不确定系统的上界自适应动态神经滑模控制

针对滑模控制中不确定上界值的问题，提出一种神经网络上界自适应学习的动态滑模控制方法。该方法将系统中不确定及干扰部分分离出来，构造不确定量的联合上界，然后分两步进行分析。当上界已知时，采用动态滑模方法设计滑模控制器；上界未知时，采用神经网络自适应学习不确定项的上界，设计了权值调整规则及动态神经滑模控制器。该控制器不仅可以保证非线性不确定系统渐近稳定，降低一般滑模控制理论分析的条件，还有效地抑制了抖振。仿真实例表明，该控制方法是正确有效的。

Upper Bound Adaptive Dynamic Neural Sliding Mode Control of Uncertain Systems

For the problem that the uncertain upper bound value must be known in the general sdudy of sliding mode control, the upper bound adaptive dynamic SM（Sliding Mode) control method based on NN(Neural Networks) is proposed. Uncertainty and disturbance of the systems are separated from the systems to construct a conjoint upper boundary of the uncertainty. The process was analysed in two steps. When the conjoint upper bound is known, the dynamic SM controller is designed.Otherwise the NN are adopted to learn adaptively the upper bound of the uncertainty. The rule of the weight adjustment and dynamic neural sliding mode controller are designed. The stability of the systems is investigated by constructing Lyapunov function. The controller can ensure the systems asymptotic stability. And it reduces the theory analysis condition of the SM control. It can also suppress the chattering effectively. The simulation examples show that the proposed dynamic neural sliding mode controller is correct and effective.