Research on Prediction of Wind Power Generation Based on K-means Clustering and Machine Learning

In order to further study the relationship between wind power generation and multiple influencing factors, this paper proposes a combined regression model for wind power generation based on K-means clustering, BP neural network and random forest algorithm. Firstly, this paper uses K-means clustering algorithm to divide the data set into two categories. Then, BP neural network and random forest were used to establish regression models for the two types of data, and grid search-K-fold cross-validation method was used to optimize the relevant parameters in the models. Finally, the average value of correlation indicators such as the determination coefficient $(R^{2})$ of BP neural network and random forest is calculated as the result of model evaluation. In this paper, 18,995 sets of wind power generation data from China’s Hulun Lake Wind Farm, Ulanqab Wind Farm and Fengshagang Wind Farm in the past three years are taken as examples to verify that the combined regression model proposed in this paper can provide more accurate and reliable prediction, and offer valuable perspectives for future research and practical applications in the realm of wind power generation.