Development Of The Platform Of The Textile And Apparel Fabrics

With the extensive use of composite materials and new textile technology, textile and garment fabrics with different fabric structure parameters and fabric style appear more and more frequently. In this paper, the author analyzes the relationship between the fabric structure parameters, mechanical parameters of fabric, fabric style and fabric properties of fabrics, and establishes a textile fabric information platform using the integrated neural network method. The author also achieves the two-way predictive function between "the parameters of fabric structure" and "fabric style subjective score", and the similarity search function based on the "fabric structure parameters".The author selects the component proportion, warp and weft yarn density, weight, thickness, width, fabric as the structural parameters of the fabric, and selects the main colors, texture, pattern, gloss, drape, smooth, soft and transparent as a subjective evaluation index of fabric style of fabric. The author cleans up the errors and redundant data of "fabric structure parameters" and "subjective evaluation of fabric style", and quantify all the descriptors and converse all the units into the general unit. The anthor puts forward a method of "unified encoding" which reflects the fabric type and the serial number in database.In order to improve the prediction accuracy, the paper adopts the integrated neural network which is composed of a "Classifier" and a "neural network", and connected by a single-layer perceptron network and multiple single linear neural network. It's necessary to divide "textile fabric sample set" and "synthetic textile fabric sample set" into subset in accordance with "unicode" and "complex coefficient" as basis.In this paper, the collaborative filtering algorithm based on the structure parameters is used to realize the function that finding the "similar fabric" to the "target fabric". The function uses fabric structure parameters to calculate the similarity between the target fabric and other fabrics, and find the most nearest k similar neighbors. Information platform can also be in accordance with the "variety of search conditions + keyword" way in all the samples for fabric retrieval.