Day-Ahead Wind Power Prediction Based on BP Neural Network Optimized by Improved Sparrow Search Algorithm

Accurate day-ahead wind power prediction is of valuable reference for the dispatching department to make the next-day power generation plan. Based on the measured data and numerical weather prediction (NWP) data collected from a certain wind farm in Hunan province, this paper uses an Improved Sparrow Search Algorithm (ISSA) for optimizing initial weights and thresholds of the BP neural network to predict the day-ahead wind power which meets the related department's requirements. Logistic chaos mapping and Levy flight strategy are integrated into traditional Sparrow Search Algorithm (SSA) to further improve the diversity of initial weights and thresholds, help sparrows to jump out of local optimum, enhance the ability for global space exploration of the algorithm, improve the accuracy of prediction results ultimately. Finally, compared with the simulation results of BP Neural Network and SSA-BP neural network, the results of ISSA-BP model are better.