Power quality disturbances analysis based on EDMRA method

An Energy Difference of Multiresolution Analysis (EDMRA) method for power quality (PQ) disturbances analysis has been proposed in this paper. At each wavelet decomposition level, the squared value of the detail information is calculated as their energy to construct the feature vector for analysis. Following the criteria proposed in this paper, different kinds of power quality disturbances can be detected, localized, and classified effectively. The choice of the decomposition levels of appropriate wavelets are of the critical importance for the EDMRA method, since they will influence quality of the reconstructed signal as well as the computational cost. It is presented in this paper that the Minimum Decomposition Level (MDL) is related to the sampling frequency by the proposed function. The comparison study among different kinds of wavelets for the EDMRA method is presented in details. The EDMRA method is scalable and has robustness characteristics in common design paradigm. It can be realized economically using wavelets with shortest length, such as Harr, Db2, Sym2 or Coif 1. Two types of noise, namely, Gaussian white noise and band limited spectrum noise are considered in this paper to show the effectiveness of the proposed method in noisy environment.