摘要
风力发电的随机性和不可控性,给发电控制环节造成控制负担,利用不同高度的风速、风向等气候因素对风力发电数据进行准确预测,有利于制定合理的调度计划。本文使用LSTM循环神经网络算法,实现了单变量预测未来一个时间点、多变量预测未来一个时间点的风电预测实例验证,实验结果发现算法在两种情况下均保持着高的预测精度。且由于单变量和多变量两种情况下预测误差相差不大,表明在实际风电预测中在多变量数据里风速本身仍起决定性作用。
The randomness and uncontrollability of wind power generation puts a control burden on the power generation control link.The use of wind speed and wind direction at different heights to accurately predict wind power data is conducive to formulating a reasonable dispatch plan.In this paper,the LSTM recurrent neural network algorithm is used to realize the verification of wind power forecasting examples in which univariate predicts a point in the future and multivariate predicts a point in the future.The experimental results show that the algorithm maintains high prediction accuracy in both cases.And because the prediction error is not much different between the univariate and the multivariate cases,it shows that the wind speed itself still plays a decisive role in the multivariate data in the actual wind power forecasting.
作者
金宇悦
康健
陈永杰
Jin Yuyue;Kang Jian;Chen Yongjie(Yisheng Innovation Education Base,North China University of Technology,Tangshan Hebei,063210;School of Electrical Engineering,North China University of Technology,Tangshan Hebei,063210)
出处
《电子测试》
2022年第2期49-51,共3页
Electronic Test
关键词
风力发电预测
风速
循环神经网络
长短期记忆神经网络
wind power forecasting
wind speed
cyclic neural network
long and short-term memory neural network
