An Identification Method of Power Quality Disturbances Using Kalman Filter and Extreme Learning Machine

To make further improvement on power quality issues, a novel power quality disturbances(PQD) classification algorithm is proposed in this paper, which is a combination technique of the Kalman Filter(KF) and the Extreme Learning Machine(ELM). KF tracks and estimates the amplitude of single or multiple disturbance signals, and two effective features are extracted from signal waveform. ELM is applied to classify the power quality events corresponding categories. Besides, Back Propagation Neural Network(BPNN) is used as a compared classification method. The superior performance of proposed method is proved by simulation. The simulation results show that the average classification accuracy of KF-ELM is 97.75%, which increases by 7.5% compared with KF-BPNN.