重組關鍵力學參數以建立棉織物懸垂係數之高預測能力的模式

The Key Physical Parameters Restructured to Establish an Adaptive Model for the High Predication of the Drape Coefficient of Cotton Fabric

王冰凝(P. N. Wang) ； 林哲彥(J. Y. Lin) ； 石天威(T. W. Shyr)

《紡織綜合研究期刊》 18卷3期 (2008/07) Pp. 1-7

https://doi.org/10.6439/TTRJ.200807.0001

摘要

本文以棉織物實證Colleir、Hu、Shyr、Niwa、和Matsudaira等五位學者所提出，以KES-F 關鍵力學參數以及關鍵組合力學參數對靜態懸垂係數所建構之統計模式。發現除了Hu所建構之模式的擬合能力較弱以外，其餘四種模式均有好的擬合能力。進一步利用顯著性分析來檢定上述五種統計模式中參數的顯著水準，將具有顯著性的參數以迴歸方法組成新的模式，再以複相關係數及殘差均方根值來比對不同模式的合適性。本研究發現以B、T、2HB、2HG/W、(G/W)1/3、和(2HG/W) 1/2 等六項具有顯著性的參數，以迴歸分析法建立新模式之複相關係數（R）為0.907，殘差均方根值（RMSE）為 0.0279。此新模式較前述五種統計模式之複相關係數為大，殘差均方根值為小，預測棉織物之懸垂係數顯著高於文獻中其他模式。

關鍵字

懸垂係數 ； KES-F系統 ； 顯著性分析

並列摘要

Five regression models, proposed from Colleir, Hu, Shyr, Niwa, and Matsudaira, for the drape coefficient of fabric with different mechanical parameters were verified using cotton fabric first. Results show that four models are good curve-fitting excepted Hu’s logarithm. The significance analysis of five proposed regression models was proceeded. The significant parameters of each proposed model were screened. Those screened significant parameters, such as B, T, 2HB, 2HG/W, (G/W) 1/3, and (2HG/W) 1/2, were used to establish an adaptive model. The multiple correlation coefficient, R, and root mean square error, RMSE through regression analysis of the adaptive model are 0.907 and root mean square error 0.0279, respectively. The multiple correlation coefficient of the adaptive model is significantly higher than the proposed models and the root mean square error of the adaptive model is significantly lower than the proposed models. It indicated that the adaptive model, based on the integration of the screened significant parameters of the proposed models, has an excellent prediction for the drape coefficient of the cotton fabric.

並列關鍵字

drape coefficient ； KES-F system ； significance analysis