Benchmark operating points determining method of operating trajectory LPV model based on gap metric

A Linear parameter varying model(LPV) is such a model structure which is featured by its linear model structure and varying model parameters. It uses several so called local linear models on equilibrium points or operating points and their corresponding nonlinear weight functions to describe the highly nonlinear dynamics of industrial process during the whole operating range. However, a universal criterion of selecting appropriate basic operating points and their local linear model has not been given. In this paper, a operating trajectory LPV model identification method based on a nonlinearity measurement called gap metric. The first step to implement this proposed method is selecting several operating points and their corresponding local linear model identified by plant test and linear identification method according to the operating trajectory of scheduling variable. Then, we compute the gap metric between any two linear models and divide the operating range. At last, several linear models in all subspaces are chosen to make up the operating points linear model set. By implementing the proposed method into the operating range dividing and model set determining of CSTR, the simulation results shows that the proposed method can get a reduced linear model set which can approximate the dynamics of CSTR well.