Intelligent Recognition of Internal Defect Signals in Composite Insulators Based on Variational Mode Decomposition and Random Forest Algorithm

As a crucial component of power transmission and distribution systems, composite insulators are susceptible to internal defects due to external factors such as manufacturing, transportation, and operating environments. The presence of these defects can lead to hazardous electrical incidents like partial discharges and insulation failures. Researchers have successfully integrated ultrasonic testing methods for detecting internal defects in composite insulators. However, manual defect identification is prone to subjective errors and is time-consuming and labor-intensive, heavily relying on expert experience. To address this issue, this paper proposes an intelligent recognition method for internal defects in composite insulators based on Variational Mode Decomposition (VMD) and the Random Forest algorithm. This method employs VMD to perform multimodal decomposition of ultrasonic signals and extract features of each component. Subsequently, the Random Forest algorithm is used to recognize these features, achieving the goal of classifying ultrasonic signals of internal defects in composite insulators. Experimental results show that the proposed method achieves a total recognition accuracy of 97% for defect types including no defects, voids, cracks, and delamination, significantly outperforming other algorithm.