Application Of Adaptive Kalman Filter Algorithm In Power Quality Disturbance Detection

The power quality problem undermines the safety and economy of power grid,as well as endangers the devices,which leads to a vast cost for power users.Hence,accurate estimation of power quality events are the premises of effective process against these events.The thesis focuses on a further study on the detection of power quality disturbances quickly and accurately.In this paper,two kinds of adaptive Kalman filter algorithms are proposed for power quality detection.Kalman filter algorithm is the development of Wiener filter,which solves the problem of state estimation in non-stationary environment and is widely used in power systems.However,the statistical characteristics of the process noise directly affects its estimation precision and the convergence speed,Therefore,to improve its performance,two adaptive methods are proposed for estimating the statistical characteristics of the process noise and the measure noise.The main research works and innovations are summarized as follows.Aiming at the problem that the statistical characteristics of traditional Kalman filter noise are difficult to determine,an adaptive Kalman filter based on maximum likelihood(KF-ML)is proposed to detect power quality disturbances.The method can optimize the noise covariance matrix and the initial condition parameters by maximum likelihood,which overcomes the problem that the traditional Kalman filter algorithm is difficult to determine the noise statistical characteristics(noise covariance matrix).When the method is used for power signal estimation,two kinds of state-space models are established according to two different state vectors.The two models are tested with the KF-ML method in different cases,such as harmonic interference,pulse interference and different signal-to-noise ratios.Simulation results of comparison between the two models are analyzed in detail.To improve the accuracy and reliability of Power quality disturbance detection in a noisy environment,an adaptive process noise covariance Kalman filter(APNCKF)is proposed for tracking the power quality disturbances.The method updates the process noise covariance matrix of the Kalman filter by maximizing the evidence density function,which enhances the stability of the filtering system.In this proposed method,the identified process noise parameters are most sensitive to signal singularity,which improves detection accuracy.Also,the method can obtain rich parameter information(amplitude,phase,etc.),which is used to detect single power quality disturbance and composite power quality disturbance.The comparison experiments with wavelet transform,Kalman filter and KF-ML are carried out.The simulation results verify the superiority of its performance.The adaptive filtering algorithm overcomes the problem that the noise statistics of traditional Kalman filtering algorithm is difficult to determine.The simulation results show that the two improved algorithms improve the accuracy of disturbance detection and provide a reliable basis for power system power quality management.