基于RBF神经网络的风电机组独立变桨控制  

Based on RBF neural-network for individual pitch control of wind turbine

作　　者：韩兵[1] 周腊吾[1] 陈浩 邓宁峰[1] 田猛[1] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]湖南世优电气股份有限公司,湘潭411100

出　　处：《中国科学：技术科学》2016年第3期248-255,共8页Scientia Sinica(Technologica)

基　　金：科技部国家国际科技合作专项(批准号:2011DFA62890)资助项目

摘　　要：由于风速的随机性、风电机组参数的时变性以及复杂的变桨系统引起的时滞性,随着风力机桨叶长度的不断增加,叶片受力拍打振动的情况越来越严重,同时造成输出功率不稳定.为改善风机变桨系统在运行区域内的动态性能,本文依据风力机空气动力学原理、风剪切特性和塔影效应,提出了基于径向基函数(RBF)神经网络自适应独立变桨距控制方法,采用RBF神经网络逼近变桨系统未知的非线性函数,通过Lyapunov方法导出神经网络自适应率,在线调整神经网络权值来改善独立变桨系统的动态性能,最后通过设计风电机组的独立变桨控制模型进行相关实验,证明基于RBF神经网络自适应独立变桨控制系统具有良好的动态性能,可以有效稳定输出功率,降低桨叶、轮毂、机舱、塔架等风电机组关键部件的疲劳载荷.Due to the randomness of wind speed, time-varying parameters of wind turbine pitch systems and delay of complex pitch control system induced, with the increasing length of the turbine blades, blade flapping vibration force by the situation is getting worse, while causing the output power instability. To improve turbine pitch system dynamic performance in the area of operations, the paper proposed based on radial basis function（RBF） neural network adaptive individual pitch control method, it using RBF neural network to approach pitch system unknown nonlinear functions, neural network adaptive rate derived by Lyapunov method, neural network weights adjusted online value to improve the dynamic performance of individual pitch control system. Finally, it design with individual pitch control model in wind turbine, experiments proved to have a good dynamic performance based on RBF neural network adaptive individual pitch control system, it can effectively stabilize the output power, lower fatigue load of blade, cabin, tower and other key components in wind turbine.

关 键 词：风机载荷 叶素 自适应率 动态性能 叶根弯矩 

分 类 号：TM315[电气工程—电机]