为实现精确的风能出力预测，保障风力发电系统稳定并网，提出了一种基于随机森林模型和支持向量回归模型的精确风力发电功率预测算法。该算法以回归树和随机森林模型为基础，对风力发电影响因素进行特征重要性评估；然后基于特征筛选理论，构建最优特征集合；最后使用最优特征集合输入支持向量回归模型，实现风力发电功率的预测。实验结果表明，相比于单独使用随机森林模型，算法大幅提高了预测精度，平均绝对误差降低了19.67%；相比于长短时神经网络模型，算法在保持同样高精度的同时，大幅降低了模型复杂度以及所需的训练时间。算法能够实现精确的风力发电功率预测，有着较为重要的理论和实际意义。

To achieve accurate wind power output prediction and ensure the stable grid connection of wind power generation systems, an accurate wind power prediction algorithm based on the random forest model and support vector regression model is proposed. The algorithm is based on regression trees and random forest models to evaluate the importance of factors affecting wind power generation. Then, based on feature selection theory, an optimal feature set is constructed. Finally, the optimal feature set is input into the support vector regression model to predict wind power generation. Experimental results show that compared to using the random forest model alone, the algorithm significantly improves the prediction accuracy with a reduction of 19.67% in average absolute error. Compared to the long short-term memory neural network model, the algorithm achieves the same high accuracy while significantly reducing the model complexity and training time required. The algorithm can achieve accurate wind power prediction, which has important theoretical and practical significance.

国家电网科技项目(2017NR58337)