Dynamic Harmonic Estimation Of Power Grid Based On Unscented Particle Filtering

Due to the rapid development of power electronics technology,electricity has become the main power source in people's daily life.When electricity provides us convenience, it also accompanies a number of power quality problems at the same time. With the widespread use of nonlinear electronic components, power grid will have a voltage distortion, a large number of harmonics is pulled into the system in the case of the of different impact load in power system, which reduce the efficiency of power equipment, harm the stability of the power system and also cause great trouble for the users. Impact caused by the power quality problems can influence the normal operation of power equipment and brings inconvenience to people's life and production. To the maximum extent, it will lead to unpredictable disaster.In the situation that increasingly sophisticated electrical equipment is used, people pay more attention on the power quality issues and do more depth research of power quality, therefore growing number of methods are used in the field of detection of power quality. Only accurately and timely detecting the parameters of harmonic followed by the timely measures taken, can the impact of harmonics problems be minimized. Therefore, finding the method to quickly and accurately estimate the harmonic parameters of power grid becomes necessary.The paper firstly proposes four kinds of typical methods for power quality detection, RMS, FT, WT, LMS and KF and makes simulation for each method. Then basic principles of unscented Kalman filter(UKF), particle filter(PF) and unscented particle filter(UPF) are illustrated. Last,a method for estimating power system dynamic harmonics is proposed based on unscented particle filter(UPF) algorithm in this paper. Firstly, the UKF algorithm is adopted to estimate the values of state variables of power system dynamic harmonics and determine the covariance of them. Then the obtained results are utilized to generate the importance density function of conventional particle filter algorithm. At last, the optimal estimation of power system dynamic harmonics is achieved using particle filter algorithm. The proposed method overcomes not only the limitation of Gaussian distribution of noise needed in unscented Kalman filter(UKF), but also the drawback that the traditional particle filter(PF)is easy to degenerate. In addition, it retains the good ability of UKF in processing nonlinearity and strong ability of PF in anti-interference.Simulation results show that the results(dynamic harmonic amplitude and phase) obtained by the proposed method are closer to the real values for both cases with Gaussian noise and non-Gaussian noise. Compared to the methods including Kalman filter(KF) and UKF, The proposed method is much more accurate.