The short-term photovoltaic power generation forecasting based on Fruit Fly Optimization Algorithm optimized BP neural network

Accurately predicting photovoltaic power generation is crucial for ensuring the safe operation of power grids and advancing solar energy development and utilization. For the issue of large errors in current prediction methods, this paper introduces a forecasting method for photovoltaic power generation. The method utilizes Fruit Fly Optimization Algorithm (FOA) to optimize Back Propagation (BP) neural network. This involves analyzing the correlations among factors impacting photovoltaic power generation and selecting relevant meteorological data via the Pearson correlation coefficient method. Fruit Fly Algorithm demonstrates rapid convergence, a minimal parameter set and ease of adjustment, rendering it applicable across various domains. Employing Fruit Fly Algorithm to optimize weights and thresholds within BP neural network leads to the final prediction outcomes. Simulation results confirm the superior prediction accuracy of the FOA-BP model for photovoltaic power generation, particularly during spring, autumn and winter, showcasing its practical utility.