A high-voltage circuit breaker short-circuit current test system key parameter identification method

Abstract

The application discloses a kind of high-voltage circuit breaker short-circuit current test system key parameter identification method, belong to the technical field of high-voltage electrical equipment test of power system, first, historical test data is collected from test system, structured database is established, and data is preprocessed;Feature selection and importance evaluation are carried out by random forest algorithm combined with SHAP value analysis framework, and key parameter screening is carried out by combining threshold method and sorting method;Based on the key parameters screened, a logistic regression prediction model is constructed;Output key parameter list and its importance score sorting, generate feature importance chart to display analysis results, based on the analysis results, provide specific optimization suggestions for test system parameter configuration.By collecting test data through the system, using feature selection and machine learning algorithm, the importance of each parameter is automatically evaluated, and the key parameters are screened out, providing a scientific basis for test system design, parameter optimization and circuit breaker performance evaluation.

Description

Technical Field

[0001] This invention belongs to the field of testing technology for high-voltage electrical equipment in power systems, and particularly relates to a method for identifying key parameters of a high-voltage circuit breaker short-circuit current testing system. Background Technology

[0002] High-voltage circuit breakers are critical protection devices in power systems, and their breaking capacity must be verified through rigorous short-circuit current tests. The test system involves various electrical parameters, including short-circuit impedance, current amplitude, arc voltage, and transient recovery voltage (TRV) waveform characteristics. These parameters collectively determine the accuracy and reliability of the test, directly affecting the assessment of the circuit breaker's breaking capacity.

[0003] Currently, parameter identification and analysis in short-circuit current tests of high-voltage circuit breakers mainly rely on engineering experience and manual judgment. Technicians subjectively select important parameters by observing test waveforms, consulting historical data manuals, and combining their own experience. This method is not only inefficient and time-consuming, but also easily limited by personal experience, leading to inconsistent or incomplete identification results. Especially when facing multivariable and highly nonlinear test systems, traditional methods struggle to quantify the impact of parameters on system performance and cannot accurately identify key parameters.

[0004] While some studies have attempted to incorporate statistical analysis or simple regression models for parameter analysis, these methods are typically based on linear assumptions, making it difficult to capture complex nonlinear relationships between parameters, and they have limited capacity to handle large-scale experimental data. Furthermore, existing methods lack mechanisms for automatically ranking and filtering parameter importance, failing to provide clear guidance for optimizing experimental systems.

[0005] Data mining technology, as an effective tool for extracting potential patterns from massive amounts of data, has demonstrated significant advantages in various industrial sectors. However, in the field of high-voltage circuit breaker testing, its application to key parameter identification is yet to be seen. Therefore, developing a method capable of automatically, accurately, and efficiently identifying key parameters of high-voltage circuit breaker short-circuit current testing systems has become an urgent technical problem to be solved in this field.

[0006] This paper addresses the comprehensive technical challenges in high-voltage circuit breaker short-circuit current testing systems, including the low efficiency, high subjectivity, and insufficient accuracy of traditional parameter identification methods, as well as the inability of existing analysis methods to effectively handle multivariate nonlinear relationships. Specifically, current technologies lack a method to automatically filter out the key parameters that have the greatest impact on system performance from a large amount of experimental data. This results in incomplete experimental analysis, unclear optimization directions, and hinders the improvement of experimental efficiency and system reliability. Summary of the Invention

[0007] Purpose of the Invention: The purpose of this invention is to provide a method for identifying key parameters in a high-voltage circuit breaker short-circuit current test system. By collecting test data through the system, and utilizing feature selection and machine learning algorithms, the importance of each parameter is automatically assessed, and key parameters are selected, providing a scientific basis for test system design, parameter optimization, and circuit breaker performance evaluation.

[0008] Technical Solution: This invention provides a method for identifying key parameters in a high-voltage circuit breaker short-circuit current test system. The test system includes a voltage source, equivalent inductance, a TRV frequency modulation device, and a high-voltage circuit breaker. The method includes the following steps:

[0009] Step 1: Collect historical test data from the test system and establish a structured database;

[0010] Step 2: Preprocess the data in the structured database;

[0011] Step 3: Use the random forest algorithm combined with the SHAP value analysis framework to perform feature selection and importance assessment on the preprocessed data;

[0012] Step 4: Based on the selected and evaluated data, key parameters are screened using a combination of thresholding and ranking methods;

[0013] Step 5: Construct a logistic regression prediction model based on the selected key parameters;

[0014] Step 6: Output a list of key parameters and their importance scores, generate a feature importance graph to display the analysis results, and provide specific optimization suggestions for the parameter configuration of the experimental system based on the analysis results.

[0015] Furthermore, step 1 specifically involves: historical test data including short-circuit impedance. Current amplitude Arc voltage TRV frequency modulation capacitor Damping resistor Delay capacitor and voltage level Test types include T100, T60, T30, and T10; auxiliary information includes ambient temperature. and relative humidity .

[0016] Further, step 2 specifically involves: systematically preprocessing the collected raw experimental data; handling missing values ​​using the median imputation method, selecting representative values ​​for imputation based on the distribution characteristics of each parameter; outlier detection using the 3σ principle, uniformly processing the data, calculating the mean and standard deviation of each parameter's numerical distribution, and removing outlier data points that deviate from the mean by more than three standard deviations; firstly, calculating the mean of each parameter's numerical distribution. and standard deviation :

[0017]

[0018]

[0019] Where n represents the number of samples, x i Indicates the first One data point;

[0020] Then eliminate those that satisfy the condition. Outlier data points were identified; for key parameters of the TRV peak, box plots were used for auxiliary judgment to remove extreme outliers exceeding 1.5 times the interquartile range; data normalization employed the Min-Max scaling method for each parameter. Calculate the minimum value of each in the entire dataset. and maximum value Then, all parameter values ​​are linearly transformed to Interval:

[0021]

[0022] Furthermore, step 3 specifically involves: using the random forest algorithm combined with the SHAP value analysis framework to evaluate feature importance and quantify the impact of each experimental parameter on system performance; the specific configuration of the random forest model is as follows: it contains 200 decision trees, with a maximum depth limit of 15 layers for each decision tree, and uses Gini impurity as the node splitting criterion. The formula for calculating Gini impurity is:

[0023]

[0024] in Category in the node proportion, The total number of categories; calculated for each feature. The accurate feature importance score is obtained by averaging the reduction in impurity across all decision trees. :

[0025]

[0026] in Features In decision tree The reduction in impurities brought about by the process The total number of decision trees;

[0027] Based on cooperative game theory, the contribution of each feature to the model output is assigned. The SHAP value is calculated by considering all possible subset combinations of features, measuring the marginal contribution of each feature to the model's predicted output when added to a specific subset.

[0028]

[0029] in For the complete set of features, For feature subset, For model prediction function, For factorial calculation, the final feature importance score A weighted average method is used to synthesize the random forest importance score and the absolute value of SHAP in a 6:4 ratio:

[0030]

[0031] Furthermore, step 4 specifically involves using a combination of thresholding and ranking methods to screen key parameters. First, an importance score threshold is set. All parameters with importance scores above the threshold are retained as the candidate set of key parameters.

[0032]

[0033] Among them, X j Represents the candidate set of key parameters. This represents the final feature importance score.

[0034] Furthermore, step 5 specifically involves: the logistic regression model adopting a form including L2 regularization, whose mathematical expression is:

[0035]

[0036] in This represents the probability of a successful interruption. to The five key parameters selected are: to These are the model coefficients.

[0037] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method of the present invention.

[0038] The present invention also discloses a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the method of the present invention.

[0039] The present invention also discloses a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method of the present invention.

[0040] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0041] 1. High efficiency: This invention can automatically process large amounts of experimental data, quickly identify key parameters, significantly improve analysis efficiency, and shorten the experimental cycle;

[0042] 2. Accuracy: Based on the feature importance assessment of random forest combined with SHAP value analysis, this invention can accurately capture the complex nonlinear relationship between parameters. It can not only identify the main influencing factors, but also discover secondary but key parameters that are easily ignored by traditional methods, providing a reliable basis for comprehensive optimization.

[0043] 3. Objectivity: Through standardized data processing procedures and unified analysis standards, analytical biases caused by human factors are effectively eliminated. Different operators can obtain consistent analytical results using this method, significantly improving the repeatability and reliability of the analytical results, and providing technical support for product quality control and test standardization.

[0044] 4. Highly instructive: It clearly defines key parameters and their degree of influence, providing direct basis for experimental system optimization, parameter tuning, and performance evaluation. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the short-circuit current test system for high-voltage circuit breakers.

[0046] Figure 2 A schematic diagram of parameters extracted from the TRV waveform;

[0047] Figure 3 The flowchart shows the overall process of identifying key parameters of a high-voltage circuit breaker short-circuit current test system based on data mining.

[0048] Figure 4 Flowchart for feature selection and importance assessment;

[0049] Figure 5 A graph showing the results of the feature importance ranking;

[0050] Figure 6 A distribution chart showing the importance of the top 5 key parameters;

[0051] Figure 7 This is a comparison chart of model performance;

[0052] Figure 8 The diagram shows the results of efficiency improvement and accuracy loss. Detailed Implementation

[0053] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0054] This invention provides a method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system based on data mining. Figure 1 This is a schematic diagram of a high-voltage circuit breaker short-circuit current testing system. The system includes a three-phase AC power supply (AC), a protective circuit breaker (CB), an equivalent inductance (L), and an equivalent resistance (R). L The test system comprises a transformer (T), a TRV frequency modulation device, and a circuit breaker under test (TB). The TRV frequency modulation device consists of a frequency modulation capacitor C1, a series damping resistor R, and a time delay capacitor C2. The test system includes core components such as a voltage source, equivalent inductance, TRV frequency modulation device, and high-voltage circuit breaker. During the test, the system collects and records various electrical parameters in real time, including key indicators such as short-circuit impedance, current amplitude, arc voltage, TRV peak value, rate of rise, and time delay. Simultaneously, it records system performance evaluation parameters, such as breaking success rate, waveform conformity, and test repeatability, forming a complete parameter-performance mapping database.

[0055] The working principle of this invention is as follows: A parameter-performance database is constructed based on historical test data. Advanced data preprocessing techniques are used to clean and standardize the raw test data. A feature selection algorithm is employed to quantify the impact weight of each electrical parameter on system performance. A machine learning model is used to verify the accuracy and effectiveness of key parameter identification. Finally, a parameter importance ranking and a set of key parameters are output, providing data support for the optimization of the test system. The specific steps are as follows:

[0056] Step 1: Data Collection and Construction

[0057] Historical test data is systematically collected from the high-voltage circuit breaker short-circuit current test system to establish a structured database. In specific implementation, the range of input parameters to be collected must first be clearly defined, such as... Figure 1 The parameters in the schematic diagram of the high-voltage circuit breaker short-circuit current test system include short-circuit impedance. Current amplitude Arc voltage TRV frequency modulation capacitor Damping resistor Delay capacitor Core electrical parameters, including system configuration parameters such as voltage level. The test types include various methods such as T100, T60, T30, and T10, and the ambient temperature... relative humidity Auxiliary information such as electrical parameters. During the data acquisition process, all electrical parameters are... Synchronous acquisition at the specified rate ensures complete and uninterrupted waveform detail. Output performance metrics need to be comprehensively recorded, such as... Figure 2 A schematic diagram of parameters extracted from the TRV waveform, including key TRV waveform parameters such as peak value. rate of increase Delay The interruption result uses binary identifiers (1 for success, 0 for failure) and waveform conformity score. The waveform conformance score is based on the GB / T 1984-2014 standard envelope and is obtained by calculating the matching degree between the actual waveform and the standard envelope. The calculation formula is as follows:

[0058]

[0059] in, The number of sampling points. For the actual TRV waveform in Voltage value at time, For the standard envelope in The voltage value at any given time. The database is constructed using an SQL database for storage, including fields such as test number, timestamp, complete input parameter set, and output performance indicators. All data acquisition strictly follows the IEC 62271-100 protocol to ensure that the data format is uniform, the source is reliable, and it is traceable.

[0060] Step 2: Data Preprocessing

[0061] The collected raw experimental data underwent systematic preprocessing to ensure data quality met the requirements of subsequent analysis. For missing value handling, the median imputation method was used, selecting the most representative values ​​based on the distribution characteristics of each parameter. Specifically, for each parameter with missing values... Calculate the median of this parameter across all valid experimental data. And assign the median to the missing position:

[0062]

[0063] If the number of missing parameters in a test record exceeds 20% of the total number of parameters, that test record will be completely removed. In the outlier detection phase, based on... Following the principle of processing, first calculate the mean of the numerical distribution of each parameter. and standard deviation :

[0064]

[0065]

[0066] Then eliminate those that satisfy the condition. Outlier data points were identified; for key parameters such as TRV peak values, box plots were used for auxiliary judgment to remove extreme outliers exceeding 1.5 times the interquartile range. Data normalization employed a Min-Max scaling method for each parameter. Calculate the minimum value of each in the entire dataset. and maximum value Then, all parameter values ​​are linearly transformed to Interval:

[0067]

[0068] This method can effectively eliminate the influence of different dimensions on the analysis results and ensure that all parameters are fairly comparable in the feature selection process.

[0069] Step 3: Feature Selection and Importance Assessment

[0070] The flowchart for feature selection and importance assessment is as follows: Figure 4 As shown, Figure 4 This paper demonstrates the feature selection principle based on random forest and SHAP analysis. It uses the random forest algorithm combined with the SHAP value analysis framework to evaluate feature importance and accurately quantify the impact of each experimental parameter on system performance. The specific configuration of the random forest model is as follows: it contains 200 decision trees, with a maximum depth limit of 15 layers per tree. Gini impurity is used as the node splitting criterion. The formula for calculating Gini impurity is:

[0071]

[0072] in Category in the node proportion, This represents the total number of categories. This is calculated for each feature. The accurate feature importance score is obtained by averaging the reduction in impurity across all decision trees. :

[0073]

[0074] in Features In the tree The reduction in impurities brought about by the process This represents the total number of decision trees. To further enhance the interpretability and reliability of the evaluation results, the SHAP value analysis framework is used to fairly allocate the contribution of each feature to the model output based on cooperative game theory. The SHAP value is calculated by considering all possible subset combinations of features, measuring the marginal contribution of each feature to the model's predicted output when added to a specific subset:

[0075]

[0076] in For the complete set of features, For feature subset, For model prediction function, Factorial calculation. The final feature importance score is calculated using a weighted average, combining the random forest importance score and the absolute value of SHAP in a 6:4 ratio.

[0077]

[0078] This approach considers both the splitting importance of features in the model and the contribution of features to specific prediction results, ensuring that the feature importance assessment is both accurate and reliable, and has clear physical meaning.

[0079] Step 4: Key Parameter Screening

[0080] The comprehensive feature importance score is calculated based on the random forest algorithm combined with SAP value analysis. A combination of thresholding and ranking methods was used to screen key parameters. First, the importance score threshold was set to... This threshold is a reasonable value obtained through multiple experiments, effectively filtering out parameters that have a negligible impact on system performance. All parameters with importance scores higher than this threshold are retained as a candidate set of key parameters.

[0081]

[0082] Ensure that no parameters with significant impact are overlooked. Then, sort the candidate parameter set in descending order of importance score to form a parameter importance ranking list, as shown in the output. Figure 5 As shown in the figure. Considering the practical application requirements of the engineering project and the operability and usability of optimizing the experimental system, the top 5 parameters were selected from the ranking list to form the final set of key parameters. The final key parameter importance distribution diagram is shown in the figure. Figure 6 As shown. This screening mechanism ensures that the selected parameters are both statistically significant and meet the requirements of simplicity and operability in engineering applications. In practical applications, a typical set of key parameters usually includes the TRV frequency modulation capacitor. Damping resistor Current amplitude Short-circuit impedance Delay capacitor These core parameters collectively determine the dynamic response characteristics of the test system and the breaking performance of the circuit breaker, providing clear technical guidance for subsequent optimization of the test system.

[0083] Step 5: Model Validation and Performance Evaluation

[0084] A logistic regression prediction model was built using the selected key parameters, and the effectiveness and practicality of the key parameter subset were verified. First, the preprocessed complete dataset was randomly divided into a training set and a training set in a 7:3 ratio. and test set Stratified sampling was used during the partitioning process to ensure that the proportions of different trial types remained consistent across the training and test sets, guaranteeing the representativeness of the data distribution and the reliability of the model evaluation. The logistic regression model employed L2 regularization, and its mathematical expression is as follows:

[0085]

[0086] in This represents the probability of a successful interruption. to The five key parameters selected are: to These are the model coefficients. Regularization coefficients. Using grid search method Optimization is performed within a certain range to prevent model overfitting and improve generalization ability. Performance evaluation primarily uses accuracy and F1 score as core metrics. Accuracy measures the overall prediction accuracy of the model.

[0087]

[0088] in It is a true positive. It is a true negative. It was a false positive. It is a false negative. The F1 score comprehensively considers precision and recall:

[0089]

[0090] Among them accuracy Recall rate By comparing the model performance using the full parameter set and key parameter subsets, this study verifies the practical value of key parameter subsets in significantly reducing model complexity while maintaining prediction accuracy (requiring an accuracy decrease of no more than 3%), demonstrating the technical advantages of key parameter selection. The model performance metrics for using the full parameter set and key parameter subsets are output as follows: Figure 7 As shown, the efficiency improvement and accuracy loss output are as follows: Figure 8 As shown in the figure, the output shows that the performance metrics are basically the same when using the full parameter set and the subset of key parameters, improving efficiency by 50% while maintaining essentially the same accuracy.

[0091] Step Six: Output and Application of Results

[0092] A complete system for outputting and applying results is established, forming a comprehensive technical solution from data analysis to engineering practice. Regarding output, a list of key parameters and their importance scores are first output, clearly displaying the importance score and ranking of each key parameter in tabular form. Simultaneously, a feature importance bar chart is generated, visually presenting the analysis results. The horizontal axis represents the parameter name, and the vertical axis represents the importance score; the length of the bars intuitively reflects the degree of parameter importance. Furthermore, a SHAP summary chart is generated, showing the direction and magnitude of each parameter's contribution to the model output, helping to understand the detailed characteristics of the parameter's influence. In terms of engineering application, specific optimization suggestions are provided for the parameter configuration of the test system based on the identification results, guiding the monitoring and adjustment strategies of key parameters during the test; technical directions are indicated for circuit breaker design improvement, highlighting the core parameters affecting performance and their optimization potential; and a rapid analysis process based on key parameters is established for fault diagnosis, enabling rapid location of potentially problematic parameters when test anomalies occur. Ultimately, a complete closed loop from data acquisition, processing and analysis to result application is formed, significantly improving the intelligence level and efficiency of the test system, providing reliable technical support for the standardization and optimization of high-voltage circuit breaker short-circuit current testing, and promoting technological progress throughout the industry.

[0093] Through the systematic execution of the above-described specific embodiments, the present invention can automatically and accurately identify key parameters from complex test data, providing reliable technical support for the optimized design, precise control, and intelligent diagnosis of short-circuit current tests of high-voltage circuit breakers. The above are merely preferred embodiments of the present invention, but do not limit the patent scope of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of the present invention's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of the present invention.

Claims

1. A method for identifying key parameters of a high-voltage circuit breaker short-circuit current testing system, characterized in that, The test system includes a voltage source, equivalent inductance, TRV frequency modulation device, and high-voltage circuit breaker, and the method includes the following steps: Step 1: Collect historical test data from the test system and establish a structured database; Step 2: Preprocess the data in the structured database; Step 3: Use the random forest algorithm combined with the SHAP value analysis framework to perform feature selection and importance assessment on the preprocessed data; Step 4: Based on the selected and evaluated data, key parameters are screened using a combination of thresholding and ranking methods; Step 5: Construct a logistic regression prediction model based on the selected key parameters; Step 6: Output a list of key parameters and their importance scores, generate a feature importance graph to display the analysis results, and provide specific optimization suggestions for the parameter configuration of the experimental system based on the analysis results.

2. The method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system according to claim 1, characterized in that, Step 1 specifically involves: historical test data, including short-circuit impedance. Current amplitude Arc voltage TRV frequency modulation capacitor Damping resistor Delay capacitor and voltage level Test types include T100, T60, T30, and T10; auxiliary information includes ambient temperature. and relative humidity .

3. The method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system according to claim 1, characterized in that, Step 2 specifically involves: systematically preprocessing the collected raw experimental data; handling missing values ​​using median imputation, selecting representative values ​​for filling in missing values ​​based on the distribution characteristics of each parameter; outlier detection using the 3σ principle, calculating the mean and standard deviation of each parameter's numerical distribution, and removing outlier data points that deviate from the mean by more than three standard deviations; firstly, calculating the mean of each parameter's numerical distribution. and standard deviation : ； ； Where n represents the number of samples, x i Indicates the first One data point; Then eliminate those that satisfy the condition. Outlier data points were identified; for key parameters of the TRV peak, box plots were used for auxiliary judgment to remove extreme outliers exceeding 1.5 times the interquartile range; data normalization employed the Min-Max scaling method for each parameter. Calculate the minimum value of each in the entire dataset. and maximum value Then, all parameter values ​​are linearly transformed to Interval: 。 4. The method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system according to claim 1, characterized in that, Step 3 specifically involves: using the Random Forest algorithm combined with the SHAP value analysis framework to evaluate feature importance and quantify the impact of each experimental parameter on system performance; the specific configuration of the Random Forest model is as follows: it contains 200 decision trees, with a maximum depth limit of 15 layers for each decision tree, and uses Gini impurity as the node splitting criterion. The formula for calculating Gini impurity is: ； in Category in the node proportion, The total number of categories; calculated for each feature. The accurate feature importance score is obtained by averaging the reduction in impurity across all decision trees. : ； in Features In decision tree The reduction in impurities brought about by the process The total number of decision trees; Based on cooperative game theory, the contribution of each feature to the model output is assigned. The SHAP value is calculated by considering all possible subset combinations of features, measuring the marginal contribution of each feature to the model's predicted output when added to a specific subset. ； in For the complete set of features, For feature subset, For model prediction function, For factorial calculation, the final feature importance score A weighted average method is used to synthesize the random forest importance score and the absolute value of SHAP in a 6:4 ratio: 。 5. The method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system according to claim 1, characterized in that, Step 4 specifically involves using a combination of thresholding and ranking methods to filter key parameters. First, an importance score threshold is set. All parameters with importance scores above the threshold are retained as the candidate set of key parameters. ； Among them, X j Represents the candidate set of key parameters. This represents the final feature importance score.

6. The method for identifying key parameters of a high-voltage circuit breaker short-circuit current test system according to claim 1, characterized in that, Step 5 specifically involves: The logistic regression model adopts a form including L2 regularization, and its mathematical expression is: ； in This represents the probability of a successful interruption. to The five key parameters selected are: to These are the model coefficients.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method of claim 1.

8. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method of claim 1.

9. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method of claim 1.

Images