Research On Harmonic Analysis Approach Of Power System Based On Neural Network

With the rapid development of industrialization,various power devices have been put into use in the power system,and there are more and more non-linear loads and impact loads in the power system.A large number of non-linear loads and impulsive loads bring huge harmonic pollution to the power grid and reduce the power quality of the power grid.For the treatment of harmonic pollution,an important prerequisite is the accurate detection of harmonic parameters,so the detection and analysis of harmonic is of great significance.Based on neural network,this thesis discusses the detection algorithm of harmonics and interharmonics.Based on neural network,this thesis discusses the detection algorithm of harmonics and interharmonics in power system,focusing on the following research:Aiming at the problems that the traditional neural network harmonic detection algorithm is prone to oscillation during network learning,and the accuracy of fundamental frequency largely affects the detection accuracy of harmonic amplitude and phase angle,a harmonic detection method based on quasi synchronous sampling algorithm and neural network is proposed.Based on the traditional algorithm,the optimization method is used to optimize the step size,and the quasi synchronous sampling algorithm is introduced to quickly obtain the accurate fundamental frequency.Simulation results show that the harmonic detection method based on quasi synchronous sampling algorithm and neural network has high accuracy,and the iteration speed has been significantly improved.Aiming at the problems of traditional neural network interharmonic detection algorithm,such as convergence speed,detection stability and accuracy,an improved neural network and root music interharmonic detection method are proposed.Based on the traditional interharmonic analysis algorithm based on neural network and root music,HQ criterion is used to estimate the number of sources.At the same time,in order to speed up the iteration speed and avoid falling into local extreme value,least square method and Gauss Newton method are used to modify the later weight of neural network and the excitation parameters of hidden layer respectively.Simulation results show that the detection accuracy and iteration speed of the improved algorithm are improved.Aiming at the problem of low detection accuracy of harmonics and interharmonics in high-intensity noise environment,a neural network interharmonic detection algorithm based on wavelet denoising is adopted.In this method,firstly,the noisy signal is de noised by threshold method,and then the de noised signal is processed by interharmonic detection algorithm.Simulation results show that this method can effectively improve the detection accuracy of harmonics and interharmonics in high noise environment.