Predicting Short-Term Photovoltaic Power Station Output Using Combination of Fruit Fly Optimization Algorithm and an Improved Adaptive Genetic Algorithm

In the process of short-term output prediction for photovoltaic power stations, it is difficult to avoid various errors. The existence of these errors severely constrains the research and development of short-term output prediction systems for photovoltaic power stations. With the increasing penetration rate of photovoltaic power stations, these errors pose a significant threat to the stable operation of the power system. Accurate prediction of photovoltaic power output can effectively reduce the impact of large-scale photovoltaic integration on the power grid and improve the grid's capacity for accommodating photovoltaic power. Furthermore, it allows for the optimal utilization of solar energy resources, leading to higher social and economic benefits. The output power of a photovoltaic power station fluctuates due to various factors. Improving the accuracy of photovoltaic system output prediction can effectively reduce the impact on the power grid after the photovoltaic system is connected, thereby enhancing the stability of the power system. In this paper, a prediction model for photovoltaic power based on a combination of the fruit fly algorithm (FOA) and adaptive genetic algorithm (AGA) optimized BP neural network is established. The results show that the BP neural network model with the combined optimization algorithm can effectively avoid the drawback of the BP neural network easily falling into local minimum points. Compared to the BP neural network model that only optimizes weights and thresholds, it improves prediction accuracy and demonstrates certain practical value.