Research Method of Ultra-short-term Wind Power Prediction Based on PSO-GRU Prediction

Wind power is an important source of electricity for the national grid. The unstable characteristics of wind make scheduling and decision-making problems for power companies. Therefore, it is necessary to improve the accuracy of predicted wind power. To solve this problem, this paper adopts the particle swarm optimization algorithm (PSO) to optimize the GRU neural network method, and selects the optimal combination of GRU hyperparameters through PSO to determine the most suitable network topology. The experimental experiment in this paper uses the measured power data of a wind farm in Inner Mongolia as the data set. And use root mean square error and mean absolute error as evaluation criteria. The experimental results show that the algorithm proposed in this paper achieves better experimental results in power prediction, and achieves higher prediction accuracy compared with BPNN, SVR and other models. It proves that the model can achieve good results in wind power prediction. It has practical application value.