The Analysis Of The Power Quality Disturbance Based On Optimizated Atom Decomposition Algorithm And RBF Neural Network

In recent years, a large number of nonlinear loads are connected to the electricity grid and there are many other disturbance sources, which further deteriorates the quality of power supply. At the same time, due to the wide application of precision instrument and intelligent equipment, more stringent requirements of power quality are put forward. Therefore, research and analysis of power quality disturbance signal is very important. This paper uses the research hotspot in the field of signal processing in recent years, atomic decomposition algorithm, to analyze all kinds of power quality disturbance signal.In this paper, firstly, reasonable discrete processing of the Gabor atomic library is put forward to solve the problem of its practical size. Secondly, aiming at the over-matching phenomenon and non-orthogonal projection problem of matching pursuit(MP) algorithm, orthogonal matching pursuit(OMP) algorithm is proposed, which can make the selected atom orthogonalized to improve the convergence of the algorithm. Thirdly, in allusion to the problem of the large amount of calculation, particle swarm optimization(PSO) algorithm is presented for the optimization. And write the program of PSO-MP and PSO-OMP, which all based on Gabor library. Take four kinds of disturbance signal for example to analyze signal reconstruction performance and convergence of the algorithm. The simulation results verified PSO-OMP algorithm has better performance.PSO-OMP de-noising method is carried out the research because of the existence of certain noise in power quality disturbance. Threshold decision method was proposed through experiment means, and it is concluded that the selection of threshold is associated with the length of the signal. Using PSO-OMP algorithm based on Gabor library for common disturbance signal denoising, and compared with wavelet de-noising. The simulation results show that the proposed method can effectively separate the noise and signal, achieve good de-noising effect.The PSO-OMP algorithm combined with RBF network is used as the classification of power quality disturbance, Firstly, fundamental wave is separated from the signal to be analysed by using the fundamental atomic library. Then, use atomic decomposition optimization algorithm to extract the characteristics of disturbance signal as the input of RBF network. The simulation examples indicate under the condition of different SNR, this article proposed method has good recognition effect on single disturbance and multiple disturbance because of its anti-noise performance.