风电场短期风速预测方法对比研究#.PDFVIP

中国科技论文在线 # 风电场短期风速预测方法对比研究 龚松建，袁宇浩，王莉，张广明* （南京工业大学自动化与电气工程学院，南京 211816） 摘要：本文介绍了三种风电场风速预测模型，分别是时间序列模型、BP 神经网络模型和小 波神经网络组合模型。时间序列ARMA模型和BP神经网络模型是风速预测中常用的模型，为 了提高预测精度，引入了小波神经网络组合模型。对三种方法在短期风速预测中的应用进行 了研究和比较，并针对不同的风速样本进行了分析，研究结果表明：在风速较平稳时，宜采 用 BP 神经网络模型进行预测；在风速波动较大时，可以采用小波神经网络模型预测。实际 运用时，应根据具体情况进行分析和判断，选择合适的模型，以取得最优预测结果。 关键词：风速预测；时间序列；神经网络；小波分析 中图分类号：TM715 Short-term wind speed forecasting method for wind farms Gong Songjian, Yuan Yuhao, Wang Li, Zhang Guangming (Institude of Automation and Electrical Engineering,Nanjing University of Technology, NanJing 211816) Abstract: This article describes the three kinds of wind speed forecasting models, namely the time-series model, BP neural network model and wavelet neural network combined model. Time-series ARMA model and BP neural network model is commonly used model in the wind speed forecast, in order to improve prediction accuracy wavelet neural network combined model has been taken into account. The three methods in the short term wind speed forecasting has been studied and compared, and wind speed from different samples has been analyzed, the results showed that: if the wind speed relatively stable, it is desirable to use BP neural network model to predict; if the wind speed fluctuated too large, it is desirable to use wavelet neural network model in the prediction. In the actual application, it is proper to use the method which is suitable for the circumstance, to select the appropriate model in order to obtain the optimal prediction. Keywords:wind speed forecasting;time series;neural network;wavelet analysis 0 引言 风力发电是解决当前突出的能源和环境问题的有效手段, 因而得到了普遍重视和大规 模开发利用, 然而风资源具有多变的特性，当风电穿透功率超过一定值时，必然