Research on the Prediction Method of MOV Deterioration State Based on Principal Component Analysis and Grid Search- Optimized Support Vector Machine Regression Algorithm

Abstract

Abstract Background: Accurate diagnosis of faults in metal oxide varistors (MOV) is crucial for the safe operation of power systems, and the deterioration of MOV under continuous pulse impacts can be more severe. To effectively improve the fault diagnosis rate, this paper proposes a fault diagnosis algorithm based on Principal Component Analysis (PCA) and Grid Search-optimized Support Vector Regression (GS-SVR).Objective, The objective of this study is to propose an effective fault diagnosis algorithm that accurately predicts the fault state of MOV under single and continuous pulse impacts, while reducing the correlation between indication indicators through dimensionality reduction. Method: The proposed experiment involves conducting a comparative test on MOV with different time intervals between impacts, on the order of 10 seconds. The data collected from this experiment, with a time resolution of 10 seconds, will be subjected to dimensionality reduction using PCA to reduce the correlation between the original indicators. Finally, the GS-SVR model will be employed to analyze and predict the effects of single and continuous pulse impacts on MOV. Results: Experimental results demonstrate that the GS-SVR model achieves a mean square error of less than 0.00057 in predicting single pulse impacts and still exhibits certain effectiveness for irregular pulse impacts, such as continuous pulses. Conclusion: The proposed fault diagnosis algorithm based on PCA and GS-SVR can effectively improve the fault diagnosis rate of MOV, and accurately predict the fault state of MOV under single and continuous impulse shock. This is of great significance to the safe operation of power system.