An integrated output space partition and optimal control method of multiple-model for nonlinear systems

Abstract

A systematic method for optimally partitioning operating range of each linear subsystem in output space under the criterion of closed-loop performance is initiated, when a multiple model approach is applied to nonlinear systems. As a result, an integrated output space partition and optimal control method is proposed. Firstly, linear input-output models are identified at given operating points and then reformulated as a hybrid model underlying each state having the same physical meaning. Secondly, the optimal state space partition is obtained according to a closed-loop control index. Finally, based on the obtained optimal state space partition, an optimal output space partition is achieved with the projection technique. Furthermore, a hybrid model-MPC strategy is designed according to the obtained multiple-model associated with its optimal output space partition. The integrated output space partition and optimal control method can improve the nonlinear system overall control performance and results of numerical simulation are provided.