MPC DESIGN FOR CONSTRAINED MULTIVARIABLE SYSTEMS UNDER ACTUATOR BACKLASH

A new approach to the design of Model Predictive Controller (MPC)
that simultaneously addresses the actuator saturation and backlash is proposed
in this paper. The discrete characteristics of the actuator backlash allows one to
reformulate the input constraints as a set of mixed integer linear inequalities. As
a result, the MPC is designed by solving a Mixed-integer Quadratic Programming
problem. Simulation results are presented to show how this new approach performs
as compared to the existing techniques of the backlash compensation when they
are applied to an industrial case study.