fault frequency characteristics of the fault signal; In the feature processing module, the first processing information includes a first preset wavelet function, a first preset decomposition layer number, a second preset wavelet function and a second preset decomposition layer number, the first preset wavelet function is obtained by the characteristics of the signal to be detected and the fault signal characteristics, the first preset decomposition layer number is obtained by the frequency characteristics to be detected and the fault frequency characteristics, the second preset wavelet function is obtained by the first preset wavelet function, and the second preset decomposition layer number is obtained by the first preset decomposition layer number; In the feature extraction module, the function type of the first preset wavelet function is obtained. When the feature extraction is performed on the signal for the first time, the preset wavelet function corresponding to any function type is randomly obtained as the most recent preset wavelet function. When the number of feature extractions is greater than or equal to two, the preset wavelet function used for the last feature extraction of the signal is obtained as the most recent preset wavelet function. The function type of the first preset wavelet function is obtained. When the function type is one, the preset wavelet function corresponding to the function type is used as the first preset wavelet function and the second preset wavelet function. When there are multiple function types and there is a preset wavelet function in the function type that is the same as the most recent preset wavelet function, When the number of decomposition layers is 1, the most recent preset wavelet function is used as the first preset wavelet function, and any preset wavelet function of a function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function; when there are multiple function types, and there is no preset wavelet function that is the same as the most recent preset wavelet function in the function types, a preset wavelet function corresponding to any function type is randomly obtained as the first preset wavelet function, and any preset wavelet function of a function type different from the first preset wavelet function is randomly obtained as the second preset wavelet function; the second preset decomposition number of layers is a random preset decomposition number of layers in the candidate preset decomposition numbers that is different from the first preset decomposition number of layers; In the deep learning model construction module, a first deep learning model is obtained based on the first preset wavelet function and the second preset wavelet function, and a second deep learning model is obtained based on the first preset number of decomposition layers and the second preset number of decomposition layers; In the fault location learning model construction module, when the first deep learning model and the second deep learning model have the same deep learning model, the same deep learning model is used as the fault location learning model; When the first deep learning model and the second deep learning model are different, using the second deep learning model as a fault location learning model; In the fault location module, after the fault location learning model is obtained, the fault point information of the preset detection area is analyzed to obtain the actual fault point of the detection area, and the fault point matching degree is obtained based on the actual fault point and the fault point information; When the fault point matching degree does not meet the preset matching degree, obtaining correction information based on the actual fault point and the fault point information, and correcting the fault location learning model by using the correction information until the fault point matching degree meets the preset matching degree; The fault location learning model building module also includes: After the fault location learning model is obtained, the historical fault signal and the historical fault feature information corresponding to the historical fault signal are obtained, the historical fault feature information is parsed through the fault location learning model to obtain the historical fault location information, the historical actual position information corresponding to the historical fault signal is obtained, the historical fault location correction parameters are obtained based on the historical actual position information and the historical fault location information, and the fault location learning model is corrected based on the fault location correction parameters.

9. A fault location device based on deep learning and wavelet transform, characterized in that: The device comprises a processor and a memory; The memory is used to store computer program code and transmit the computer program code to the processor; The processor is used to execute the fault location method based on deep learning and wavelet transform according to claims 1 to 7 according to the instructions in the computer program code.

10. A computer medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the fault location method based on deep learning and wavelet transform described in claims 1 to 7 is implemented.

Images