风电系统最大风能追踪的智能模型预测控制  

Intelligent model predictive control of wind power system to trace maximal wind energy

作　　者：刘吉宏 吕跃刚[1] 郭鹏[1] 徐大平[1] 

机构地区：[1]华北电力大学控制科学与工程学院,北京102206

出　　处：《计算机工程与应用》2011年第9期228-232,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.50677021);教育部重点项目基金(No.105049)~~

摘　　要：根据最大风能捕获原理,额定风速以下风能的最大追踪可以通过控制双馈感应发电机(DFIG)跟踪最优转速来实现。以变速恒频双馈风力发电系统为研究对象,研究了额定风速以下风能的最大追踪控制问题。首先针对双馈发电机强耦合、强非线性、机理模型复杂的特点,采用支持向量机(SVM)理论建立了智能预测模型;然后利用反馈校正的方法对预测输出进行修正,构成控制闭环;最后利用粒子群优化算法(PSO)调整参数少、演化群体小、计算速度快的优点容易地求出最优控制序列,较好地解决了滚动优化计算中的"瓶颈问题"。仿真结果验证了所采用的预测模型具有比较好的抗干扰能力和泛化能力,预测控制算法能够实现控制目标。Based on the principle of maximum wind energy capture,the maximum wind energy tracing below the rated wind speed can be realized by controlling the speed of Double-Fed Induction Generator（DFIG） to track the optimal speed.The variable speed constant frequency double-fed wind power system is taken as the research object and the control problem of maximum wind energy tracing below the rated wind speed is studied.Firstly,according to the characteristics of strong coupling, strong non-linearity and the complexity of mechanism model for DFIG,the intelligent predictive model is built by using Support Vector Machine（SVM）.Secondly,the predictive output is revised by feedback correction and the closed control loop is structured.Finally,due to the advantages that Particle Swarm Optimization（PSO） algorithm has the fewer regulated parameters, the small evolution groups and the quick calculation speed,the optimal control sequences are obtained easily and the＂bottleneck problem＂in the rolling optimization calculation is solved better.The simulation results validate that the adopted predictive model has better anti-disturbance and generalization abilities,and the predictive control algorithm can realize the control objects.

关 键 词：非线性智能模型预测控制 最大风能捕获 双馈感应发电机(DFIG) 转速控制 粒子群优化 支持向量机(SVM) 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TM614[自动化与计算机技术—控制科学与工程]