面向直驱风力发电系统的控制算法优化设计  

Optimization Design of Control Algorithms for Direct Drive Wind Power Generation System

作　　者：万抒策 苏人奇 钱韫辉 WAN Shuce;SU Renqi;QIAN Yunhui(Huaneng Clean Energy Research Institute,Beijing 102209,China;Huaneng(Zhejiang)Energy Development Co.,Ltd.,Clean Energy Branch,Zhejiang 310014,China;Shanghai Qince Electrical and Mechanical Engineering Technology Co.,Ltd.,Shanghai 200434,China)

机构地区：[1]中国华能集团清洁能源技术研究院有限公司,北京102209 [2]华能(浙江)能源开发有限公司清洁能源分公司,杭州310014 [3]上海擎测机电工程技术有限公司,上海200434

出　　处：《自动化与仪表》2023年第9期46-50,共5页Automation & Instrumentation

摘　　要：针对直驱风力发电机组的功率输出不稳定、频率波动超出安全范围问题,该文提出了一种基于深度学习网络预测控制的直驱风力发电系统控制策略。以直驱风力发电系统的频率误差和输出功率为优化目标函数,并将系统的频率变化及风力发电机的转矩限制作为约束条件,利用时间卷积神经网络(TCN)与门控循环单元(GRU)构建预测模型。可行性验证结果表明,在低风速及高风速下有功功率输出的标准差分别为0.04 MW及0.05 MW,轴转矩标准差分别为0.01 MNm与0.05 MNm,推力标准差分别为0.04 MN和0.05 MN,证明了该策略可有效降低直驱风力发电系统输出功率的频率波动。In response to the problem of unstable power output and frequency fluctuations exceeding the safe range of direct drive wind turbines,this paper proposes a control strategy for direct drive wind power generation systems based on deep learning network predictive control.Taking the frequency error and output power of the direct drive wind power generation system as the optimization objective function,and taking the frequency variation of the system and the torque limit of the wind turbine as the constraint conditions,a prediction model is constructed using time convolutional neural network(TCN)and gate recurrent unit(GRU).The feasibility verification results indicate that the standard deviations of active power output at low and high wind speeds are 0.04 MW and 0.05 MW,respectively,the standard deviations of shaft torque are 0.01 MNm and 0.05 MNm,and the standard deviations of thrust are 0.04 MN and 0.05 MN,respectively.This proves that the strategy can effectively reduce the frequency fluctuations of output power in direct drive wind power generation systems.

关 键 词：直驱风力发电系统 风电场 时间卷积神经网络 门控循环单元 优化目标函数 模型预测控制 

分 类 号：TM315[电气工程—电机] TP18[自动化与计算机技术—控制理论与控制工程]