An off-line Model Reduction-based Technique for On-line Linear MPC Applications for Nonlinear Large- Scale Distributed Systems

摘要

Linear Model Predictive Control (MPC) has been effectively applied for many processrnsystems. However, linear MPC is often inappropriate for controlling nonlinear largescalernsystems. To overcome this, model reduction methodology has been exploited tornenable the efficient application of linear MPC for nonlinear distributed-parameterrnsystems. An implementation of the proper orthogonal decomposition method combinedrnwith a finite element Galerkin projection is first used to extract accurate non-linear loworderrnmodels from the large-scale ones. Then a Trajectory Piecewise-Linear method isrndeveloped to construct a piecewise linear representation of the reduced nonlinear model.rnLinear MPC, based on quadratic programming, can then be efficiently performed on thernresulting system. The stabilisation of the oscillatory behaviour of a tubular reactor withrnrecycle is used as an illustrative example to demonstrate our methodology.