| Summary: | 博士 === 逢甲大學 === 土木及水利工程博士學位學程 === 101 === As the improvement of personal computer and the popularity of GIS (Geographic Information System), the development of comprehensive techniques with GIS and artificial intelligence (AI) algorithm is becoming an important trend in environment-related research. There are two purposes in this dissertation: first, according to each of different environmental behavior-pattern (EBP) problems applies AI algorithm which has consistent logic in order to find solutions; second, establish a comprehensive framework to sort and conclude the essence of each EBP problem, and figuring out the distinctions of EBP expert knowledge.
This research uses the informal commercial activities of the Feng-Chia night market as the experimental EBP target for operating a series of case study experiment. The AI techniques applied in this research are: artificial neural network (ANN), self-organizing map (SOM), genetic algorithm (GA), cellular automata (CA) and Bayesian probability, and using CA-GIS as the integration data base. Additionally, special AI procedures are offered in order to resolve specific issues, like GAANN (GA+ANN), NNSOM (ANN+SOM) and (Bayesian probability+ANN). In the end, we use two kinds of axis system: one is the inference system of the occurrence order of EBP problems, and the other is the distinction system of EBP expert knowledge to establish &;quot;the systematic framework for investigating EBP based on AI techniques&;quot;. This systematic framework can integrate all case study experiments, and conclude the essences of EBP problems. The EBP knowledge of resolving, producing and managing can be more objective, systematic and practical.
|
|---|
