正则化自适应匹配追踪电能质量数据重构方法

Novel reconstruction method of power quality data based on regularized adaptive matching pursuit algorithm

为了有效改善传统方法应用在电能质量信号的采集和压缩方面所面临的资源浪费以及重构性能较差等问题,提出了一种基于正则化自适应匹配追踪的电能质量数据重构方法,此方法利用压缩感知理论,对电能质量信号进行采样和压缩并行处理。首先,对感知矩阵中的原子进行一次挑选并且计算相关系数,将挑选出的原子索引值存入至候选集中。然后,在电能质量信号稀疏度K不作为先验条件的前提下,对候选集中原子的数目进行自适应地调节,并运用正则化的处理过程完成支撑集的二次挑选,用步长逐步逼近信号的稀疏度进而准确重构出电能质量原始信号。仿真实验结果表明,信号的重构精度高于98.2%,并且能够保存原始电能质量信号大部分能量,重构信噪比高,均方误差小。

In order to solve the problems of resource wasting and low reconstruction performance that traditional methods face when used in the acquisition and compression of power quality signals. This paper proposes a reconstruction method of power quality data based on regularized adaptive matching pursuit algorithm. This method uses compressive sensing theory to conduct the sampling, compression and processing of the power quality signals. Firstly, the atoms in the perception matrix are selected first time and the correlation coefficients are calculated. The selected index values of the atoms are then kept in the candidate set. Then, under the precondition that the sparsity K of the power quality signals is not used as the prior condition, this method adaptively adjusts the number of the atoms in the candidate set and realizes the second time selection of the support set with regularization processing. The step size is used to gradually approach the proper sparsity K of the signals; and then, the original power quality signals can be reconstructed precisely. Simulation experiment re- sults show that the reconstruction accuracy of the signals is higher than 98.2% and the most energy of the original power quality signals can be preserved. Also, the signal to noise ratio of the reconstructed signals is high and the mean square error is small.