臺灣博碩士論文加值系統

本文提出 GANARXSI (GA-based NARX System Identification
software，應用 C++ 語言完成之視窗軟體) 做為 NARX 系統鑑別的工具
．首先應用 GANARXSI ，於 SGA ( Simple Genetic Algorithm) 下鑑
別數個 NARX 模式並比較各種 GA 運算子組合之優劣．其次應用前述所得
之結果做為 GAMAS (GA based on Migration and Artificial
Selection) 之運算子組，用以鑑別非線性離散及連續系統．由GAMAS 所
得之系統模式， 其表現令人滿意．此外，提出一新的突變運算子，
truncate mutation ，用以增進 NARX系統鑑別之效能．由模擬之例子中
得知在某些 GA 運算子組合中，其確實有助於 NARX 系統之鑑別．

In this thesis, a program named GANARXSI (GA-based NARX
System Identification software) written in C++
language with graphic user interface (GUI) is proposed and
utilized for NARX system identification. We employed GANARXSI
to identify models of several NARX system equations and compared
the combinations of GA operators under SGA. We also used it to
identify nonlinear discrete and continuous systems with well
combinations of GA operators under GAMAS. The behavior of
obtained system models is satisfying. In addition, a novel
mutation operator called truncate mutation is proposed to
improve the performance of NARX system identification. From
the simulated cases, it reveals that the truncate mutation
is indeed helpful for NARX system identification in GA.