Automatic switching control of transformer simulated fault resistor network under integrated signal processing

Abstract

Using a simple threshold judgment method, only coarse switching based on fixed resistance values is possible, resulting in low efficiency in switching control of transformer simulated fault resistance networks. Therefore, this paper studies automatic switching control of transformer simulated fault resistance networks under integrated signal processing. A multi-channel signal acquisition module is used to collect core signals such as three-phase voltage, current, and temperature of the transformer. After wavelet threshold denoising, feature extraction, and normalization, standardized feature vectors are obtained. Based on an improved RBF kernel function-based SVM model, trained with a fault sample dataset, typical transformer faults are accurately identified. According to the fault type, a dynamic resistance parameter calculation model is constructed using equivalent circuits, incorporating correction terms such as temperature and loss to calculate the fault resistance value. Through a switchable resistor array and closed-loop control strategy, the actuator is driven to complete the resistance network switching, achieving automatic switching control. Experimental results demonstrate that when the fault severity reaches 75%, the switching success rates of the four typical fault resistor networks all exceed 95%. Compared with the control method, the response time is reduced by up to 18 ms when the number of resistors is 23, significantly enhancing the efficiency of the transformer simulated fault resistor network switching control.