人工智慧最佳化程式模組系統之開發-以基因演算法為基礎

Using artificial intelligence to develop computer programs module system based on genetic algorithm.

林俊儒(Jyun-ru Lin)

指導教授 ： 吳獻堂

朝陽科技大學/理工學院/營建工程系/碩士(2012年)

摘要

在目前資訊科技進步的時代，人工智慧是極為重要的一環，其代表著人類長久以來的夢想，電腦不再只是計算與處理資料的機器，加入了人工智慧的電腦能做出同等人類思考模式的行為，幫忙人類處理需要認知、推論、甚至判斷等問題。基因演算法為人工智慧分支之一，求解效率與準確性已經多方面被證實是快速且正確的演算法，其應用於最佳化問題上，求解領域可從基本的單目標到多目標最佳化，而電腦硬體與軟體程式語言的進步帶動了人工智慧的發展，而所有有關人工智慧演算法的應用或研究皆要有符合其個別限制條件與個別目標的電腦程式，即不同的問題有著其專屬的電腦程式。雖然目前已經有以基因演算法為基礎的套裝軟體(例如：Evolver、SUGAL、Gene Hunter)，但其在使用上有諸多限制、或無法完全發揮基因演算法的效能，本研究開發一套以基因演算法為基礎的開放原始程式碼架構標準化程式模組系統，以樂高積木的概念將程式予以積木方式模組化，提供使用者能在短時間內依其題目需求自行組合以符合其限制條件的電腦程式，當有模組無法滿足其需求時，亦可自行修改原始程式碼，以符合特殊需要。

關鍵字

基因演算法

並列摘要

In an era with advanced information technology, the artificial intelligence is an important factor. It represents the long-time dream of human beings. The computer is no longer a machine of calculation and data processing. The computer integrated with artificial intelligence could perform in a way with the same thinking model of human beings. It can help people to solve problems which need cognition, inference and even judgement. The genetic algorithm is a branch of artificial intelligence. The solution efficiency and accuracy have proved from many aspects that it is a quick and correct algorithm. As for the optimization of application, the solution field can be applied from the optimization of single target to multi-target. The development of hardware and software of computers results in the development of the artificial intelligence. All application or research of the algorithms of artificial intelligence shall be in line with unique restrictive conditions and computer programs aiming at individual targets, that is, different issues need different exclusive computer programs. Though software kits based on genetic algorithm have come into existence (such as Evolver, SUGAL and Gene Hunter), however, there are many restrictions, or the efficacy of genetic algorithm can’t be unleashed completely. This study developed a standard program module system with open source code based on genetic algorithm. The concept of LEGO is applied and the program is modularized with the way of building blocks, providing options for the users to combine the computer program in accordance with the restrictive condition. In case the modular system can’t conform to the specific circumstance, the user can modify the source code to satisfy the special requirements accordingly.

並列關鍵字

Genetic Algorithms

參考文獻

16. 葉哲邵，「以Visual Basic開發遠端控制系統於影像量測實驗之應用」，碩士論文，淡江大學，臺北，2007。

5. 吳獻堂，「動態營建資源即時配送最佳化模式之開發-以混凝土配送為例」，博士論文，國立成功大學，台南，2006。

13. 洪暉凱，「建築物外牆磁磚整磚計畫自動化之研究」，碩士論文，朝陽科技大學，台中，2010。

14. 徐瑞旻，「共振柱試驗程式視窗化之研究」，碩士論文，國立成功大學土木工程學系，台南，2002。

19. 蔡永勤，「應用頻率域改良型基因演算法與遞迴式改良型基因演算法於結構動力系統識別」，碩士論文，朝陽科技大學，台中，2010。