Research On Detection And Compression Of Power Quality Disturbance Based On The Atoms Sparse Decomposition

With the continuous development of power grid construction and load diversification, power quality problems have been increasingly prominent, which makes analysis and improvement of power quality become an important research topic. Atomic sparse decomposition can realize the adaptive decomposition of signals by constructing an over complete atomic library, and has received extensive attention in signal analysis of power system. Based on the idea of atomic sparse decomposition, power quality disturbance analysis, data compression, system harmonic detection are researched in this paper.A large number of atoms in complete atomic librar y and the matching pursuit algorithm needing to traverse the entire atomic library for finding the optimal solution, result in massive calculation. Intelligent optimal matching pursuit algorithms are selected to reduce the algorithmic complexity. However, the extraction accuracy is obviously decreased. And considering the short-sighted and error accumulation effect of matching pursuit algorithm, parameters-extracting inaccurate problem and irrelevant components problem will occur, interfering the accurate detection of signals. An improved PSO dynamic search(particle swarm optimization dynamic search, PSO-DS) algorithm is proposed to improve these problems mentioned above, which according to prior information provided by fast Fourier transform and wavelet transform to optimize the search ranges of parameters and the search method of best atomics. Simulation analysis shows that the improved algorithm can effectively extract the features of signals by doing several times of decomposition, and avoid the irrelevant components, realizing the improvement of detection accuracy and the simplicity and accuracy of signal expression.With the intelligent and information construction of power grid, a large number of power quality data bring great difficulty for data storage and transmission effectively. Atoms sparse decomposition can reconstruct signals by decomposition parameters and atomic library. Compared with the original signals, the decomposition parameters are very small, and could achieve a high compression ratio. This paper further studies the compression and reconstruction performance of power quality disturbances based on the PSO-DS algorithm. And the compression and reconstruction performance are evaluated according to the compression rate, signal to noise ratio, the mean square error percentage and energy recovery coefficient. Simulation analysis shows that a high to 98% for compression rate, more than 99% for energy recovery coefficient, 39 dB for the signal to noise ratio, could be achieved with the PSO-DS algorithm. When compression rate is 95%, the signal to noise ratio can reach 55 d B, and there is 15%~20% higher than wavelet packet compression algorithm when the noise ratio is constant.Finally, this paper studies the application of harmonic detection based on the atomic sparse decomposition method. Four aspects including atomic library construction, algorithm realization, algorithmic complexity and detection performance are analyzed. Besides, comparison analysis between the algorithm and the harmonic detection method based on instantaneous reative power has been done, which verifies the effectiveness and superiority of atomic sparse decomposition in harmonic detection.