Phase commutation failure type judgment method and device, equipment and storage medium
By extracting and weighting the fault voltage signal of the high-voltage DC transmission system, the fault type is accurately judged, and the electrical fluctuations and power outage accidents caused by phase commutation failure are solved, and the fault type is accurately identified and prevented.
Patent Information
- Application Number
- CN202411239085.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-12-24
AI Technical Summary
Phase commutation failure in high-voltage DC transmission systems is often caused by AC side faults, resulting in electrical fluctuations and DC lockout, which may cause large-scale power outages. It is difficult for the existing technology to effectively determine the type of fault.
By obtaining multiple fault voltage signals at the fault point, determining the trajectory circle, performing a series of feature extraction and weighting enhancement processing, extracting attention weights, and calculating weights to determine the fault type.
Accurately judge the type of fault, avoid re-commutation failure, prevent electrical fluctuations and DC locking, and avoid large-scale power outages.
Smart Images
Figure CN119178957A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power transmission system fault judgment, and in particular to a method, device, equipment and storage medium for judging the type of commutation failure fault. Background Art
[0002] HVDC transmission has significant advantages in power transmission and plays an important role in the cross-regional transmission of clean energy in my country. However, commutation failures are inevitable in high-voltage DC transmission systems (LCC-HVDC) based on grid-commutated converters, most of which are caused by AC side faults. As the interaction between AC and DC systems becomes increasingly close, commutation failures may cause fluctuations in electrical quantities such as commutation bus voltage and DC current, or even cause DC locking and power oscillation, resulting in large-scale power outages. Summary of the invention
[0003] Based on this, it is necessary to propose a method, device, equipment and storage medium for judging the type of commutation failure fault in order to address the above problems.
[0004] A method for determining a commutation failure fault type, the method comprising: Acquire multiple fault voltage signals of the fault point at different times; Determine a corresponding trajectory circle according to the multiple fault voltage signals; Performing a first convolution process on the trajectory circle to obtain a first feature map; Performing first-order feature extraction on the first feature map to obtain a first-order feature map; Performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; Performing third-order feature extraction on the second-order feature map to obtain a third-order feature map; Performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The attention weight of the enhanced feature map is extracted, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0005] In one embodiment, determining the corresponding trajectory circle according to the multiple fault voltage signals comprises: Convert each fault voltage signal into a corresponding complex function; determining an analytical signal and an instantaneous phase according to the fault voltage signal and the complex function; Determine an upper envelope sequence according to the analytical signal; The trajectory circle is determined according to the upper envelope sequence and the variation of the instantaneous phase.
[0006] In one embodiment, performing first-order feature extraction on the first feature map to obtain the first-order feature map includes: Normalizing the first feature map to obtain a second feature map; Performing ReLU activation processing on the second feature map to obtain a third feature map; Perform maximum pooling processing on the third feature map to obtain the first-order feature map.
[0007] In one embodiment, performing second-order feature extraction on the first-order feature map to obtain the second-order feature map includes: Performing a second convolution process on the first-order feature map to obtain a fifth feature map; Performing a third convolution process on the first-order feature map to obtain a sixth feature map; Performing a fourth convolution process on the sixth feature map to obtain a seventh feature map; The second-order feature map is obtained by fusing the fifth feature map and the seventh feature map.
[0008] In one embodiment, performing third-order feature extraction on the second-order feature map to obtain the third-order feature map includes: Performing feature compression on the second-order feature map to obtain compressed features; Performing an excitation operation on the compression feature to obtain a weight of the compression feature; Recalibrating the second-order feature map according to the second-order feature map and the weight to obtain a recalibration weight; A plurality of recalibration weights constitute the third-order feature map.
[0009] In one embodiment, performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map includes: The third-order feature map is sequentially subjected to convolution processing and average pooling processing to obtain the enhanced feature map.
[0010] In one embodiment, extracting the attention weight of the enhanced feature map includes: Obtaining a query vector of the enhanced feature graph, a key vector of the enhanced feature graph, and a value vector of the enhanced feature graph; The attention weight is determined according to the query vector, the key vector, and the value vector.
[0011] A commutation failure fault type judgment device, the device comprising: An acquisition module, used to acquire multiple fault voltage signals of the fault point at different times; A determination module, configured to determine a corresponding trajectory circle according to the plurality of fault voltage signals; A first processing module, used for performing a first convolution process on the trajectory circle to obtain a first feature map; A first extraction module, used for performing first-order feature extraction on the first feature map to obtain a first-order feature map; A second extraction module is used to perform second-order feature extraction on the first-order feature map to obtain a second-order feature map; A third extraction module is used to extract third-order features from the second-order feature map to obtain a third-order feature map; A second processing module is used to perform weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The fourth extraction module is used to extract the attention weight of the enhanced feature map, and perform weighted sum calculation on the value vector corresponding to the attention weight to obtain a weighted value, wherein the weighted value corresponds to the fault type.
[0012] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps: Acquire multiple fault voltage signals of the fault point at different times; Determine a corresponding trajectory circle according to the multiple fault voltage signals; Performing a first convolution process on the trajectory circle to obtain a first feature map; Performing first-order feature extraction on the first feature map to obtain a first-order feature map; Performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; Performing third-order feature extraction on the second-order feature map to obtain a third-order feature map; Performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The attention weight of the enhanced feature map is extracted, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0013] A computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following steps: Acquire multiple fault voltage signals of the fault point at different times; Determine a corresponding trajectory circle according to the multiple fault voltage signals; Performing a first convolution process on the trajectory circle to obtain a first feature map; Performing first-order feature extraction on the first feature map to obtain a first-order feature map; Performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; Performing third-order feature extraction on the second-order feature map to obtain a third-order feature map; Performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The attention weight of the enhanced feature map is extracted, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0014] The present application obtains multiple fault voltage signals at different times of the fault point; determines the corresponding trajectory circle according to the multiple fault voltage signals; performs a first convolution process on the trajectory circle to obtain a first feature map; performs a first-order feature extraction on the first feature map to obtain a first-order feature map; performs a second-order feature extraction on the first-order feature map to obtain a second-order feature map; performs a third-order feature extraction on the second-order feature map to obtain a third-order feature map; performs a weighted enhancement process on the third-order feature map to obtain an enhanced feature map; extracts the attention weight of the enhanced feature map, and performs a weighted and calculated calculation on the value vector corresponding to the attention weight to obtain a weighted value, and the weighted value corresponds to the fault type. A series of feature extraction operations are performed on multiple fault voltage signals to obtain an enhanced feature map, and then the weighted value is determined according to the enhanced feature map. The fault type can be determined based on the weighted value. Once the fault type is determined, subsequent preventive work can be carried out for that fault type, avoiding the occurrence of another commutation failure, and also avoiding fluctuations in electrical quantities such as the commutation bus voltage and DC current, which may cause DC locking and power oscillation and result in large-scale power outages. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0016] in: Figure 1 An application environment diagram of a method for determining the type of commutation failure fault in one embodiment; Figure 2 is a flow chart of a method for determining the type of commutation failure fault in one embodiment; Figure 3 is a schematic diagram of a trajectory circle in one embodiment; Figure 4 It is a structural block diagram of a commutation failure fault type judgment device in one embodiment; Figure 5 FIG. 4 is a structural block diagram of a computer device in one embodiment. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] High-voltage direct current transmission has significant advantages in electric energy transmission and plays an important role in the cross-regional transmission of clean energy in my country. However, commutation failures are inevitable in high-voltage direct current transmission systems (LCC-HVDC) based on grid-commutated converters, most of which are caused by faults on the AC side. As the interaction between AC and DC systems becomes increasingly close, commutation failures may cause fluctuations in electrical quantities such as the commutation bus voltage and DC current, or may even cause DC lockout and power oscillation, resulting in large-scale power outages. In order to solve the above technical problems, the present application provides a method for determining the type of commutation failure fault. Figure 1 FIG. 1 is a diagram of an application environment for determining the type of commutation failure in an embodiment. Figure 1 The commutation failure fault type judgment method is applied to a commutation failure fault type judgment system. The commutation failure fault type judgment system includes a terminal 110 and a server 120. The terminal 110 and the server 120 are connected via a network. The terminal 110 may be a desktop terminal or a mobile terminal. The mobile terminal may be at least one of a mobile phone, a tablet computer, a laptop computer, etc. The server 120 may be implemented as an independent server or a server cluster consisting of multiple servers. The terminal 110 is used to obtain multiple fault voltage signals of the fault point at different times, and the server 120 is used to determine the corresponding trajectory circle according to the multiple fault voltage signals; perform a first convolution process on the trajectory circle to obtain a first feature map; perform a first-order feature extraction on the first feature map to obtain a first-order feature map; perform a second-order feature extraction on the first-order feature map to obtain a second-order feature map; perform a third-order feature extraction on the second-order feature map to obtain a third-order feature map; perform a weighted enhancement process on the third-order feature map to obtain an enhanced feature map; extract the attention weight of the enhanced feature map, and perform a weighted and calculated operation on the value vector corresponding to the attention weight to obtain a weighted value, wherein different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0019] like Figure 2 As shown, in one embodiment, a method for determining the type of a commutation failure fault is provided. The method can be applied to both a terminal and a server. This embodiment is described by taking the application to a terminal as an example. The method for determining the type of a commutation failure fault specifically includes the following steps: S10: Acquire multiple fault voltage signals of the fault point at different times; S20: determining a corresponding trajectory circle according to the multiple fault voltage signals; S30: performing a first convolution process on the trajectory circle to obtain a first feature map; S40: performing first-order feature extraction on the first feature map to obtain a first-order feature map; S50: performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; S60: extracting third-order features from the second-order feature map to obtain a third-order feature map; S70: performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; S80: extracting the attention weight of the enhanced feature map, and performing weighted sum calculation on the value vector corresponding to the attention weight to obtain a weighted value, wherein different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0020] The present application obtains multiple fault voltage signals at different times of the fault point; determines the corresponding trajectory circle according to the multiple fault voltage signals; performs a first convolution process on the trajectory circle to obtain a first feature map; performs a first-order feature extraction on the first feature map to obtain a first-order feature map; performs a second-order feature extraction on the first-order feature map to obtain a second-order feature map; performs a third-order feature extraction on the second-order feature map to obtain a third-order feature map; performs a weighted enhancement process on the third-order feature map to obtain an enhanced feature map; extracts the attention weight of the enhanced feature map, and performs a weighted and calculated calculation on the value vector corresponding to the attention weight to obtain a weighted value, and the weighted value corresponds to the fault type. A series of feature extraction operations are performed on multiple fault voltage signals to obtain an enhanced feature map, and then the weighted value is determined according to the enhanced feature map. The fault type can be determined based on the weighted value. Once the fault type is determined, subsequent preventive work can be carried out for that fault type, avoiding the occurrence of another commutation failure, and also avoiding fluctuations in electrical quantities such as the commutation bus voltage and DC current, which may cause DC locking and power oscillation and result in large-scale power outages.
[0021] Specifically, in step S20, determining the corresponding trajectory circle according to the multiple fault voltage signals includes: S201: Convert each fault voltage signal into a corresponding complex function; S202: determining an analytical signal and an instantaneous phase according to the fault voltage signal and the complex function; S203: determining an upper envelope sequence according to the analytical signal; S204: Determine the trajectory circle according to the upper envelope sequence and the change amount of the instantaneous phase.
[0022] The specific expression for obtaining the trajectory circle is as follows: (1) (2) (3) (4) (5) In the formula, is the fault voltage signal, is the complex function corresponding to the fault voltage signal, for, for, for, To analyze the signal, is the imaginary part of the complex function, is the instantaneous phase, is the instantaneous amplitude, Upper envelope sequence, is the trajectory circle, is the trajectory circular polar diameter, is the polar angle. The instantaneous amplitude obtained by Hilbert can accurately express the variation law of the signal amplitude in the time-frequency domain, and can better reflect the local characteristics of the signal. The larger the instantaneous amplitude, the larger the oscillation amplitude. Taking the commutation failure caused by the three-phase short circuit on the inverter side as an example, the following is drawn: Figure 3 In the trajectory circle shown, the numbers on the outer circle represent the polar angle, the line from the center of the circle to the outer circle is the polar diameter, the oscillating part represents the time when the fault occurs, and the smooth part represents the time before and after the fault occurs.
[0023] Specifically, in step S40, performing first-order feature extraction on the first feature map to obtain the first-order feature map includes: S401: performing normalization processing on the first feature map to obtain a second feature map; S402: Perform ReLU activation processing on the second feature map to obtain a third feature map; S403: Perform maximum pooling processing on the third feature map to obtain the first-order feature map.
[0024] The specific expression for obtaining the first-order feature map is as follows: (6) (7) (8) (9) (10) (11) In the formula, is the first feature map, is the second feature map, For the nth convolution kernel, with the input channel Connected, the location is ( , ), is the trajectory circle, is the bias of the roll and, is the height of the first feature map, is the width of the first feature map, C For the channel, K is the size of the convolution kernel, For the The mean of the feature maps, H is the height of the trajectory, W is the width of the track, For the The variance of the feature map, represents the stability constant, is the scaling parameter, Table is the shift parameter, is the third characteristic graph, is the first-order feature map, is the size of the pooling window, i is the row in the matrix, j are the columns in the matrix, for i The maximum value of for j The maximum value of .
[0025] In step S50, performing second-order feature extraction on the first-order feature map to obtain the second-order feature map includes: S501: Perform a second convolution process on the first-order feature map to obtain a fifth feature map; S502: Performing a third convolution process on the first-order feature map to obtain a sixth feature map; S503: performing a fourth convolution process on the sixth feature map to obtain a seventh feature map; S504: Fusing the fifth feature map and the seventh feature map to obtain the second-order feature map.
[0026] Specifically, the specific formula for obtaining the second-order feature map is as follows: (12) (13) (14) (15) In the formula, is the first-order feature map, is the fifth characteristic graph, is the ReLU activation function; To combine the features of branch 1 and branch 2 (i.e., flatten operation), is the first convolutional block of branch 1 The weight matrix of the filters, is the first convolutional block of branch 1 The bias parameters of the filter, is the first convolutional block of branch 2 The weight matrix of the filters, is the first convolutional block of branch 2 The bias parameters of the filter, is the first convolution block of the second layer of branch 2 The weight matrix of the filters, is the first convolution block of the second layer of branch 2 The bias parameters of the filter, is the fifth characteristic graph, is the sixth characteristic graph, is the seventh characteristic graph, Second-order feature map.
[0027] In step S60, the third-order feature extraction is performed on the second-order feature map, and the obtained third-order feature map includes: S601: Perform feature compression on the second-order feature map to obtain compressed features; S602: performing an excitation operation on the compression feature to obtain a weight of the compression feature; S603: recalibrating the second-order feature map according to the second-order feature map and the weight to obtain a recalibration weight; S604: A plurality of recalibrated weights constitute the third-order feature map.
[0028] Specifically, the specific formula for obtaining the third-order characteristic graph is as follows: (16) (17) (18) (19) In the formula, is the second-order feature map, is the compression feature, is the weight, is the third-order feature map, is the fully connected layer parameter, is the activation function, is the recalibration weight, g is the parameterized gating mechanism, is the sigmoid activation function, is the weight parameter of the fully connected layer 1, is the weight parameter of the fully connected layer 2.
[0029] In step S70, performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map includes: performing convolution processing and average pooling processing on the third-order feature map in sequence to obtain the enhanced feature map.
[0030] Specifically, the specific expression for obtaining the enhanced feature map is as follows: (20) In the formula, y To enhance the feature map, is the third-order feature map, is the activation function, is the weight and b is the bias parameter.
[0031] In step S80, extracting the attention weight of the enhanced feature map includes: S801: Obtain a query vector of the enhanced feature graph, a key vector of the enhanced feature graph, and a value vector of the enhanced feature graph; S802: Determine the attention weight according to the query vector, the key vector and the value vector.
[0032] Specifically, the specific formula for obtaining the weighted value is as follows: (twenty one) (twenty two) In the formula, y To enhance the feature map, V (Value) is the value vector of the enhanced feature map, is the attention weight, Q (Query) is the query vector of the enhanced feature graph, K (Key) is the key vector of the enhanced feature map, is the weighted value, is the dimension of the key vector.
[0033] The present application obtains multiple fault voltage signals at different times of the fault point; determines the corresponding trajectory circle according to the multiple fault voltage signals; performs a first convolution process on the trajectory circle to obtain a first feature map; performs a first-order feature extraction on the first feature map to obtain a first-order feature map; performs a second-order feature extraction on the first-order feature map to obtain a second-order feature map; performs a third-order feature extraction on the second-order feature map to obtain a third-order feature map; performs a weighted enhancement process on the third-order feature map to obtain an enhanced feature map; extracts the attention weight of the enhanced feature map, and performs a weighted and calculated calculation on the value vector corresponding to the attention weight to obtain a weighted value, and the weighted value corresponds to the fault type. A series of feature extraction operations are performed on multiple fault voltage signals to obtain an enhanced feature map, and then the weighted value is determined according to the enhanced feature map. The fault type can be determined based on the weighted value. Once the fault type is determined, subsequent preventive work can be carried out for that fault type, avoiding the occurrence of another commutation failure, and also avoiding fluctuations in electrical quantities such as the commutation bus voltage and DC current, which may cause DC locking and power oscillation and result in large-scale power outages.
[0034] The present application also provides a commutation failure fault type judgment device, such as Figure 4 As shown, the device comprises: An acquisition module 100 is used to acquire multiple fault voltage signals at different times of the fault point; A determination module 200, configured to determine a corresponding trajectory circle according to the plurality of fault voltage signals; A first processing module 300, configured to perform a first convolution process on the trajectory circle to obtain a first feature map; A first extraction module 400, configured to perform first-order feature extraction on the first feature map to obtain a first-order feature map; A second extraction module 500 is used to extract second-order features from the first-order feature map to obtain a second-order feature map; A third extraction module 600 is used to extract third-order features from the second-order feature map to obtain a third-order feature map; The second processing module 700 is used to perform weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The fourth extraction module 800 is used to extract the attention weight of the enhanced feature map, and perform weighted sum calculation on the value vector corresponding to the attention weight to obtain a weighted value, wherein the weighted value corresponds to the fault type.
[0035] Figure 5 FIG. 1 shows an internal structure diagram of a computer device in an embodiment. The computer device may be a terminal or a server. Figure 5As shown, the computer device includes a processor, a memory and a network interface connected via a system bus. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor may implement a method for determining the type of a commutation failure. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor may implement a method for determining the type of a commutation failure. Those skilled in the art will appreciate that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0036] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps: S10: Acquire multiple fault voltage signals of the fault point at different times; S20: determining a corresponding trajectory circle according to the multiple fault voltage signals; S30: performing a first convolution process on the trajectory circle to obtain a first feature map; S40: performing first-order feature extraction on the first feature map to obtain a first-order feature map; S50: performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; S60: extracting third-order features from the second-order feature map to obtain a third-order feature map; S70: performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; S80: Extracting the enhanced feature map y The attention weight is obtained, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0037] In one embodiment, a computer-readable storage medium is provided, storing a computer program, wherein when the computer program is executed by a processor, the processor performs the following steps: S10: Acquire multiple fault voltage signals of the fault point at different times; S20: determining a corresponding trajectory circle according to the multiple fault voltage signals; S30: performing a first convolution process on the trajectory circle to obtain a first feature map; S40: performing first-order feature extraction on the first feature map to obtain a first-order feature map; S50: performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; S60: extracting third-order features from the second-order feature map to obtain a third-order feature map; S70: performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; S80: Extracting the enhanced feature map y The attention weight is obtained, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
[0038] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0039] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0040] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A method for determining the type of commutation failure fault, characterized in that: The method comprises: Acquire multiple fault voltage signals of the fault point at different times; Determine a corresponding trajectory circle according to the multiple fault voltage signals; Performing a first convolution process on the trajectory circle to obtain a first feature map; Performing first-order feature extraction on the first feature map to obtain a first-order feature map; Performing second-order feature extraction on the first-order feature map to obtain a second-order feature map; Performing third-order feature extraction on the second-order feature map to obtain a third-order feature map; Performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The attention weight of the enhanced feature map is extracted, and the value vector corresponding to the attention weight is weighted and calculated to obtain a weighted value, where different weighted values correspond to different fault types, and the fault type is determined by the weighted value.
2. The method for determining the type of commutation failure according to claim 1, characterized in that: Determining the corresponding trajectory circle according to the multiple fault voltage signals comprises: Convert each fault voltage signal into a corresponding complex function; determining an analytical signal and an instantaneous phase according to the fault voltage signal and the complex function; Determine an upper envelope sequence according to the analytical signal; The trajectory circle is determined according to the upper envelope sequence and the variation of the instantaneous phase.
3. The method for determining the type of commutation failure according to claim 1, characterized in that: The performing first-order feature extraction on the first feature map to obtain the first-order feature map comprises: Normalizing the first feature map to obtain a second feature map; Performing ReLU activation processing on the second feature map to obtain a third feature map; Perform maximum pooling processing on the third feature map to obtain the first-order feature map.
4. The method for determining the type of commutation failure according to claim 1, characterized in that: The extracting second-order features from the first-order feature map to obtain the second-order feature map comprises: Performing a second convolution process on the first-order feature map to obtain a fifth feature map; Performing a third convolution process on the first-order feature map to obtain a sixth feature map; Performing a fourth convolution process on the sixth feature map to obtain a seventh feature map; The second-order feature map is obtained by fusing the fifth feature map and the seventh feature map.
5. The method for determining the type of commutation failure according to claim 1, characterized in that: The third-order feature extraction is performed on the second-order feature map to obtain a third-order feature map including: Performing feature compression on the second-order feature map to obtain compressed features; Performing an excitation operation on the compression feature to obtain a weight of the compression feature; Recalibrating the second-order feature map according to the second-order feature map and the weight to obtain a recalibration weight; A plurality of recalibration weights constitute the third-order feature map.
6. The method for determining the type of commutation failure according to claim 1, characterized in that: The step of performing weighted enhancement processing on the third-order feature map to obtain an enhanced feature map comprises: The third-order feature map is sequentially subjected to convolution processing and average pooling processing to obtain the enhanced feature map.
7. The method for determining the type of commutation failure according to claim 1, characterized in that: The step of extracting the attention weight of the enhanced feature map comprises: Obtaining a query vector of the enhanced feature graph, a key vector of the enhanced feature graph, and a value vector of the enhanced feature graph; The attention weight is determined according to the query vector, the key vector, and the value vector.
8. A commutation failure fault type judgment device, characterized in that: The device comprises: An acquisition module, used to acquire multiple fault voltage signals of the fault point at different times; A determination module, configured to determine a corresponding trajectory circle according to the plurality of fault voltage signals; A first processing module, used for performing a first convolution process on the trajectory circle to obtain a first feature map; A first extraction module, used for performing first-order feature extraction on the first feature map to obtain a first-order feature map; A second extraction module is used to perform second-order feature extraction on the first-order feature map to obtain a second-order feature map; A third extraction module is used to extract third-order features from the second-order feature map to obtain a third-order feature map; A second processing module is used to perform weighted enhancement processing on the third-order feature map to obtain an enhanced feature map; The fourth extraction module is used to extract the attention weight of the enhanced feature map, and perform weighted sum calculation on the value vector corresponding to the attention weight to obtain a weighted value, wherein the weighted value corresponds to the fault type.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.