Research On The Object Predict Method Of Garment Seam Pucker Grade Based On Fuzzy Neural Network

With the development of garment quality criterions and normal measurement methods in the world and the requirement of Quick Response System in garment manufacture, it is necessary to set up an objective evaluation system for garment quality prediction based on the fabric low stress mechanical properties. Fabric sewing ability is one of important factors that influence the quality of garment appearance and manufacture quality. In this paper, we proposed an effective evaluation method to predict garment seam pucker grade.Garment seam pucker is the wrinkle degree of fabric under low stress, and relates closely with the mechanical properties of fabric. Firstly, in this paper, we use FAST (Fabric Assurance by Simple Testing) system to measure the mechanical properties of commonly using fabric, and apply the correlation analysis method to observe the relations of fabric FAST mechanical properties and fabric sewing ability. Secondly, we use PCA and KPCA method to analyze these high dimension and nonlinear FAST mechanical properties data (reducing dimensions of test data). So we extract the corresponding features (PC and KPC) as the input of neural network. Thirdly, the extracted PC and KPC of fabric FAST data were classified by the algorithm of fuzzy kernel clustering (KFCM) and get fuzzy partition matrix U and clustering center V. At last, we pass the fuzzy partition matrix U and clustering center V to ordinary RBFNN to construct a KFCM-RBFNN neural network by adjust the width and center of radial basis function of hidden nodes. The modification makes the width (receipt field) and center of hidden nodes of RBFNN become more effectively optimizing and controlling. Our experimental results demonstrate that the proposed system could efficiently be used as a fabric sewing ability prediction system with high accuracy and is robust for various structures and mechanical properties of filmy silk and middle thick wool fabric.On the basis of the formers' research, the proposed system by us can be used efficiently to an objective evaluation of garment seam based on KFCM-RBFNN. The prediction results of sewing ability will be used to instruct the garment factory to manufacture effectively. So the prediction system proposed by us improves the market competition capability of garment enterprises.