基于NARX神经网络预测及模糊控制的互联电网CPS鲁棒控制策略研究  

CPS robust control strategy research based on NARX neural network pre-sentient algorithms and fuzzy logic controller

作　　者：李挺[1] 雷霞[1] 张学虹[2] 孔祥清[1] 刘庆伟[1] 柏小丽[3] 

机构地区：[1]西华大学电气信息学院,四川成都610039 [2]广安电业局,四川广安638000 [3]泸州电业局,四川泸州646000

出　　处：《电力系统保护与控制》2012年第14期58-62,68,共6页Power System Protection and Control

基　　金：国家自然科学基金项目(51007006);四川省科技厅应用基础项目(2009JY0139);西华大学2009人才培养项目(R0920906);四川省教育厅重点项目(11Za002)~~

摘　　要：对于传统CPS控制策略难以满足互联电力系统对鲁棒性和适应性的要求,提出了一种将NARX神经网络预测算法和模糊逻辑控制器相结合的控制方法。配合CPS下的传统PI控制器,根据CPS控制参数的预测值与当前值之间的偏差值,实现对AGC机组的预控制。利用Matlab的Simulink仿真软件建立了一个双区域电力系统的控制模型。仿真结果表明,新的控制方法不仅达到了改善CPS控制效果的目的,并且提高了CPS1,CPS2指标的考核率,减少了机组的调节次数,降低了运行费用,取得了一定的经济效益。Traditional CPS controlling strategy is hard to meet the demand for robustness and adaptability in interconnected power system. A new control method combining NARX neural network pre-sentient algorithms with fuzzy logic controller is proposed. Cooperate with traditional PI controller based on Control Performance Standard （CPS） and realize the pre-control in AGC generator according to the error between predicted value and current value of CPS control parameters. By applying the simulating software MATLAB （Simulink）, this paper sets up a two-area control system of power grids. According to the results, new control strategy can not only enhance CPS control effect, but also improve the CPS1 and CPS2 standard assessment rate, achieving some economic benefits by reducing the regulation times and operating costs of units. This work is supported by National Natural Science Foundation of China （No. 51007006）.

关 键 词：控制性能标准 鲁棒性 NARX神经网络预测算法 模糊逻辑控制器 预控制 

分 类 号：TM76[电气工程—电力系统及自动化]