摘要
为了更好地提升风力发电系统的安全性与经济性,提高风速预测的精确度,笔者提出了一种长短神经网络(Long-Short Term Memory,LSTM)模型用于风速预测。首先模型使用防止过拟合和提高模型泛化能力的归一化方法对数据进行处理;其次将时间序列数据转换为监督学习问题;最后使用对时间序列预测有着较好鲁棒性的LSTM模型对风速进行预测。实验结果表明,相较于支持向量回归法,笔者提出的LSTM在风速预测上具有更高的预测精度。
In order to improve the security and economy of the wind power system,this paper proposes a LSTM model for wind speed prediction.Firstly,the normalized method is adopted to prevent overfitting and improve the generalization ability of the model,so as to process the data in the model.Then,the time series data are transformed into supervised learning problems.Finally,the LSTM model with better robustness for time series prediction is used to forecast the wind speed.Compared with the support vector regression method,the proposed LSTM has higher accuracy in wind speed prediction.
作者
涂安龙
胡萧书萌
邓万雄
TU Anlong;HUXIAO Shumeng;DENG Wanxiong(School of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan Hubei 430065,China)
出处
《信息与电脑》
2021年第12期142-145,共4页
Information & Computer