Generalized predictive control of DEAP actuator based on RBF

摘要

Dielectric Electro-active Polymer (DEAP) materials generally have strong hysteresis, nonlinearities, and uncertainties, these characteristics making the effective control of its actuators quite difficult. A generalized predictive control strategy based on RBF neural network was proposed. The nonlinear characteristics of a DEAP actuator are approximated by RBF neural network. On this basis, the generalized predictive control strategy is adopted to achieve the predictive control of DEAP actuator via rolling optimization and feedback adjustment. Simulation results indicate that the proposed method can weaken the interference caused by nonlinearities of DEAP materials, has adaptability, robustness compared with the traditional generalized predictive control approach.