Time-Scaling Adptive Notch Filter Based Dynamic Phasor Estimator For Power Systems

The grid construction in China has gradually formed the national simultaneous cross-regional network, large area interconnected power system make the operation more economic, but at the same time dynamic process of interconnected systems become more complex. Wide-area Measurement System (WAMS) based on Synchronized Phasor Units (PMU) can effectively improve State Analysis and real-time monitoring of large interconnected power systems, which has a high requirement of synchronized phasor measurement in dynamic or transient conditions. But he current technology is not suitable in dynamic conditions of the power system, especially under conditions of low-frequency oscillation. In this paper the author pursuit to get a accurate and real-time dynamic phasor estimator based on Adaptive Notch Filter (ANF) to fit requirements of WAMS in power system dynamic or transient conditions.As the frequency changes with time in the dynamic conditions, the author use frequency tracking technology to reduce the non-synchronous sampling error, realized by the ANF based on low-pass filter. Firstly, the frequency tracking performance of ANF method is studied, with compared to adaptive Discrete Fourier Transform (DFT) method and a recursive wavelet method. Test results show that:frequency measurement of ANF has a higher accuracy than that of the other two methods in the presence of noise and harmonics, and the error brought by the impact of transient decaying DC component is a smaller and Shorter duration; ANF method is able to achieve accurate frequency tracking results in one or two cycle, which is rapider than that of recursive wavelet method, and not like DFT method produce the oscillation caused by frequency mutations. In addition, the author studies the multiple frequency tracking performance of parallel adaptive notch filters. Test results show that the multi-channel parallel notch filters can achieve accurate tracking of multiple frequencies, and no additional delay brought to the algorithm because the channels work in parallel.The ANF method can track time-varying frequency in the dynamic conditions, but the coupling between parameter update law and the filter itself make it difficult to realize estimation of amplitude and phase. Based on traditional ANF algorithm, the author analyzes a new time scaling adaptive notch filter (TSANF) based dynamic phasor estimator method. Firstly, the author provide time scaling to traditional ANF algorithm in continuous time, make appeal to new dynamic equations, which is equivalent to linear time-invariant systems, and then time discrete is made. The new time scaling ANF remove the coupling between frequency parameter and the filter itself, make it easy for dynamic phasor estimator. Test and simulation are made in dynamic (transient) conditions which contain amplitude or frequency change of the signal, with compared to the adaptive DFT method and recursive wavelet method. The results show that in conditions of amplitude mutations or the existence of decaying DC, the TSANF method will produce a phasor estimator error in the initial transient, which gradually disappear with the weakening the transient process, the caused error is relatively smaller and shorter duration compared to the other 2 methods; in conditions of mutation or linear change for frequency, due to tracking results slightly lag the true frequency, thus the TSANF method will produce a phasor estimator error in the dynamic (transient) process, but the estimator error is less than the other 2 methods.Finally, the author analyzes the applicability of TSANF based dynamic phasor estimator under the conditions of low-frequency oscillations. While amplitude and frequency of electrical signals are changed, not like the signal model of traditional definition, the author makes a definition of dynamic signal under low-frequency oscillation, and uses TSANF based dynamic phasor estimator to match dynamic signal under the conditions of single and multi-mode low frequency oscillation. By the way, multiple parallel TSANF is applied for multi-mode low frequency oscillation singles. Then test and simulation is made. The results show that the TSANF method can receive more accurate results compared to adaptive DFT method in simple single mode low-frequency oscillation; in the condition of low-frequency oscillations and then short-circuit fault happened, due to effects of mutations in the transient, the TSANF method will generate a frequency mutation error in the initial when fault occurs, but the error exists a very short time, which lead to a relatively small effect on phasor estimator; in the condition of multi-mode low frequency oscillation, results show that the multiple parallel TSANF can extract to low-frequency envelope be added to the of sine wave, so it can receive dynamic phasor in multi-mode low frequency oscillation, although estimator error will increase when frequency range a relatively large change in short time.