风电功率统计建模及预测

Modeling and Prediction of Wind Power Statistics

摘要: 本文针对风电功率的预测问题，分别采用时间序列法、人工神经网络、灰色预测法对未来机组输出的电功率建立了三种合理预测模型，并通过对各种模型的误差分析，进一步提出了改进的方案使其预测更加的准确，在最后将模型推广到n台风电机组并给出了合理的模型。我们通过分析知道阻碍风电功率实时预测精度进一步改善的主要因素是因为实测数据本身就存在各种不可避免的误差，得出风电功率的预测精度无法得到无限提高的结论。

Abstract: In this paper, time series, artificial neural network, gray forecast are adopted to predict the power of wind turbines. By means of setting up three types of rational prediction models and performing error analysis, this paper presents an improvement scheme to make predictions more accurate. Finally, the models are applied to several wind turbines. The analysis demonstrates that the orig-inal measurement data often contain various errors, so the prediction accuracy of wind power cannot be infinitely increased.