CLCC system fault identification method and system based on LDA and SVM
By applying the fault identification methods of LDA and SVM in the CLCC system, the problem of difficult to identify the fault of the converter valve is solved, and the accurate identification of fault types and improvement of system stability is achieved.
Patent Information
- Application Number
- CN202411381413.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2025-02-11
AI Technical Summary
The converter valve in the CLCC system is prone to failure, making it difficult to accurately determine the type and cause of the fault, which affects the stability of the system.
The fault identification method based on LDA and SVM is adopted, and fault data is obtained, fault data is preprocessed and feature extraction is performed, and the LDA model is used for dimensional compression and data screening, input fault characteristics are generated, and fault types are automatically identified using the SVM classifier.
It realizes accurate identification of fault types of CLCC system, improves the efficiency and accuracy of fault identification, and improves the stability and safety of the system.
Smart Images
Figure CN119416041A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of direct current transmission, and in particular to a CLCC system fault identification method and system based on LDA and SVM. Background Art
[0002] The traditional thyristor-based grid-commutated converter HVDC transmission system has significant advantages such as long-distance large-capacity transmission and fast controllable active power. However, since the thyristor is a semi-controlled device, it needs to rely on the AC system to provide sufficient reverse voltage to restore its blocking capability. When the AC system fails, the LCC converter is prone to commutation failure, resulting in a surge in DC current in a short period of time, thereby significantly reducing the DC power output. In order to solve the problem of commutation failure, DC transmission technology containing controlled commutated converters (CLCC) has been used in engineering practice. CLCC not only has a controllable shutdown capability to avoid commutation failure, but also has the advantages of large capacity and low cost of LCC, which significantly improves the stability of the power grid.
[0003] CLCC is a controllable grid-commutated converter system based on a hybrid of fully controlled IGBTs and semi-controlled thyristors. It has more internal power electronic devices than LCC, and its internal structure and control logic are more complex. It is more likely to have short circuits or grounding faults due to insulation problems in the converter valve. Therefore, when a fault occurs in the converter valve, it is difficult to determine the type and cause of the fault, which hinders the stability of CLCC use. Summary of the invention
[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a CLCC system fault identification method and system based on LDA and SVM which can accurately identify the fault type and has high efficiency.
[0005] The purpose of the present invention can be achieved by the following technical solutions:
[0006] In a first aspect, the present invention provides a CLCC system fault identification method based on LDA and SVM, comprising the following steps:
[0007] Acquire fault data when a fault occurs in a DC converter valve of the CLCC system, pre-process the fault data, and extract fault waveform feature data;
[0008] The LDA model is used to perform dimension compression and data screening on the fault waveform feature data to generate input fault features;
[0009] The input fault feature is used as the input of the trained SVM classifier to automatically identify the fault type.
[0010] Furthermore, the fault data includes overvoltage data or overcurrent data.
[0011] Furthermore, the preprocessing includes waveform standard normalization processing.
[0012] Furthermore, the fault waveform characteristic data includes pulse peak value, frequency component and / or pulse rise time.
[0013] Furthermore, the LDA model performs dimension compression and data screening by calculating the intra-class scatter matrix, the inter-class scatter matrix and the projection matrix.
[0014] Furthermore, the projection matrix satisfies the requirement of maximizing the divergence between fault data sets and minimizing the divergence within a class.
[0015] Furthermore, the fault types include AB phase short circuit, VT1 valve short circuit, VT1 valve short circuit grounding fault, 6-pulse anode grounding fault, 6-pulse cathode grounding fault, 6-pulse short circuit fault and normal situation.
[0016] Furthermore, the method further comprises:
[0017] When the fault type is normal, the corresponding fault data is expanded into the training sample to achieve data sample expansion and online adjustment and optimization of the model.
[0018] In a second aspect, the present invention provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the CLCC system fault identification method based on LDA and SVM as described above.
[0019] In a third aspect, the present invention provides a CLCC system fault identification system based on LDA and SVM, comprising:
[0020] A data preprocessing module, used to obtain fault data when a fault occurs in a DC converter valve of the CLCC system, preprocess the fault data, and extract fault waveform feature data;
[0021] A dimension reduction processing module, used for performing dimension compression and data screening on the fault waveform feature data by using an LDA model to generate input fault features;
[0022] The classification and recognition module is used to use the input fault feature as the input of the trained SVM classifier to automatically identify the fault type.
[0023] Compared with the prior art, the present invention can accurately identify the fault type, which has important engineering practical significance for improving the intelligent level of valve control in DC systems and the safe operation of converter stations. The present invention has the following beneficial effects:
[0024] 1. The present invention pre-processes the fault data collected when a fault occurs in the DC converter valve of the CLCC system, extracts the fault features, and uses the SVM classifier for classification and identification, which can more specifically implement fault type identification with high identification accuracy.
[0025] 2. After extracting the fault feature data, the present invention uses the LDA model to perform dimension compression and data screening on the fault waveform feature data, which can be more effectively and accurately applied to subsequent classification processing and improve the efficiency of classification and recognition. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a schematic diagram of the process of the present invention;
[0027] Figure 2 A schematic diagram of a fault occurrence location in an embodiment of the present invention;
[0028] Figure 3 Schematic diagram of waveform data of a training set in an embodiment of the present invention;
[0029] Figure 4 This is a schematic diagram of classification results of a classifier in an embodiment of the present invention;
[0030] Figure 5 This is a schematic diagram of a fault waveform when an AB phase short circuit occurs in an embodiment of the present invention;
[0031] Figure 6 Schematic diagram of waveform of VT1 valve short circuit fault in an embodiment of the present invention;
[0032] Figure 7 This is a schematic diagram of waveforms of a VT1 valve short circuit grounding fault occurring in an embodiment of the present invention;
[0033] Figure 8 Schematic diagram of waveform of 6-pulse anode grounding fault in an embodiment of the present invention;
[0034] Fig. 9 This is a schematic diagram of a 6-pulse cathode grounding fault occurring in an embodiment of the present invention;
[0035] Fig.10 Schematic diagram of a 6-pulse short-circuit fault waveform in an embodiment of the present invention. DETAILED DESCRIPTION
[0036] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0037] Example 1
[0038] refer to Figure 1As shown, this embodiment provides a CLCC system fault identification method based on LDA and SVM, which is an artificial intelligence fault identification method for a converter valve of a high-voltage direct current transmission system including a controllable commutation converter, comprising the following steps:
[0039] In step S1, a SVM (Support Vector Machine) classifier is constructed and trained based on simulation data and historical data;
[0040] In step S2, fault data when a fault occurs in a DC converter valve of the CLCC system is obtained;
[0041] In step S3, the fault data is preprocessed to extract fault waveform characteristic data;
[0042] In step S4, the LDA (Linear Discriminant Analysis) model is used to perform dimension compression and data screening on the fault waveform feature data to generate input fault features;
[0043] In step S5, the input fault feature is used as the input of the trained SVM classifier to automatically identify the fault type.
[0044] In the above step S2, the collected fault data may include overvoltage data, overcurrent data, etc.
[0045] In the above step S3, the preprocessing includes waveform standard normalization processing, and calculating the characteristic data with large differences between different faults. The fault waveform characteristic data may include pulse peak value, frequency component, pulse rise time, etc.
[0046] In the above step S5, the fault types include AB phase short circuit, VT1 valve short circuit, VT1 valve short circuit grounding fault, 6-pulse anode grounding fault, 6-pulse cathode grounding fault, 6-pulse short circuit fault and normal situation.
[0047] In a preferred embodiment, the method further includes: when the fault type is a normal condition, the corresponding fault data is expanded into the training sample to achieve data sample expansion and online adjustment and optimization of the model.
[0048] The above fault identification method includes two main parts: LDA dimension reduction and SVM classification.
[0049] Linear discriminant analysis (LDA) is a supervised learning algorithm for classification problems. It is mainly used for dimensionality reduction and classification, especially for mapping high-dimensional data to low-dimensional space in classification tasks. It considers category information while reducing dimensionality, so it is more efficient than algorithms such as logistic regression and principal component analysis (PCA), and is suitable for multi-classification problems such as fault identification.
[0050] LDA projects the data into a low-dimensional space, maximizes the inter-class variance and minimizes the intra-class variance, and finds a projection direction so that the projected data points can be better classified on the new coordinate axis. LDA is a supervised data dimensionality reduction method that can better capture the characteristics of fault data.
[0051] LDA involves the calculation of intra-class scatter matrix, inter-class scatter matrix, eigenvalues and eigenvectors. w ) is used to measure the degree of dispersion between samples of the same type. The calculation method is as follows:
[0052]
[0053] where c is the number of fault categories, S i is the scatter matrix of the i-th type fault data, and the calculation formula is:
[0054]
[0055] X i is the set of all fault samples of the i-th category, μ i is the mean vector of the i-th type fault samples.
[0056] The inter-class scatter matrix (S b ) is used to measure the degree of discreteness between samples of different classes. The calculation method is as follows:
[0057]
[0058] N i is the number of samples in the i-th class, μ i is the mean vector of the i-th class sample, and μ is the overall mean of all samples. Solve the generalized eigenvalue and find the best projection direction to maximize the inter-class divergence of the fault data set and minimize the intra-class divergence.
[0059] S b ω=λS w ω
[0060] ω is the eigenvector, λ is the corresponding eigenvalue, and the eigenvectors corresponding to the largest 7 eigenvalues are selected to form the projection matrix W. The data is projected onto the selected eigenvector to obtain the reduced-dimensional data:
[0061] X 1 =XW
[0062] X is the original data matrix, X 1 is the data matrix obtained after dimensionality reduction.
[0063] Distribute the fault data of different types as much as possible, use SVM to find an optimal hyperplane in the feature space, let this hyperplane maximize the interval between different fault categories, and establish a support vector group for fault condition data and normal condition data. Use the SVM model to find an optimal hyperplane in the feature space. This hyperplane can maximize the interval between different fault categories. The support vectors are the data points closest to this hyperplane. Use SVM to establish classifiers M, F 1 , F 2 , F 3 , F 4 , F 5 , F 6 , F 7 They represent AB phase short circuit, VT1 valve short circuit, VT1 valve short circuit grounding fault, 6-pulse anode grounding fault, 6-pulse cathode grounding fault, 6-pulse short circuit fault and normal situation. The fault location is as follows: Figure 2 As shown. Constitute 7 support vector groups SV 1 , SV 2 , SV 3 , SV 4 , SV 5 , SV 6 , SV 7 , realize the construction of classifier M, F 1 , F 2 , F 3 , F 4 , F 5 , F 6 , F 7 They are counted as a, b, c, d, e, f, g in the classifier respectively.
[0064] Table 1 Output sample number
[0065]
[0066] SVM-based classifier training and usage process reference Figure 1 As shown in Table 1, ag represents the fault training set, and class a represents n groups of fault data with the fault type being AB phase short circuit, where X a1 The first group of fault data represents the fault type of AB phase short circuit, and the rest are the same. The data with known fault types are first processed by LDA to reduce the dimension of the data. a -y g The input is fed into the classifier M, and the fault category is finally output. The input of the classifier M is the fault data after LDA dimension reduction, and the output is the fault number of the judged sample.
[0067] Before using SVM to determine the fault type, it is necessary to first use the fault training set to train the SVM model. This embodiment takes the fault at the outlet current as an example. The training set input waveform data is as follows: Figure 3 As shown, curve a represents an AB phase short circuit fault, curve b represents a VT1 valve short circuit fault, curve c represents a VT1 valve short circuit grounding fault, curve d represents a 6-pulse anode grounding fault on the y side, curve e represents a 6-pulse cathode grounding fault on the y side, and curve e represents a 6-pulse short circuit fault.
[0068] If the judgment result is a normal waveform without fault, the normal waveform can be re-input into the SVM model to achieve data sample expansion and online adjustment and optimization of the model.
[0069] After training the SVM model, 7 sets of waveform data are randomly extracted from the database, numbered 1-7, and input into the fault recognition model. The output a represents the fault type of AB phase short circuit, b represents the fault type of VT1 valve short circuit, c represents the fault type of VT1 valve short circuit grounding, d represents the fault type of y side 6 pulse anode grounding, e represents the fault type of y side 6 pulse cathode grounding, f represents the fault type of 6 pulse short circuit, and g represents normal without fault. These 7 sets of waveform data are input into the classifier M, and the following can be obtained: Figure 4 The classification results and verification results show that the fault types of the first and second groups of data are AB phase short circuit, the third and fourth groups are 6-pulse short circuit, the fifth group is a normal situation without fault, the sixth group is VT1 valve short circuit and grounding, and the seventh group of data is a 6-pulse cathode grounding fault on the y side. The judgment results are consistent with the actual fault results.
[0070] Taking the output current as an example, when a short circuit occurs between phases AB, the fault waveform is as follows: Figure 5 As shown, the VT1 valve short circuit fault waveform is as follows Figure 6 As shown, the VT1 valve short circuit grounding fault waveform is as follows Figure 7 As shown, the 6-pulse anode grounding fault waveform is as follows Figure 8 As shown, 6 pulse cathode grounding fault Fig. 9 As shown, the 6-pulse short-circuit fault waveform is as follows Fig.10 As shown in the figure, the measured fault waveform data is processed by LDA and then input into the trained SVM model to obtain the fault identification result.
[0071] If the above method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0072] Example 2
[0073] The present embodiment provides a CLCC system fault identification system based on LDA and SVM, including a data preprocessing module, a dimensionality reduction processing module and a classification identification module, wherein the data preprocessing module is used to obtain fault data when a fault occurs in a DC converter valve of the CLCC system, preprocess the fault data, and extract fault waveform feature data; the dimensionality reduction processing module is used to use an LDA model to perform dimensionality compression and data screening on the fault waveform feature data to generate input fault features; the classification identification module is used to use the input fault features as input of a trained SVM classifier to automatically identify and obtain the fault type.
[0074] The rest is the same as in Example 1.
[0075] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.
[0076] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0077] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0078] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0079] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A CLCC system fault identification method based on LDA and SVM, characterized in that: The following steps are involved: Acquire fault data when a fault occurs in a DC converter valve of the CLCC system, pre-process the fault data, and extract fault waveform feature data; The LDA model is used to perform dimension compression and data screening on the fault waveform feature data to generate input fault features; The input fault feature is used as the input of the trained SVM classifier to automatically identify the fault type.
2. The CLCC system fault identification method based on LDA and SVM according to claim 1 is characterized in that: The fault data includes overvoltage data or overcurrent data.
3. The CLCC system fault identification method based on LDA and SVM according to claim 1, characterized in that: The preprocessing includes waveform standard normalization processing.
4. The CLCC system fault identification method based on LDA and SVM according to claim 1, characterized in that: The fault waveform characteristic data includes pulse peak value, frequency component and / or pulse rise time.
5. The CLCC system fault identification method based on LDA and SVM according to claim 1, characterized in that: The LDA model performs dimension compression and data screening by calculating the intra-class scatter matrix, the inter-class scatter matrix and the projection matrix.
6. The CLCC system fault identification method based on LDA and SVM according to claim 5, characterized in that: The projection matrix satisfies the requirement that the inter-class divergence of the fault data set is maximized and the intra-class divergence is minimized.
7. The CLCC system fault identification method based on LDA and SVM according to claim 1, characterized in that: The fault types include AB phase short circuit, VT1 valve short circuit, VT1 valve short circuit grounding fault, 6-pulse anode grounding fault, 6-pulse cathode grounding fault, 6-pulse short circuit fault and normal situation.
8. The CLCC system fault identification method based on LDA and SVM according to claim 7, characterized in that: The method further includes: When the fault type is normal, the corresponding fault data is expanded into the training sample to achieve data sample expansion and online adjustment and optimization of the model.
9. A computer-readable storage medium, characterized in that: It comprises one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs comprise instructions for executing the CLCC system fault identification method based on LDA and SVM as claimed in any one of claims 1 to 8.
10. A CLCC system fault identification system based on LDA and SVM, characterized in that: include: A data preprocessing module, used to obtain fault data when a fault occurs in a DC converter valve of the CLCC system, preprocess the fault data, and extract fault waveform feature data; A dimension reduction processing module, used for performing dimension compression and data screening on the fault waveform feature data by using an LDA model to generate input fault features; The classification and recognition module is used to use the input fault feature as the input of the trained SVM classifier to automatically identify the fault type.