Nonlinear Robust Control Design And Application For Complex System Of IPMC Artificial Muscles

The ionic polymer metal composite(IPMC) artificial muscle actuator belongs to category electroactive polymers(EAP), compared to the traditional actuators, the IPMCs have the following merits: large strain and stress induced electrically, light in weight, small and simple mechanisms, small electric consumption, and low drive voltage, which have been widely used in the developments of bio-robots and biomedical devices. To control precise movements in the specific application, the displacement and bending of IPMC artificial muscles are very important in the application fields. Because the IPMC actuator is a highly nonlinear plant, and also relate to electricity, chemistry and power in the dynamic process. Moreover, the control performance is affect by the parameter variations, various disturbances, and constraint control input, and it is not controlled well by the traditional methods. Therefore, it is so curial that how to design controller to guarantee the robust stability and tracking in the presence of model uncertainties and constraint input.In this research, a white-box dynamic model based on physical mechanism is investigated. Considering control accuracy, robust stability and quick response, several kinds of control methods are proposed to design the stable tracking system. In order to control precisely the IPMCs’ position, first, the system with uncertainties and without uncertainties are investigated by sliding mode control, and the chattering caused in the control process is compensated by the boundary method. Second, the parameters of sliding mode controller are optimized by PSO to improve further accuracy, obtain higher quick response, and guarantee stronger robustness. Finally, PI based neural network control is proposed to the obtained stabilized system, the parameters of tracking controller are optimized by neural network. For the every scheme, the robust stability and tracking conditions is discussed. The effectiveness of the proposed methods are confirmed by the simulation results.