Optimal Prediction Intervals of Wind Power Generation Based on FA-ELM

The process of wind power generation has great uncertainty, which is one of the important sources of adverse impact on the stable operation of power grid. In fact, the traditional point prediction ignores this adverse effect, and the probability and fluctuation range of the predicted value is failed to be obtained. This paper presents a probability interval prediction method based on firefly algorithm (FA) and extreme learning machine (ELM). Firstly, elm network is established to generate the prediction interval, and then the coverage probability and sharpness of the generated interval are optimized by seeking the optimal output weight. Finally, the prediction interval of the set confidence level is obtained. Experiments show that compared with other single prediction algorithms, this hybrid prediction algorithm can improve the prediction accuracy and reduce the prediction risk. At the same time, the experimental data verify the feasibility and effectiveness of the method, which shows that the method has a certain practical application value in power system.