Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/2621
 標題: 類神經網路在光彈圖像分析之應用Application of Neural Networks on Photoelastic Field Analysis 作者: 陳昱佑 關鍵字: 類神經網路;光彈;應力量測;影像處理 出版社: 機械工程學系 摘要: 本研究是以倒傳遞式類神經網路的方法分析光彈圖像；將穿透式光彈儀取得的光彈圖像，經由影像處理，取得神經網路所需的資料R、G、B三原色，用這些資料來訓練類神經網路，再用訓練完成的類神經網路來分析光彈圖像，也就是藉由RGB三原色求得條紋階數(fringe order)；利用這個方式，可以讓我們很方便的作應力量測。 在材料力學的書上，可以找到已知的應力場模型，並可以算出應力場的理論值，但是在實際應用上，我們判斷光彈圖像的方式是以人的肉眼判斷，這種定性分析只有對光彈圖像有基礎的人才有辦法作到，而且只是做約略的判斷，不能正確指出該處的應力，若要以定量的方式分析光彈圖像，可以用本研究提出的方法，達到方便與快速的定量分析。This paper describes application of back propagation neural network in photoelastic analysis. We use image process to get training data, R、G、B value, of neunal networks from photoelastic field. When the neural networks has been trained by those data, we can use the neural networks to analyse photoelastic field. We can get fringe orders from R、G、B value. The method this paper offer is convenient for us to measure stresses. We can find stress field model already known by us from the book, named mechanics of materials. It is convenient for us to calculate stress fields from the knowns. Really,we judge fringe order by its photoelastic field in eyes. The qualitative analysis is only used by the man of photoelasticity fundamental and is a probable judgment. We can’t point out exact stresses somewhere. It is convenient and efficient that this method offer us the method analysing photoelastic field in quantification. URI: http://hdl.handle.net/11455/2621 Appears in Collections: 機械工程學系所

TAIR Related Article