伴随型非线性系统的自适应RBF神经网络补偿控制

为了抵消伴随型非线性系统中的非线性项,可以设计控制器对非线性系统精确线性化.通常由于系统中存在外界不确定性因素导致系统模型的不确定,而不能直接设计控制器.利用"RBF神经网络能以任意精度逼近连续函数"的原理,对系统模型中的不确定项进行自适应辨识,并将辨识结果提供给控制器,从而实现伴随型非线性系统的神经网络自适应补偿控制.将控制器应用于起重机吊重摆角子系统,对摆角进行控制.实验结果表明:吊重摆角及其角速度约在5s后,得到了很好的控制,并且控制器对系统模型的不确定项的逼近误差约在5s时达到0;控制器对系统的不确定性因素和系统参数变化均具有很强的鲁棒性.

Companion nonlinear system control based on adaptive RBF network compensation

In order to counteract the nonlinear term in companion nonlinear system, a controller can be designed to precisely linearize the nonlinear system. Generally, there are uncertain factors existing outside the system which lead to the system model uncertainty, so the controller cannot be designed directly. The uncertain term in the system model was adaptively identified by using the principle of that RBF neural network could approximate any continuous function with any precision. The identified result was provided to the controller and it realized the adaptive compensation control of the companion nonlinear system based on neural network. The designed controller was used to control swing angle subsystem of crane-load system. Experiment results showed that the swing angle of the load and the angular velocity of the swing angle were well controlled in about 5 s, and the approximation error of the uncertain term in the system model could reach zero at about 5 s; the designed controller had strong robustness against the uncertain factors of the system and the change of the system parameters.