Abstract:

Objective For the rapid reconfiguration requirement of multi-variety and small-batch clothing production system under the characteristics of modular intelligence. A method based on back propagation (BP) neural network was proposed to predict module man-hours and optimize the application of mixed mode component module production scheduling. The research results can be utilized to optimize production scheduling, predict man-hours and assign processes, and provide reference for quick quotation and production planning.

Method shirt; Taking shirts produced by an enterprise as the research object, a sample set of shirt module man-hour was established, and the influence factors of standard man-hour were analyzed to build a man-hour prediction model. Production of two shirts of the same color and different styles was taken as an example to achieve the arrangement of production by using modules in the mixed assembly line, and the arrangement effect was analyzed.

Results In order to measure the accuracy of prediction results more intuitively, the man-hour prediction model was constructed for all module groups and verified one by one. Based on each evaluation index, the prediction accuracy of the model was evaluated, and the prediction results of 11 types of module groups were obtained. From the perspective of model fitting effect, the accuracy of fit of all module groups was above 0.81. From the perspective of the prediction time value of test samples, the average absolute error of each module group was within 9 s and the error value of 8 module groups was not more than 5 s. The model prediction accuracy rate reached more than 90% peaking at 94.89%. Taking the combinatorial splitter module of class 8 module group as an example, the scatter plot was adopted to compare the real value and predicted value data of the test set samples. The values of the two samples were close to each other for most samples, and the error was within the range of ±10 s. The paired sample T test was adopted to analyze the error between the actual value and the predicted value of the class 8 main process, and the significance Sig. value was greater than 0.05, indicating no significant difference between the two, and the multidimensional proof model achieved good results in the prediction of standard man-hour. Finally, the process allocation of the hybrid production line was achieved based on the module group for the two shirts. The results showed that the average absolute error of each module group was within 9 s and the error value of class 8 module group was not more than 5 s. Compared with single-style production, the efficiency of mixed production mode was more than 90%, the efficiency of modular optimization reached 95.55%, the balance delay rate was reduced by 44.04%, and the smoothing index was reduced by 50.09%.

Conclusion A method based on BP neural network is proposed to predict the man-hour of module group and apply it to the production scheduling optimization of mixed mode components. Based on the 11 types module group of shirt parts, the BP neural network prediction model is built through the weight analysis of the influence factors of standard man-hour. In order to verify the accuracy of the prediction results, the man-hour prediction model was constructed for all module groups, and the optimal network structure was obtained by verifying one by one. The paired sample T-test was conducted for the actual and predicted values of each process of the class 8 module group, and the results showed that Sig. values were all greater than 0.05, indicating no significant difference between them. Finally, based on the module group, the process allocation of the hybrid production line is realized for the two shirts. The results of this study can be used for production scheduling optimization, man-hour prediction and process allocation, which can meet the application requirements of mixed model processing in enterprises to a certain extent, and provide reference for rapid quotation and production planning.

Key words: shirt, BP neural network, standard man-hour, man-hour prediction, module production, flow arrangement