SVM-based wind power plant power transmission system fault detection and identification model and system

Abstract

The invention discloses a wind power plant power transmission system fault detection and identification model and system based on an SVM, and relates to the technical field of power systems, and the construction of the model comprises the steps: S1, defining a problem and a target; s2, data acquisition; s3, data preprocessing; s4, constructing an SVM (Support Vector Machine) model; s5, feature selection and dimension reduction; s6, performing model evaluation; s7, carrying out model integration; and the model is carried in a fault detection and identification system. According to the model and the system, a fault detection and identification model is constructed by using the identification capability of a support vector machine (SVM), the model can automatically learn and identify various fault signals from a large amount of data, and high accuracy can be maintained even under the condition that the data are non-linearly separable, so that the real-time monitoring capability of the system carrying the model can ensure quick response, and the fault detection and identification efficiency is improved. The time from fault detection to fault processing is greatly shortened, so that risks and potential economic losses are reduced, and the operation efficiency of the wind power plant is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power systems, and in particular to a fault detection and identification model and system for a wind farm power transmission system based on SVM. Background Art

[0002] In the operation of wind farms, the reliability of the transmission system is crucial. Timely detection and identification of faults can significantly reduce downtime and maintenance costs, and ensure the continuity of power supply. Traditional fault detection methods mainly rely on manual inspections and simple threshold trigger systems, which often fail to detect faults in a timely manner and have difficulty dealing with complex fault types and hidden faults. In addition, the models built into traditional methods usually lack sufficient flexibility and accuracy in data processing and fault prediction, and cannot effectively process large-scale data or identify complex fault modes.

[0003] The shortcomings of the prior art are mainly reflected in:

[0004] Slow response time: Manual inspections take a long time to react and rely on manual experience, which is prone to errors.

[0005] Limited accuracy: Traditional rule-based models are not very accurate when dealing with complex or unknown failures.

[0006] Insufficient data utilization: A large amount of data obtained from sensors has not been fully analyzed and utilized, and the fault prediction capability is limited. Summary of the invention

[0007] In view of the deficiencies of the prior art, the present invention provides a wind farm power transmission system fault detection and identification model and system based on SVM to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a wind farm transmission system fault detection and identification model based on SVM, the construction of the model includes the following contents:

[0009] S1. Define the problem and goal

[0010] Clarify the problem and the goal of the model. For the wind farm transmission system, the fault types include line breakage, insulator damage, and voltage drop. The goal of the model is to accurately detect and identify the fault type, define the fault category, and determine the evaluation criteria and performance indicators for fault detection so that maintenance and repair can be carried out in a timely manner to ensure the stable operation of the transmission system.

[0011] S2. Data Collection

[0012] Collect data from multiple data sources at the wind farm, including sensor data such as current, voltage, power, temperature, vibration, operation logs, and system alarm records;

[0013] S3. Data preprocessing

[0014] Data preprocessing includes data cleaning, data standardization, and data feature engineering to improve data quality and reliability;

[0015] S4. Construction of SVM model

[0016] Define the objective function of SVM, introduce a specific kernel function (K) based on the characteristics of the wind farm transmission system, and perform parameter optimization and model training;

[0017] S5. Feature selection and dimensionality reduction

[0018] Perform feature selection and dimensionality reduction to improve the model's operating efficiency and avoid overfitting. Use principal component analysis and linear discriminant analysis to reduce the dimension of the feature space while maintaining the main variability of the data.

[0019] S6. Model Evaluation

[0020] After the model training is completed, the performance of the model is evaluated on an independent test set;

[0021] S7. Model Integration

[0022] The evaluated SVM model is integrated into the fault detection and identification system, and the model receives data, performs fault detection and issues alarms in real time.

[0023] To further optimize the technical solution, in step S3, data preprocessing specifically includes:

[0024] Data cleaning to remove outliers and noise;

[0025] Data standardization and normalization are performed to make the data at the same level to facilitate SVM processing;

[0026] Data feature engineering: extracting data features that are helpful for fault identification, such as statistical features and frequency domain features;

[0027] During the data preprocessing process, the proportion of fault data in the collected data is small, and oversampling or undersampling technology is used to deal with the data imbalance problem.

[0028] To further optimize the technical solution, in step S4, the objective function of the SVM is defined, and the objective function is as follows:

[0029]

[0030] Where W is the weight vector, (C) is the penalty parameter, (ξ i ) is a slack variable, indicating the degree of misclassification of the (i)th data point;

[0031] Based on the characteristics of the wind farm transmission system, a specific kernel function (K) is introduced, which is as follows:

[0032]

[0033] in, and is the value of samples (i) and (j) on the (k)th feature, and the parameter (p k ) is a feature-dependent index that adjusts the sensitivity of each feature to the fault type, (d) is the total number of features, and (γ) is the width parameter of the kernel function that controls the local sensitivity of the function.

[0034] To further optimize the technical solution, in step S3, based on the introduction of a specific kernel function (K), parameter optimization and model training are performed to obtain the process of the SVM model as follows:

[0035] Assume a training data set in is an input vector with (d)-dimensional features, y i ∈{-1, 1} is the corresponding class label;

[0036] Perform parameter optimization to train the SVM model:

[0037]

[0038] The constraints are as follows:

[0039]

[0040] Among them, (φ(x i )) is the input vector mapped to the high-dimensional space by the kernel function (K), (b)) is the bias term;

[0041] Establish a decision function, and the final decision function is used to classify new observation data (x):

[0042]

[0043] where (α i) is the Lagrange multiplier obtained by solving the dual problem during training.

[0044] Further optimizing the technical solution, in step S5, the principal component analysis method transforms the data by finding the direction with the largest variance in the data. Assuming that we have a feature space of (d) dimensions, the goal is to find a feature subspace of lower dimension, assuming (k) dimensions, while retaining the variance of the data;

[0045] The principal component analysis method is as follows:

[0046] Dataset:

[0047]

[0048] in, is the mean of the original data (X);

[0049] Covariance matrix calculation:

[0050]

[0051] Eigenvalue decomposition:

[0052] SV=VΛ

[0053] Where (V) is the eigenvector and (Λ) is the corresponding eigenvalue;

[0054] Select the principal components:

[0055] Select the eigenvectors corresponding to the largest (k) eigenvalues;

[0056] Projection data:

[0057] X reduced =X centered V k

[0058] Among them, (V k k) is the selected feature vector.

[0059] Further optimizing the technical solution, in step S5, the linear discriminant analysis method is used to find the best projection direction so that the intra-class variance is minimized and the inter-class variance is maximized;

[0060] The linear discriminant analysis method is as follows:

[0061] The intra-class hash matrix (S w ) and the inter-class hash matrix (S b ):

[0062]

[0063] Where (C) is the number of categories, (n c ) is the number of samples of category (c), (μ c ) is the mean of category (c), (μ) is the overall mean;

[0064] Optimization problem:

[0065]

[0066] Solve the eigenvalue problem:

[0067]

[0068] The eigenvector corresponding to the largest eigenvalue is selected as the projection direction.

[0069] Further optimizing the technical solution, in step S6, the evaluation indicators include accuracy, recall, F1 score, and receiver operating characteristic curve (ROC) and area under the curve (AUC);

[0070] The method for model evaluation is as follows:

[0071] Data splitting, using a separate test set to evaluate the model;

[0072] Cross-validation, using cross-validation techniques to reduce chance and overfitting;

[0073] Sensitivity analysis, analyzing the sensitivity of the model to different fault types;

[0074] Model comparison, compare the current model with other baseline models or previous models to evaluate their relative performance.

[0075] A wind farm transmission system fault detection and identification system based on SVM integrates the above-mentioned wind farm transmission system fault detection and identification model based on SVM, including a data acquisition module, a data preprocessing module, a feature selection and dimension reduction module, an SVM optimization training module, a real-time monitoring detection module and a performance evaluation module.

[0076] To further optimize the technical solution, the data acquisition module is also closely integrated with the on-site SCADA system to receive sensor and equipment status information in real time and obtain data from operation logs and system alarm records. The data will be used to train and verify the fault detection and identification model.

[0077] To further optimize the technical solution, the system also includes an update and maintenance module. Based on changes in the operating environment and technical conditions of the wind farm, the update and maintenance module supports regular updating and adjustment of the SVM model, and retrains the model to adapt to new failure modes and data changes, while collecting user feedback to optimize the accuracy and response speed of the model.

[0078] Compared with the prior art, the present invention provides a wind farm power transmission system fault detection and identification model and system based on SVM, which has the following beneficial effects:

[0079] The SVM-based wind farm transmission system fault detection and identification model and system constructs a fault detection and identification model by utilizing the recognition capability of support vector machines (SVM). The model can automatically learn and identify various fault signals from a large amount of data, and can maintain high accuracy even when the data is nonlinearly separable. The real-time monitoring capability of the system equipped with the model ensures rapid response, greatly shortening the time from fault detection to processing, thereby reducing risks and potential economic losses, which not only improves the operating efficiency of the wind farm, but also reduces maintenance costs and enhances the overall reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 A schematic diagram of the construction process of a wind farm power transmission system fault detection and identification model based on SVM proposed by the present invention;

[0081] Figure 2 A schematic flow chart of a principal component analysis method in a wind farm power transmission system fault detection and identification model based on SVM proposed by the present invention;

[0082] Figure 3 A schematic flow chart of a linear discriminant analysis method in a wind farm power transmission system fault detection and identification model based on SVM proposed by the present invention;

[0083] Figure 4 A flow chart of a model evaluation method in a wind farm power transmission system fault detection and identification model based on SVM proposed by the present invention;

[0084] Figure 5 This is a structural schematic diagram of a wind farm transmission system fault detection and identification system based on SVM proposed by the present invention. DETAILED DESCRIPTION

[0085] The following will be combined with 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.

[0086] Embodiment 1:

[0087] See also Figure 1 A wind farm transmission system fault detection and identification model based on SVM is proposed. The construction of the model includes the following contents:

[0088] S1. Define the problem and goal

[0089] Clarify the problem and the goal of the model. For the wind farm transmission system, the fault types include line breakage, insulator damage, and voltage drop. The goal of the model is to accurately detect and identify the fault type, define the fault category, and determine the evaluation criteria and performance indicators of fault detection, such as accuracy, recall rate, and F1 score, so as to carry out maintenance and repair in a timely manner and ensure the stable operation of the transmission system.

[0090] S2. Data Collection

[0091] Collect data from multiple data sources at the wind farm, including sensor data such as current, voltage, power, temperature, vibration, operation logs, and system alarm records.

[0092] In this embodiment, it is ensured that the collected data can fully reflect the operating status of the system and include data before and after the failure occurs. The quality of the data directly affects the performance of the model, so the integrity and accuracy of the data must be ensured.

[0093] S3. Data preprocessing

[0094] Data preprocessing includes data cleaning, data standardization, and data feature engineering to improve the quality and reliability of data.

[0095] In this embodiment, data preprocessing specifically includes:

[0096] Data cleaning to remove outliers and noise;

[0097] Data standardization and normalization are performed to make the data at the same level to facilitate SVM processing;

[0098] Data feature engineering: extracting data features that are helpful for fault identification, such as statistical features and frequency domain features;

[0099] During the data preprocessing process, the proportion of fault data in the collected data is small, and oversampling or undersampling technology is used to deal with the data imbalance problem.

[0100] S4. Construction of SVM model

[0101] The objective function of SVM is defined, and based on the characteristics of the wind farm transmission system, a specific kernel function (K) is introduced, and parameter optimization and model training are performed.

[0102] In this embodiment, the objective function of SVM is defined as follows:

[0103]

[0104] Where w is the weight vector, (C) is the penalty parameter, (ξ i ) is a slack variable, indicating the degree of misclassification of the (i)th data point;

[0105] Based on the characteristics of the wind farm transmission system, a specific kernel function (K) is introduced, which is as follows:

[0106]

[0107] in, and is the value of samples (i) and (j) on the (k)th feature, and the parameter (p k ) is a feature-dependent index that adjusts the sensitivity of each feature to the fault type, (d) is the total number of features, and (γ) is the width parameter of the kernel function that controls the local sensitivity of the function.

[0108] Feature dependence index (p k ): The introduction of this parameter takes into account the different importance of different features in fault diagnosis. For example, a small change in voltage may indicate certain fault types more significantly than the same change in temperature, so the voltage feature (p k ) may be greater than the temperature characteristic (p k ).

[0109] Adaptive penalty parameter (C i ): Traditional SVM uses a single penalty parameter (C), but in our model, each sample has an adaptive (C i ), adjusted according to the labels and features of the samples, making the model more sensitive to key fault types.

[0110] Dynamically adjusted (γ) parameter: We design (γ) to be dynamically adjusted, which will be automatically adjusted according to the statistical characteristics of the real-time data stream (such as mean and standard deviation) to adapt to changes in the environment and maintain the sensitivity and accuracy of the model.

[0111] This enables the SVM model to not only more accurately identify and classify faults in wind farm transmission systems, but also adapt to environmental changes and improve the robustness and adaptability of the model.

[0112] Furthermore, based on the introduction of a specific kernel function (K), parameter optimization and model training are performed to obtain the process of the SVM model as follows:

[0113] Assume a training data set in is an input vector with (d)-dimensional features, y i ∈{-1, 1} is the corresponding class label;

[0114] Perform parameter optimization to train the SVM model:

[0115]

[0116] The constraints are as follows:

[0117]

[0118] Among them, (φ(x i )) is the input vector mapped to the high-dimensional space by the kernel function (K), (b) is the bias term;

[0119] Establish a decision function, and the final decision function is used to classify new observation data (x):

[0120]

[0121] where (α i) is the Lagrange multiplier obtained by solving the dual problem during training.

[0122] In this SVM model:

[0123] Feature dependency index (pk): allows the model to adaptively weight different features based on different sensitivities to fault types.

[0124] Adaptive penalty parameter (C i ): Provides personalized misclassification penalties for each data point and increases attention to key data points.

[0125] Dynamic kernel width parameter (γ(t)): Adaptively adjusts according to changes in real-time data to ensure that the model maintains optimal performance in a changing environment.

[0126] This model can not only more accurately identify various fault types in the wind farm transmission system, but also adaptively adjust its performance to cope with changes in the environment and data characteristics. In this way, the model can achieve higher accuracy and stability in practical applications.

[0127] S5. Feature selection and dimensionality reduction

[0128] Feature selection and dimensionality reduction are performed to improve the operating efficiency of the model and avoid overfitting. Principal component analysis and linear discriminant analysis are used to reduce the dimension of the feature space while maintaining the main variability of the data.

[0129] In this embodiment, the principal component analysis method transforms the data by finding the direction with the largest variance in the data. Assuming that we have a feature space of (d) dimensions, the goal is to find a feature subspace of lower dimension, assumed to be (k) dimensions, while preserving the variance of the data;

[0130] like Figure 2 As shown, the principal component analysis method is as follows:

[0131] Dataset:

[0132]

[0133] in, is the mean of the original data (X);

[0134] Covariance matrix calculation:

[0135]

[0136] Eigenvalue decomposition:

[0137] SV=VΛ

[0138] Where (V) is the eigenvector and (Λ) is the corresponding eigenvalue;

[0139] Select the principal components:

[0140] Select the eigenvectors corresponding to the largest (k) eigenvalues;

[0141] Projection data:

[0142] X reduced =X centered V k

[0143] Among them, (V k ) is the selected feature vector.

[0144] In this embodiment, the linear discriminant analysis method is used to find the best projection direction so that the intra-class variance is minimized and the inter-class variance is maximized;

[0145] like Figure 3 As shown, the linear discriminant analysis method is as follows:

[0146] The intra-class hash matrix (S w ) and the inter-class hash matrix (S b ):

[0147]

[0148] Where (C) is the number of categories, (n c ) is the number of samples of category (c), (μ c ) is the mean of category (c), (μ) is the overall mean;

[0149] Optimization problem:

[0150]

[0151] Solve the eigenvalue problem:

[0152]

[0153] The eigenvector corresponding to the largest eigenvalue is selected as the projection direction.

[0154] In practical applications, the principal component analysis method is first used to reduce the data dimension to reduce noise and reduce the amount of calculation. Then, the linear discriminant analysis method is used to further extract the features that are most helpful in distinguishing different fault categories. Through this combined method, we can retain the most critical information for fault identification while removing unnecessary information, thereby improving the accuracy and efficiency of the SVM model in fault detection of wind farm transmission systems.

[0155] S6. Model Evaluation

[0156] After the model is trained, the performance of the model is evaluated on an independent test set.

[0157] In this embodiment, the evaluation indicators include precision, recall, F1 score, and receiver operating characteristic curve (ROC) and area under the curve (AUC);

[0158] like Figure 4 As shown, the method of model evaluation is as follows:

[0159] Data splitting, using a separate test set to evaluate the model;

[0160] Cross-validation, using cross-validation techniques to reduce chance and overfitting;

[0161] Sensitivity analysis, analyzing the sensitivity of the model to different fault types;

[0162] Model comparison, compare the current model with other baseline models or previous models to evaluate their relative performance.

[0163] The evaluated data are shown in Table 1.

[0164] Evaluation Metrics data Accuracy 0.92 Recall rate (Recall)-line break 0.90 Recall rate (Recall) - insulator damage 0.95 Recall rate (Recall)-voltage drop 0.88 Precision-Line Break 0.93 Precision-Insulator Damage 0.92 Precision-Voltage Drop 0.90 F1 Score - Line Break 0.91 F1 Score - Insulator Damage 0.93 F1 Score-Voltage Drop 0.89 Area Under the ROC Curve (AUC) - Overall 0.94 Area Under the ROC Curve (AUC) - Line Breaks 0.92 Area Under the ROC Curve (AUC) - Insulator Damage 0.96 Area under the ROC curve (AUC) - voltage drop 0.90

[0165] Table 1

[0166] Several key evaluation metrics, including accuracy, recall, precision, and F1 score, are listed in Table 1, which are evaluated for different types of faults (line break, insulator damage, voltage drop). In addition, we also calculate the area under the ROC curve (AUC) for the overall and each fault type, which is a common metric to measure the overall performance of the model, especially when dealing with unbalanced datasets.

[0167] Other factors need to be considered during evaluation, such as model training time, prediction time, resource consumption, etc., to ensure the feasibility and efficiency of the model in practical applications.

[0168] S7. Model Integration

[0169] The evaluated SVM model is integrated into the fault detection and identification system, and the model receives data, performs fault detection and issues alarms in real time.

[0170] Embodiment 2:

[0171] See also Figure 5 A wind farm transmission system fault detection and identification system based on SVM is integrated with the wind farm transmission system fault detection and identification model based on SVM described in Example 1, including a data acquisition module, a data preprocessing module, a feature selection and dimension reduction module, an SVM optimization training module, a real-time monitoring detection module and a performance evaluation module.

[0172] The data acquisition module is also closely integrated with the on-site SCADA system to receive sensor and equipment status information in real time and obtain data from operation logs and system alarm records. The data will be used to train and verify the fault detection and identification model.

[0173] in:

[0174] Data preprocessing module, which processes the raw data to make it suitable for model processing. It includes data cleaning (removing outliers and noise), standardization (normalization to make the data at the same level) and feature engineering (extracting statistical features and frequency domain features that are helpful for fault identification). Data preprocessing also involves dealing with data imbalance issues, such as enhancing the model's ability to identify rare fault categories through oversampling or undersampling techniques.

[0175] Feature selection and dimensionality reduction module, in order to improve the efficiency and performance of the model, the feature selection and dimensionality reduction module screens and optimizes data features. It uses techniques such as principal component analysis (PCA) and linear discriminant analysis (LDA) to reduce the number of features while retaining the most representative information as much as possible. This helps to reduce computational complexity, improve model operation efficiency, and reduce the risk of overfitting.

[0176] SVM optimization training module, which trains the SVM model based on the selected kernel function and the optimal parameters obtained through grid search. The training process includes the use of cross-validation techniques to evaluate model performance and ensure that the model has good generalization ability. This module is the core of the system and directly determines the accuracy and efficiency of fault detection.

[0177] Real-time monitoring and detection module, which receives new sensor data in real time and uses the trained SVM model for fault detection. When a potential fault is detected, the system automatically triggers an alarm and notifies the maintenance team. The efficiency and accuracy of this module is critical to preventing greater losses from faults.

[0178] Performance evaluation module, which is responsible for regularly evaluating the fault detection performance of the system, using indicators such as accuracy, recall, F1 score, ROC and AUC for evaluation.

[0179] The system also includes an update and maintenance module. Based on changes in the operating environment and technical conditions of the wind farm, the update and maintenance module supports regular updates and adjustments to the SVM model, retrains the model to adapt to new failure modes and data changes, and collects user feedback to optimize the accuracy and response speed of the model.

[0180] The beneficial effects of the present invention are:

[0181] The SVM-based wind farm transmission system fault detection and identification model and system constructs a fault detection and identification model by utilizing the recognition capability of support vector machines (SVM). The model can automatically learn and identify various fault signals from a large amount of data, and can maintain high accuracy even when the data is nonlinearly separable. The real-time monitoring capability of the system equipped with the model ensures rapid response, greatly shortening the time from fault detection to processing, thereby reducing risks and potential economic losses, which not only improves the operating efficiency of the wind farm, but also reduces maintenance costs and enhances the overall reliability of the system.

[0182] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0183] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A wind farm transmission system fault detection and identification model based on SVM, characterized in that: The construction of the model includes the following contents: S1. Define the problem and goal Clarify the problem and the goal of the model. For the wind farm transmission system, the fault types include line breakage, insulator damage, and voltage drop. The goal of the model is to accurately detect and identify the fault type, define the fault category, and determine the evaluation criteria and performance indicators for fault detection so that maintenance and repair can be carried out in a timely manner to ensure the stable operation of the transmission system. S2. Data Collection Collect data from multiple data sources at the wind farm, including sensor data such as current, voltage, power, temperature, vibration, operation logs, and system alarm records; S3. Data preprocessing Data preprocessing includes data cleaning, data standardization, and data feature engineering to improve data quality and reliability; S4. Construction of SVM model Define the objective function of SVM, introduce a specific kernel function (K) based on the characteristics of the wind farm transmission system, and perform parameter optimization and model training; S5. Feature selection and dimensionality reduction Perform feature selection and dimensionality reduction to improve the model's operating efficiency and avoid overfitting. Use principal component analysis and linear discriminant analysis to reduce the dimension of the feature space while maintaining the main variability of the data. S6. Model Evaluation After the model training is completed, the performance of the model is evaluated on an independent test set; S7. Model Integration The evaluated SVM model is integrated into the fault detection and identification system, and the model receives data, performs fault detection and issues alarms in real time.

2. A wind farm power transmission system fault detection and identification model based on SVM according to claim 1, characterized in that: In step S3, data preprocessing specifically includes: Data cleaning to remove outliers and noise; Data standardization and normalization are performed to make the data at the same level to facilitate SVM processing; Data feature engineering: extracting data features that are helpful for fault identification, such as statistical features and frequency domain features; During the data preprocessing process, the proportion of fault data in the collected data is small, and oversampling or undersampling technology is used to deal with the data imbalance problem.

3. A wind farm power transmission system fault detection and identification model based on SVM according to claim 1, characterized in that: In step S4, the objective function of SVM is defined, and the objective function is as follows: Where w is the weight vector, (C) is the penalty parameter, and (ξi) is the slack variable, which indicates the degree of misclassification of the (i)th data point; Based on the characteristics of the wind farm transmission system, a specific kernel function (K) is introduced, which is as follows: in, and is the value of samples (i) and (j) on the (k)th feature, and the parameter (p k ) is a feature-dependent index that adjusts the sensitivity of each feature to the fault type, (d) is the total number of features, and (γ) is the width parameter of the kernel function that controls the local sensitivity of the function.

4. A wind farm power transmission system fault detection and identification model based on SVM according to claim 1, characterized in that: In step S4, based on the introduction of a specific kernel function (K), parameter optimization and model training are performed to obtain the SVM model as follows: Assume a training data set in is an input vector with (d)-dimensional features, y i ∈{-1, 1} is the corresponding class label; Perform parameter optimization to train the SVM model: The constraints are as follows: Among them, (φ(x i )) is the input vector mapped to the high-dimensional space by the kernel function (K), (v) is the bias term; Establish a decision function, and the final decision function is used to classify new observation data (x): Among them, (α i ) is the Lagrange multiplier obtained by solving the dual problem during training.

5. The wind farm power transmission system fault detection and identification model based on SVM according to claim 1 is characterized in that: In step S5, the principal component analysis method transforms the data by finding the direction with the largest variance in the data. Assuming that we have a feature space of (d) dimensions, the goal is to find a feature subspace of lower dimension, assuming (k) dimensions, while preserving the variance of the data; The principal component analysis method is as follows: Dataset: in, is the mean of the original data (X); Covariance matrix calculation: Eigenvalue Decomposition: SV=VΛ Where (V) is the eigenvector and (Λ) is the corresponding eigenvalue; Select the principal component: Select the eigenvectors corresponding to the largest (k) eigenvalues; Projection data: X reduced =X centered V k Among them, (V k ) is the selected feature vector.

6. A wind farm power transmission system fault detection and identification model based on SVM according to claim 1, characterized in that: In step S5, a linear discriminant analysis method is used to find the best projection direction so that the intra-class variance is minimized and the inter-class variance is maximized; The linear discriminant analysis method is as follows: The intra-class hash matrix (S ω ) and the inter-class hash matrix (S b ): Where (C) is the number of categories, (n c ) is the number of samples of category (c), (μ c ) is the mean of category (c), (μ) is the overall mean; Optimization Problem: Solve the eigenvalue problem: The eigenvector corresponding to the largest eigenvalue is selected as the projection direction.

7. A wind farm power transmission system fault detection and identification model based on SVM according to claim 1, characterized in that: In step S6, the evaluation indicators include accuracy, recall, F1 score, and receiver operating characteristic curve (ROC) and area under the curve (AUC); The method for model evaluation is as follows: Data splitting, using a separate test set to evaluate the model; Cross-validation, using cross-validation techniques to reduce chance and overfitting; Sensitivity analysis, analyzing the sensitivity of the model to different fault types; Model comparison, compare the current model with other baseline models or previous models to evaluate their relative performance.

8. A wind farm power transmission system fault detection and identification system based on SVM, integrating the wind farm power transmission system fault detection and identification model based on SVM according to any one of claims 1 to 7, characterized in that: It includes data acquisition module, data preprocessing module, feature selection and dimension reduction module, SVM optimization training module, real-time monitoring and detection module and performance evaluation module.

9. A wind farm power transmission system fault detection and identification system based on SVM according to claim 8, characterized in that: The data acquisition module is also closely integrated with the on-site SCADA system to receive sensor and equipment status information in real time and obtain data from operation logs and system alarm records. The data will be used to train and verify the fault detection and identification model.

10. The wind farm power transmission system fault detection and identification system based on SVM according to claim 8, characterized in that: The system also includes an update and maintenance module. Based on changes in the operating environment and technical conditions of the wind farm, the update and maintenance module supports regular updates and adjustments to the SVM model, retrains the model to adapt to new failure modes and data changes, and collects user feedback to optimize the accuracy and response speed of the model.

Images