Research On Power Quality Data Reconstruction Method Based On Compressive Sensing

With the rapid development of science and technology and the continuous expansion of the scale of power system, a large number of new power electronic equipment was put into power grid. It brings a lot of power quality problems. Therefore, it is very important to take effective measures to carry out real-time monitoring of power quality and the analysis of the power quality problems in a timely manner. Sampling and compression of power quality data is the premise of its analysis and processing. The traditional power quality data sampling and compression methods are based on the Shannon sampling theorem. It points out that the sampling frequency of the power quality disturbance signals must be more than two times higher than that of the original signals’ frequency. It leads to a large amount of data and brings a lot of pressure to the storage and transmission of data. The emergence of compressive sensing theory effectively solves such problems. Signals are sampled at a value which is far below the Nyquist sampling frequency. The sampled data are what we need to deal with. Finally, the original signals are reconstructed by applying some suitable reconstruction algorithm.In this paper, firstly, the basic theory of compressive sensing theory and the three important aspects of it are discussed. Secondly, according to the definition of power quality and the international standard, different kinds of power quality signal models are constructed in MATLAB. According to the characteristics of different power quality signals, the feasibility of power quality signal analysis and processing based on compressive sensing theory is analyzed.This paper focuses on the reconstruction of power quality data. Based on the compressive sensing theory, two kinds of power quality data reconstruction methods are proposed. They are regularized adaptive matching pursuit(RAMP) reconstruction method and backtracking-based adaptive orthogonal matching pursuit(BAOMP) reconstruction method. The sparsity of most power quality signals is unknown. The sparsity adaptive ability of these two methods is just able to solve this problem. The final simulation results show that the RAMP method is applied to the reconstruction of most power quality signals and the reconstruction performance index basically meet the requirements of power quality signal analysis. But for the transient oscillation signal and voltage pulse signal, the reconstruction performance index of the RAMP method is not good. The BAOMP method uses its backtracking characteristic to effectively improve the reconstruction effect of the above mentioned two signals. Comparison of simulation results show that the RAMP and BAOMP power quality data reconstruction method which are based on compressive sensing theory are superior to the traditional greedy reconstruction methods on reconstruction rate, reconstructed signal to noise ratio(SNR) and mean square error(MSE) aspects. In the field of power quality data analysis and processing, they reflect a certain application value.