Summary: | 碩士 === 國立臺灣大學 === 化學工程學系研究所 === 86 === This thesis aims to present the construction of the multi-linear model and
multi-linear model based control system design. The multi-linear model
divides the possible operating regime into several sub-regimes and each
linear model represents the dynamic behavior of the specific sub-regime.
The multi-linear model combines several linear models according to the
corresponding weighting to a global model. The weighting of the linear
model depends on states. In this thesis, the decision of the weighting
magnitude uses the concept of fuzzy logic inference. Three methods of
combining these linear models are provided: model output combination, model
parameter combination and model output change combination.
The multiloop contol strategy is widely used to control an MIMO system. For
a nonlinear MIMO system, multiloop control based on the multi-linear model
is presented in this thesis. One controller is designed for each
sub-regime, and the current control output is obtained by the weighting sum
of all sub-controller outputs, parameters or output changes. Finally, the
control strategy is illustrated by two chemical processes. One is the
mixing tank with temperature and level control and the other is the
methanol-ethanol high purity distillation column.
|