机构地区: 北京科技大学计算机与通信工程学院
出 处: 《哈尔滨工程大学学报》 2006年第B07期120-123,共4页
摘 要: 针对难以解决的纯滞后非线性系统控制,提出一种基于自调整模糊神经网络控制的辨识Smith预估方法,采用模糊神经网络与PID控制动态复合,保持了模糊控制较强的鲁棒性和神经网络可以任意逼近非线性系统的能力以及PID调解器消除静态误差的优点.同时利用神经网络进行参数在线辨识以构成Smith预估器,适应了被控对象的实时变化.在热连轧中的仿真结果证明了该方法的有效性. Against the pure time-delay nonlinear system control, an identifiable Smith pre-estimation control scheme based on self-adjusting fuzzy neural network control is proposed. Fuzzy neural network and PID control are integrated, manifesting stronger robustness of fuzzy control, ability of neural network to willfully approach the nonlinear system and the high steady accuracy of PID adjuster. Smith predictor is constituted by neural network recognizing the parameter on-line, The simulation result of the hot rolling shows that the method is effective when parameter is changed.
关 键 词: 自调整 模糊神经网络 参数辨识 预估 非线性系统
领 域: [自动化与计算机技术] [自动化与计算机技术]