Research On Relay Protection Algorithm For Transmission Lines Base On Kalman Filter

Microcomputer relay protection, whose core technology is signal processing, has been widely used in power system. Fourier algorithms were utilized in relay protection equippments to extract the fundamental components of the measurement signals. However, the Fourier methods can not reflect signals’ time-domain characteristics, and they have weak ability to process the non-stationary signals with local characteristics. Moreover, the decaying DC components and harmonic components have great impact on the performance of Fourier methods. These drawbacks reduce the speeds and accuracies of the traditional algorithms. Therefore, the study of new microcomputer relay protection algorithm based on signal processing technology is of great significance. Based on the classical signal processing methods, including Kalman filter, unscented Kalman filter (UKF), wavelet transform, mathematical morphology (MM) filter, new microcomputer realy protection algorithms for transmission line were studied in this paper.Firstly, two microcomputer protection algorithms including wavelet-Kalman filter and morphology-Kalman filter are presented in this paper. These algorithems adopt the wavalet transform or mathematical morphological filter to decompose the measurment signal into smooth signals and detailed signals in multi-sacle. The smooth signals are obtained by filtering out the noise of the measured signals and the detailed signals contain transient noises of the measured signals. Smooth signals are applied to update the observations of Kalman filter to improve its convergence rate, and detailed signals are employed to calculate the statistical characteristics of the measurement noises to improve the convergence precision of the filter. Kalman filter is then utilized to process the signal, which has already filter the transient noise, to estiamte the fundamental component of the signal in real-time.Since the microcomputer protection algorithm based on linear Kalman filter is sensitive to the frequency offset of measured signals, new algorithm based on UKF is presented. The algorithm establishes a nonlinear mathematical model which takes the frequency and DC offset of fault signal into account. UKF is then adopted to estimate the frequency, DC offset, and the fundamental component of the signals using the nonlinear model. Finally, we combine the MM filter and UKF and propose the morphology-UKF filter based realy protection alorithm. Study cases are carried out in Matlab/Simulink, and the simulation results demostrate the effectiveness of the proposed algorithms.