| Summary: | 碩士 === 國立成功大學 === 化學工程學系碩博士班 === 92 === Nonlinear processes can be classified into two types. The first type of processes consists of nonlinear static and dynamic parts, whereas the second type consists of a linear dynamic part and a nonlinear static part. For the first type of processes, we can obtain a parameterized transfer function via extended linearization, where the parameter represents the steady-state input or output of the process. We then design a nonlinear IMC (internal model control) controller based on the IMC theory. To realize the parameterized IMC controller, we propose to convert the transfer function representation of the controller into a state-space representation by means of the residual matrix approach. Simulation results with high-order level-tank systems reveal that the proposed nonlinear IMC controller outperforms the corresponding linear IMC controller for set-point changes. We have the third-order level-tank system model by process identification. The proposed IMC design can also be incorporated with a parameterized model provided by process identification. This is verified on a third-order level-tank process.
For the second type of processes, we assume that identification as a Hammerstein model is appropriate and provide an inverse function to cancel out the static nonlinearity of the model. A linear IMC controller can then be used to control the resulting system. In this thesis, three approaches are investigated to arrive at an inverse function of the static nonlinearity. Simulation with a continuous stirred tank reactor system reveals that the proposed method works well for set point changes.
|
|---|
