An Adaptive Approach Based on KPCA and SVM for Real-Time Fault Diagnosis of HVCBs

• Published in: IEEE transactions on power delivery Vol. 26; no. 3; pp. 1960 - 1971
• Main Authors: Jianjun Ni, Chuanbiao Zhang, Yang, S. X.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01-07-2011; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptation model; Applied sciences; Connection and protection apparatus; Data models; Electrical engineering. Electrical power engineering; Electrical power engineering; Exact sciences and technology; Fault detection; Fault diagnosis; high-voltage circuit breakers (HVCBs); Kernel; kernel principal component analysis (KPCA); Mathematical models; Miscellaneous; Power networks and lines; Real time; Real time systems; Recognition; Samples; Similarity; Statistical methods; Studies; support vector machine (SVM); Support vector machines; Testing. Reliability. Quality control; support vector machine (SVM); Updating; Experimental study; Vector method; Algorithm; Adaptive method; Kernel method; Circuit breaker; Fault diagnosis; High voltage; Electrical network; high-voltage circuit breakers (HVCBs); kernel principal component analysis (KPCA); Diagnostic aid; Fault diagnostic; Principal component analysis; Defect detection
• ISSN: 0885-8977; 1937-4208
• DOI: 10.1109/TPWRD.2011.2136441

High-voltage circuit breakers (HVCBs) play an important role in power systems, which can control and ensure the power grids are working properly. Real-time fault diagnosis of HVCBs is an essential issue for power systems. In this paper, a novel approach based on an adaptive kernel principal component analysis (KPCA) and support vector machine (SVM) is proposed for real-time fault diagnosis of HVCBs. In the proposed approach, a sample reduction algorithm based on a similarity degree function is proposed to analyze the similarity between the samples, and the redundant data can be eliminated. An adaptive KPCA method is used for the fault detection of HVCBs based on squared prediction error statistics. An SVM is used to carry out the fault recognition. Two spare data areas are set up for fault detection and recognition modeling. The data in the spare date areas are updated continuously, and the detection and recognition models are updated subsequently to improve the adaptivity of the diagnosis models and reduce the diagnosis error. The proposed approach can deal with various situations of the fault diagnosis for HVCBs. The experimental results show that the proposed approach is capable of detecting and recognizing the faults efficiently.