Abstract:

Objective Under the background of digital economy, the popularity of individual needs promotes the diversified development of clothing styles, which brings new opportunities and challenges to clothing production. At present, the clothing market as a whole presents a "multi-variety, small batch, short cycle" production mode. In order to reduce production difficulty caused by excessive style changes and to reduce production costs, industrial customization is oriented to customer demand while taking into account the production speed and economic benefits, in which modular production is one of the effective means to achieve this production mode.

Method Using the fuzzy clustering of equivalence relation, the method of building module group of shirt processing components was achieved. The typical shirt styles produced in recent years were taken as the research object, the common styles in production were sorted and classified. The main shapes and processing methods were summarized, the processing modules and stitch types were classified and coded, and the shirt modules were divided under the production situation of short flow.

Results After the basic module group and processing technology are summarized, the classification of clothing modules is quantitatively analyzed and studied. First of all, the complex process is preliminarily screened. Processing examples of different modules : 0 indicates that the module does not use this process, 1 indicates that the module will use this process for processing. The truncated matrices under different λ thresholds are established by fuzzy hierarchical matrix. The modules are clustered from large to small, and different truncated matrices are divided into different truncated matrices. A total of 19 kinds of clustering results were obtained for all types of parts (parts) modules, with a total of 42 parts (parts). On the basis of preliminary screening, fuzzy F-statistic formula was used to calculate the corresponding values of different clustering results. The optimal solution is obtained when the module group of shirt production process is divided into 11 classes. According to the results of F-statistic quantitative analysis, the division of the final module group clustering results is obtained.The final clustering results are basically consistent with the actual production, and the module processing technology in the same module group is basically similar.

Conclusion The theoretical method of this research is extended to cost accounting, wage payment, quality assessment and other aspects, and provides certain reference value for the production of clothing production arrangement, construction period forecast and other production links. In the following research, we will focus on the research direction of module family time prediction based on BP neural network and the optimization application of module production scheduling for mixed mode components.

Key words: shirt, clothing style, process similarity, fuzzy clustering, module group, modularization production