Research on Transient Power Quality Disturbance Classification Based on Wavelet Hidden Markov Tree and Support Vector Machine

Abstract

Aiming at the problem that the modes of transient power quality disturbance are complex and difficult to classify, a new model based on Wavelet Hidden Markov Tree (WHMT) and Support Vector Machine is proposed. Taking into account the inherent properties of wavelet transform, WHMT has the merit of using a probabilistic model to capture key characteristics of considered signal, and achieves significant denoise effect at the same time. Expectation maximization algorithm is utilized for fitting the WHMT to observational data. Then, denoising is accomplished, and Support Vector Machine is utilized to classify the extracted features. Simulation experiments under different noisy environments are designed to verify the performance. Results show that this method achieves high recognition accuracy and strong anti-noise ability with fewer features.