Dynamic Harmonic Estimation Based On Optimized Unscented Kalman Filter Model

With the development of power electronics technology, due to the increasing use of nonlinear loads in power systems, periodical distortions in current and voltage waveforms become more, harmonic injection has been a growing power quality concern. The presence of harmonic distortion in power lines results in greater power losses in the distribution systems, sometimes, in operation failures of electronics equiments, especially existence of harmonics in power system could cause serious problems such as harmonic resonace in partial circuit, overvotaltage, harmonic distortion,even endanger the safety of users. However, load sensings are more and more, the use of PLC, computers and precise instruments in power systems that is heavily based on power-electronic converters also contributes to the growing concern for better estimation to ensuer power quality. Hence, to develop techniques to remove unwanted harmonic distortion in power systems is necessary. Timely and accurate detection for harmonic signal, reducing the harmonic caused by misoperation of relay protection and automatic device, so as to improve the efficiency of the power equipment, reduce the electricity cost.The paper proposes four kings of typical method of power quality estimation: Root Mean Square, FFT and Wavelet Packet Transform and Adaptive Recursive Least Squares, and various simulation results of the above algorithms. Then, Elaborated the basic principle of Kalman filter and Unscented Kalman filter, and various simulation results of them. Due to the traditional Unscented Kalman filter method in which the above-mentioned two kinds of covariance are taken as constants. We propose a particle swarm optimized unscented Kalman filter(PSOUKF) method to estimate the power system dynamic harmonics. By using the improved particle swarm optimization algorithm with species classification and dynamic learning factor, we optimize the state noise covariance and the measurement noise covariance of the unscented Kalman filter(UKF) so as to sufficiently take the impacts of power system noise on dynamic harmonic estimation into account. The proposed method overcomes the deficiency of low dynamic harmonic estimation accuracy in the traditional UKF method in which the above-mentioned two kinds of covariance are taken as constants. Simulation results show that the proposed PSOUKF is more effective than Kalman filter(KF) and UKF, and PSOUKF can improve the dynamic harmonic estimation accuracy without increasing the computational complexity.