基于相关向量机的风电功率实时预测研究

风能的波动性和随机性给风电功率预测带来了很大的影响，准确合理的预测可以使系统可靠、持续、稳定运行。提出一种基于相关向量机的超短期风电功率预测方法。相关向量机是在贝叶斯理论的基础上提出的一种概率学习模型，与支持向量机相比，相关向量机具有概率模型稀疏、核函数计算量小等优点。对滚动多步预测模型进行了分析，建立了相关向量机的风电功率预测模型。利用该方法对吉林西部若干风电场进行功率预测，结果表明，所提出的预测模型能有效地提高预测精度，对工程有较高的应用价值。

Real Time Prediction of Wind Power Based on Relevance Vector Machine

The volatility and randomness of wind have significant impacts on the wind power prediction. Accurate and reasonable prediction can ensure a reliable, continuous and stable system operation. An ultra-short term wind power prediction method is proposed based on relevance vector machine. The relevance vector machine is a probability learning model based on the Bayesian theory, and compared with the support vector machine(SVM), it has the advantage of sparse probability model and minor kernel function calculation. Then, based on an analysis of the rolling multi-step prediction model, a short-term wind power prediction model of relevance vector machine is developed. The model has been implemented for wind power prediction to a lot of wind farms in Jilin province, and the testing results show that the proposed model can effectively improve the prediction accuracy and is valuable for engineering implementation.