【摘要】 将CMAC网络用到具体非线性系统的预测控制研究中,并且在控制中将CMAC建立的模型输出值与测量输出值进行综合,代替量测输出用于控制中,从而降低辨识器与控制器对未建模动态的敏感性,加强控制器的适应能力和鲁棒性。对一类CSTR系统的仿真结果表明,该预测控制策略响应快且容易实现,对于改善非线性预测控制不失为一种有益的尝试。更多还原

【Abstract】 A local structure of CMAC results in faster learning of nonlinear systems. The nonlinear system is identified by CMAC neural network. Since the output sequence used in the controller design is the combination of the system output and the model output, so the sensitivity of the identifier and controller is cut down, and the adaptive ability and robustness of the controller is increased. Simulation results of a CSTR system predictive controller show that the method has a quick response speed and good robustness, so it is helpful to improve the capabilities of nonlinear predictive controller.更多还原