Inverse-backlash-compensation Neural Network Nonlinear Predictive Control and Simulation

摘要

A discrete-time predictive inversion compensation scheme has been studied for backlash compensation in nonlinear systems. The controller uses the dynamic inversion technique with neural networks for inverting the backlash nonlinearity in the feed-forward path. The neural network controller is trained with multi-step predictive optimization algorithm. Another neural network is used for constructing the predictive model. It is not necessary for control designers to have accurate information about the plant dynamics. Results are presented from the application of such a control scheme to a nonlinear system with backlash input. It is found to be very efficient at cancelling the deleterious effects of actuator backlash.