一种自适应CMAC神经元网络控制器及其在水轮机调速器中的应用

提出了一种自适应CMAC神经元网络控制器的结构. 该控制器的核心是一个两维的存贮区间, 它采用一个参考模型和直接自适应律来获得在线训练信号, 而相应存贮单元的更新采用一阶学习律. 最后以水轮机调速器仿真实验系统来检验它的控制性能, 并与普通的PID控制器比较, 结果证明, 该控制器有较强的学习能力及较强的鲁棒性.

Adaptive CMAC controller for hydraulic turbine speed governor

An adaptive CMAC controller for hydraulic turbine governor is discussed. The adaptive CMAC controller has a two dimensional memory area, the horizontal memory location is selected by the speed of the hydraulic turbine and the vertical memory location is selected by the speed error between the turbine speed and the input signal. All of these two dimensions are linearly coded. The first order learning law and the conception of generalization is used to update the memories of the CMAC. A reference model and the direct adaptive law are employed to obtain the on-line training signal for CMAC. In order to check the control quality of the adaptive CMAC controller, a hydraulic turbine speed governor simulation experiment system is used and an ordinary PID controller is used to compare with it. The experimental results show that the adaptive CMAC controller works well and has ability to resistant the disturbance of the system parameter change.