Multilinear model decomposition of MIMO nonlinear systems and its implication for multilinear model-based control

In order to accomplish the multilinear model decomposition of MIMO nonlinear processes with multiple scheduling variables, a systematic division algorithm based on gap metric together with a supporting dichotomy gridding algorithm is proposed by using the gap metric as a measuring tool. For a prescribed distance level, this gap metric based division algorithm effectively decomposes a MIMO nonlinear system into a set of linear subsystems which provide enough model information for multilinear model-based controller design without linear model redundancy. Based on the linear models, a set of linear MPC controllers are designed and combined into a global controller for setpoint tracking control. Two benchmark nonlinear processes are studied to demonstrate the effectiveness of the proposed method.

► Multilinear model decomposition and control of MIMO nonlinear systems with multiple scheduling variables is investigated.
► A dichotomy gridding algorithm is proposed by using the gap metric as a measuring tool.
► A systematic division algorithm based on gap metric is proposed for MIMO nonlinear systems.
► The proposed division algorithm decomposes a MIMO nonlinear system into a set of linear subsystems effectively.
► The subsystems provide enough model information for multilinear controller design without model redundancy.