頁籤選單縮合
題名 | 以神經網路模式診斷鋼筋混凝土建築物之龜裂=A Neural Network Model to Diagnose Cracks in Reinforced Concrete Buildings |
---|---|
作者 | 趙鏡如; | 書刊名 | 技術學刊 |
卷期 | 12:2 1997.06[民86.06] |
頁次 | 頁227-237 |
分類號 | 441.557 |
關鍵詞 | 專家系統; 知識庫; 神經網路; 模糊集合; 回溯傳遞; 因果圖; Expert system; Knowledge base; Neural networks; Fuzzy sets; Back propagation; Cause-and-effect diagram; |
語文 | 中文(Chinese) |
中文摘要 | 以神經網路模式診斷龜裂是結合因果圖、模糊集合及神經網路的概念而成,由於 判斷龜裂原因的知識和經驗都具有模糊性、不確定性及不完整性,不易作有系統的歸納和整 理。因此利用因果圖方法將龜裂的特性和原因予以分類歸納,並以模糊集合描述其特性和原 因的模糊關係,此模糊關係矩陣可以為神經網路的訓練範例,待訓練完成後的神經網路即成 為診斷用的知識庫,來推斷龜裂發生的原因。文中以三個案例說明神經網路模式在推斷龜裂 原因上的可行性。 |
英文摘要 | A neural network model based on the concept of cause-and-effect diagraming, fuzzy set theory and neural networks is presented. The knowledge and experience of diagnosing cracks is fuzzy, uncertaint and incomplete, and it is hard to set up the relationships between the causes of the cracking and its characteristics systematically. A cause and effect diagram was employed to deal with the classification of the relationships between the causes and the effects of cracking. Fuzzy set theory was applied to describe our knowledge of cracking. The fuzzy relation matrices are training data for neural networks, that is, the knowledge base for cracks diagnosis. Three examples are presented to demonstrate the feasibility of the model in diagnosing crack formation in reinforced concrete buildings. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。