ABSTRACT
Due to the discontinuous physical property of the control actuators, the state space of such a dynamical system is divided into many subdomains. For each subdomain, the flow of such a system is governed by the corresponding subsystem. The state boundary between the adjacent subdomains is called the physical switching boundary. The controller is designed to switch when the subsystem of such a discontinuous dynamical system is switched in order to have the optimum control performance. Since the ambiguity and uncertainty of modeling, the mathematical expressions for describing the discontinuous physical properties of the control actuators may not be accurate. Since the nominal switching boundary where the controller really switches is not exactly the corresponding physical switching boundary, the mismatch between the subsystem and the corresponding controller will occur and it may seriously affect the control performance. Therefore, a boundary estimation algorithm is proposed to estimate the physical switching boundaries based on the model reference control and error backpropagation. The simulation results show that the adaptive sliding mode control with the boundary estimation algorithm has superior control performance and strong robustness to deal with the internal uncertainty, the external interference, and the boundary ambiguity.
