Research and Implementation of Children's Wear Model Based on Improved BP Algorithm

Abstract

In order to alleviate the extreme shortage, casual resignation of clothing designers, and the continuous improvement of the need for individual customization, we have come up with an improved BP algorithm which using children's easy-to-measure parts size data to forecast the size of complex parts required for different children's wear. We made and produced the predicted data to meet the actual requirements. This study compares the predicted results of the improved BP algorithm with those of the original BP algorithm and other algorithms, proving that the improved BP algorithm is with stronger stability, higher accuracy and faster convergence speed. The data required for pattern making is predicted by the improved BP algorithm, and the error between the forecast data and the actual data is as high as 99% within 0.3 cm, which is more advantageous than other algorithms in learning various clothing types and realizing monomer customization.