Research on Neural Network MPPT Algorithm Based on DE and Dichotomy

2019 4th Asia Conference on Power and Electrical Engineering

能源学;电工学

Li, Xiaojiao^1 ; Qi, Xuanxuan^1

School of Electrical Engineering, Xi'An Jiaotong University, Xi'an, China^1

关键词: Global optimal solutions;    Gradient Descent method;    Improved differential evolutions;    Local optimal solution;    Maximum power point;    Maximum Power Point Tracking;    Oscillation phenomenon;    Photovoltaic power generation systems;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/486/1/012111/pdf
DOI  :  10.1088/1757-899X/486/1/012111

来源: IOP

【 摘 要 】

Aiming at the shortcomings of traditional MPPT method, such as slow tracking speed and oscillation at maximum power point, this paper combines neural network with dichotomy to propose a new maximum power point tracking method for photovoltaic power generation system. And the traditional neural network uses the gradient descent method to solve the problem that the parameters are easy to enter the local optimal solution. In this paper, the improved differential evolution method is used to solve the global optimal solution. The neural network is used to track the vicinity of the maximum power point, and then the dichotomy is used to further approach the maximum power point. The simulation results show that compared with the traditional MPPT method, BP neural network and dichotomy can track the maximum power point faster, avoid the oscillation phenomenon, and have faster tracking speed and higher tracking accuracy.