T. Fan and C.W. de Silva
Flexible-link manipulator, model predictive control, dynamic modelling, motion control, system identification, non-minimum phase systems
This paper presents a strategy of model predictive control (MPC) for motion control of flexible-link robot manipulators. The present development of the MPC algorithm concerns two major aspects: development of a dynamic model and design of a real-time MPC controller. The nonlinear dynamic model of the system is derived based on the Euler–Lagrange equations of motion. More realistic boundary conditions are used in the dynamic model of the system. Local linearization is used to derive a linear model of the system at an operating point, and the model is re-linearized for large operating point changes. The unknown system payload is identified online using a prediction-error method (PEM). The MPC controller is designed based on this linear model. A computationally efficient, multi-stage MPC algorithm with guaranteed nominal stability is developed to facilitate real-time implementation of the overall scheme. Physical implementation of the developed MPC algorithm in a prototype flexible-link manipulator system (FLMS) is explored. The performance of the MPC scheme is evaluated using computer simulations and experimentation of the prototype FLMS. The results show that the developed control algorithm can effectively control the motion of the FLMS.
Important Links:
Go Back
