Analysis Of Power Quality Disturbance Signals Using Atomic Decomposition

To improve the power quality, first and foremost is the ability to analyze vast amount information obtained from real-time monitoring of the grid, and correctly detect, compress, extract the feature as well as classify the data. In this paper, an atomic decomposition based method is proposed to meet the feature extraction and compression requirement of power quality disturbance signal, the presented method is applied to variety of common power quality disturbance for analysis feature extraction and combined with wavelet algorithm for data compression.In this paper, a variety of power quality analysis methods are summarized, the power quality problem and related theories are introduced. Based on those work, the atomic decomposition based power quality disturbance data analysis method is proposed.Then, in-depth study of atomic decomposition method is preformed, pointing out atomic decomposition's superiority over Fourier Transform and wavelet analysis. Three important concepts of atomic decomposition, including Gabor over-complete dictionary and Matching Pursuits, are introduced in detail.As to the realization of the algorithm, by considering the characteristics of power quality disturbance signals, the damped sinusoids model is given and the transformation between Gabor atom and damped sinusoids is realized as well as the parameter optimization. In addition, the termination condition of the algorithm has been improved by using the initial residual energy as the threshold for the termination. This method is applied for feature extraction of power quality disturbance signal, it can precisely extract the starting and ending moment of disturbances as well as their magnitude, frequency and other features. Thus, the analytic representation of the various components is obtained.Finally, this paper introduces an atomic decomposition and wavelet combined data compression method. A small-scale dictionary of atoms was established according to the priori knowledge of power quality disturbance signal, the main components are extracted by suing atomic decomposition which lead to a high compression ratio. Then, the remaining irregular part was compressed with wavelet transform.The proposed algorithm is programmed by MATLAB and applied to various PQ disturbances analysis. Simulation results confirm its effectiveness and reliability.