Multiple model predictive control of robot manipulator

A multiple model predictive control method based on membership function was proposed according to nonlinear characteristics of robot manipulator. An appropriate scheduling variable was selected according to the characteristics of the robot manipulator. The operation space of the robot manipulator was divided into several subspaces, the robot manipulator was linearized at equilibrium point in each subspace, and linear sub-models were built in each subspace, and the multiple model presentation of the robot manipulator was developed. Then, local predictive controllers were designed according to each linear sub-model, and make it satisfy with control requirements in the subspace. Finally, the local predictive controllers were combined by trapezoidal membership functions into a global multiple model predictive controller to control the robot manipulator. The simulation results show that the control performance of the global multiple model predictive controller based on the membership functions is superior to conventional PD controller when the robot manipulator was working in a wide operating range, so as to realize the desired control goal.