Improved Back Propagation Neural Network Algorithm for Photovoltaic Maximum Power Point Tracking Control Research

Aiming at the problems of slow tracking speed and poor tracking accuracy of the current maximum power point tracking (MPPT) technology for photovoltaic arrays, this paper proposes a photovoltaic MPPT algorithm based on improved quantum particle swarm optimization optimized back propagation (IQPSO-BP) neural network. The QPSO algorithm using adaptively varying contraction-expansion (CE) coefficient endows BP neural networks with optimal weights and thresholds, which solves the defects of traditional BP neural networks with random initial weights and thresholds that cause the prediction results to fall into local optima.This improves the convergence speed of traditional BP neural networks and reduces the prediction error. The simulation results show that the prediction error of the IQPSO-BP MPPT algorithm is significantly reduced compared to the traditional perturb and observe(P&O) MPPT algorithms, and the tracking speed is greatly improved, which can track the global maximum power more quickly and accurately when the solar irradiance is changed.