基于半物理仿真的RBF神经网络滑模控制

针对一类不确定时滞系统研究滑模控制的实现问题。对于实际应用对象的时滞特性采取了特殊的线性变换，将原时滞系统转化为无时滞系统。通过设计二次型性能指标计算得到了最优的切换函数，并使用RBF神经网络实现了滑模控制的自适应等效控制，保证了系统能够克服扰动，系统状态在有限时间能够到达滑模面。系统仿真验证了该方法的有效性和稳定性。

RBFNN Siding Mode Control Based on Semi-physical Simulation

Based on Matlab RTW semi-physical simulation platform, a siding mode control algorithm is presented for a kind of uncertain time-delay system. Through a particular linear transformation, the original uncertain time-delay system is first transformed into a delay-free system. Based on the transformed system, a design method of optimal sliding mode with a quadratic performance index minimized is proposed. An appropriate control law which is approached by a RBF Neural Network(RBFNN) and the weight of the network is tuned on line using adaptive algorithm to force the system states to reach the sliding manifold in finite time. Simulation results show the efficiency and superiority of the proposed method.