一种基于延时补偿的电流跟踪控制新方法  

A Novel Current Tracking Control Method Based on Delay Compensation

作　　者：吴敬兵[1,2] 罗安[1] 杨晓峰 方璐[1] 马伏军[1] 徐先勇[4] 杨翠翠[1] 

机构地区：[1]湖南大学电气与信息工程学院长沙410082 [2]福州电业局福州350009 [3]陕西地方电力集团有限公司宝鸡供电分公司宝鸡721001 [4]湖南省电力公司科学研究院长沙410007

出　　处：《电工技术学报》2012年第4期123-130,共8页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金(60774043);国家973计划(2009CB219706)资助项目

摘　　要：提出一种基于π补偿Smith预估器和算法改进型神经网络的电流跟踪控制方案。π补偿预估器的引入有效地补偿了系统控制中的延时,提高了系统的稳定性能和响应速度;算法改进型神经网络用于优化PI控制器的参数,以提高系统的补偿精度。同时,利用ITAE准则给出π补偿预估器参数与PI控制器参数间的数学关系式。通过算法改进型神经网络可以同时优化两个控制器的参数,避免了将PI控制器参数与预估器参数分开独立识别的局面,并降低了PI控制器和π补偿Smith预估器对电网参数的敏感依赖性。当电网负载发生变化时,能够利用改进的神经网络算法实时地寻求到最优的控制器参数,达到理想的控制效果。仿真和实验验证了所提控制策略的有效性。This paper proposes a novel current tracking control scheme,which is based on π compensation Smith predictor and neural network with a modified weight algorithm.The introduction of π compensation Smith predictor has some advantages,such as effectively eliminating system delay,improving system stability and response speed.The neural network is used for optimizing PI controller parameters and then improving control system precision.Meanwhile,obtaining the relation of π compensation Smith predictor parameters and PI controller parameters by ITAE criterion（Integrated Time Absolute Error criterion）,which can be utilized to get parameters estimation of two controllers,that avoids the case of separately identifying two controller parameters,and reduces the sensitive dependence on the grid parameters of two controllers.When the loads change,the best parameters of controller can be obtained by improved weights algorithm of neural network.Simulation and experimental results verify the effectiveness of proposed method.

关 键 词：滤波器 延时 粒子群算法 π补偿预估器 神经网络 

分 类 号：TN713[电子电信—电路与系统]