Summary: | 碩士 === 元智大學 === 電資與資訊工程研究所 === 86 === This thesis investigates the temperature control problem for
hotrolling process during finishing milling. We aim to preserve
the nonlinear time-varying characteristics of the plant model
in controller design. When the plant model is nonlinear, the
common approach in controller design is to linearize the model;
an alternative is to convert the model into a fuzzy model with
locally linear rules. In our problem, we find that the nonlinear
thermal model can be used directly for controller design. It is
unnecessary to linearize the model or convert it into a fuzzy
model. We propose optimal controllers based on the thermal model,
and PID controllers based on optimal control solution. In the
special case when the thermal model cannot provide the control
solution, it can be converted into a fuzzy model. With the
locally linear rules, control solutions for fuzzy models can
always be found, and model nonlinearity is preserved. We also
propose optimal controllers based on the fuzzy model and the
related PID controllers. Compare to the controllers obtained
after model linearization, our controllers perform better and
are easier to design. Ideal controller parameters can be
obtained directly without trials and errors.
|