基于LSTM风速区间预测的风力发电机再切入控制

Recut-in Control of Wind Turbine based on LSTM Wind Speed Interval Prediction

台风风况时,风速变化率大,风力发电机出力受限,除需要考虑提高风速的预测精度之外,还应着重考虑再切入控制方案的设置问题。而台风时风速采样数据稀少,传统利用周期性、相似性进行预测的效果并不理想和瞬时风速在切出风速上下波动时,单独采用风速死区的方法失效,严重影响风机安全性和可靠性。因此,提出一种基于递归神经网络LSTM的风速区间预测方法,通过深度神经网络对采样风速进行超短期预测,并把预测结果的置信水平,作为风速预测区间的上下限,进而实现风速的区间预测。结合实例证明提出的方法有效地降低了风力发电机的切出次数,提升了风机安全性和可靠性。

Because during typhoon the speed of wind is a high rate of change and the output of wind turbines is limited,not only should researchers consider the accuracy of wind speed prediction,but also focus on setting up the control scheme again.However,sampling data of wind speed are scarce during typhoon,and the effect of traditional periodic and similarity prediction is not ideal.In addition,wind speed fluctuates around the cutting wind speed instantaneously,so that only employing dead zone is invalid,which seriously affects the safety and reliability of fans.A prediction of wind speed interval based on LSTM is proposed in this paper.Using the deepen neural network predicts the wind speed in super short time,and take the confidence level of prediction results as the Upper and lower limits.Combined with practical examples,it is proved that the proposed method can effectively reduce the number of cutting times of wind turbines,and also improve the availability and reliability of wind turbines.