整合智慧式系統在鋼結構寬翼樑設計之應用

Abstract

在電腦輔助工程的發展上，最早先有以傳統程序式的程式來協助完成一些 程序式的問題，而後，又有專家系統的出現以解決一些非程序式的問題。 但，不論是傳統程式或是專家系統，均無法由過去的經驗中自動學習，以 提高下次處理類似問題時的正確性與效率。而近幾年來才又又被人們重新 重視的類神經網路雖然具有上述的學習能力，可是卻嚴重缺乏像專家系統 般清晰易懂的(Explicit)知識表示能力。是以，本研究即在嘗試藉由整合 專家系統與類神經網路之優點，並利用一黑板式介面系統，發展出一套具 有自動學習能力的「整合智慧式系統」。此外，本研究論文係以AISC的 LRFD鋼結構寬翼樑設計規範做為系統驗証的範例。而由範例之執行之結果 ，可顯示本整合智慧式系統在電腦輔助工程上為一成功且可行的發展方向 。 Computer programs have been widely used in the domain of structral design and analysis. However,programs do not learn anything from the data they process;that is , the additional knowledge contained in the results is not feedback to improve the process. The objective of this thesis is to develop an integrated intelligent structural design system by integrating a knowledge-based system(KBS),an artificial neural network model(ANN),and a blackboard-liked knowledge-base interface. The ANN adds accuracy and error tolerance to the KBS and the KBS is used to explain the ANN work behaves as it does and feed the results back to ANN for additional learning. The integrated intelligent structural design system has been implemented in a PC using KAPPA-PC expert system shell and C language. Moreover, the system has been appiled to the domain of detailed structural design steel structures. The results show that the integrated intelligent structural design systems is a possible approach in the future development of computer aided engineering.