基于RBF神经网络的超导风力发电系统变速控制研究  

VARIABLE SPEED CONTROL STUDY OF SCSG WIND TURBINE SYSTEM BASED ON RBF NEURAL NETWORK

作　　者：左姗[1] 宋庆旺 王磊[2] 宋永端[2] 

机构地区：[1]电子科技大学智能系统和可再生能源技术实验室,成都611731 [2]重庆大学智能系统与新能源技术研究院,重庆400044

出　　处：《系统科学与数学》2014年第2期145-157,共13页Journal of Systems Science and Mathematical Sciences

基　　金：国家高技术研究发展计划(863计划)(SS2012AA052302);国家重点基础研究发展计划(973计划)(2012CB215202);国家自然科学基金会(51205046)资助课题

摘　　要：超导同步发电机(SCSG)以其功率密度大,重量轻,体积小,寿命长的特点,已经被用来研究超大型海上风力发电机组.文章首先介绍了一种将高温超导材料应用于大型风力发电机组的技术,分析了高温超导发电机的结构和特点,基于FAST和Matlab/Simulink软件平台搭建SCSG风力发电机组的数学模型和速度控制模块,其次针对SCSG模型存在高阶,非线性时变的特性,在速度控制环中引入了基于Delta学习规则的单神经元自适应PID控制算法,其中RBF网络用于控制参数的辨识整定.模拟实验结果表明,相对于传统PID控制,该算法可以使SCSG速度更好地跟踪风速,保持稳定的最佳叶尖速比和最大功率系数,从而实现最大风能利用.Superconducting synchronous generator （SCSG） has been used to develop large-scale offshore wind turbine because of its high energy density and the advantages in the weight and size.This paper firstly introduces a technology of high-temperature superconducting materials used in large-scale wind turbine,and analyze the structure and characteristics of high-temperature superconducting generator.Then taking into account the high-order,nonlinear,time-varying characteristics of SCSG model,a control algorithm of single neuron adaptive PID controller is introduced based on Delta learning regulation,in which the radial basis function （RBF） neural network is used to identify the undetermined portion.The paper applies the improved algorithm in the speed control of SCSG wind generation system.Simulation results show that compared to traditional PID control,SCSG speed with improved algorithm can better track the wind speed,maintain a more stable optimal tip speed ration and maximum power coefficient,for achieving the maximum wind energy utilization.

关 键 词：超导同步发电机(SCSG) 风力发电 单神经元PID控制 改进RBF网络 鲁棒性 

分 类 号：TM614[电气工程—电力系统及自动化]