| Summary: | 碩士 === 國立成功大學 === 電機工程學系 === 102 === An observer/Kalman filter identification (OKID) method-based adaptive mechanism for tuning weighting matrices of multi-objective cost function is newly proposed in this thesis. An efficient algorithm is newly derived for the new tracker-design for the unknown system with an input-output feed-through term and input/state/output constraints. First, the unknown linear/nonlinear system containing an input-output feed-through term is identified by the observer/Kalman filter identification (OKID) method to have the equivalent mathematical model, then the controller is analyzed and designed by the equivalent mathematical model. The linear analogue quadratic performance index is modified to contain the term of input, state, and output constraints. The linear analogue quadratic performance index with input, state, and output constraints can be directly discretized into an equivalent discrete function, so that the obtained quadratic sub-optimal digital tracker can preserve the performance of the linear analogue quadratic performance index. In order
to make the exceeding input, state and output update quickly and accurately, an OKID-based adaptive mechanism for tuning the weighting matrices is constructed. Finally, an OKID-based adaptive mechanism for tuning weighting matrices of the new tracker-design for the unknown system with an input-output feed-through term and input, state, and output constraints is proposed.Examples show the usefulness of the proposed design.
|
|---|
