Traditional pattern making models are mainly linear models. This kind of model has many shortcomings. As Back Propagation (BP) neural network using simple nonlinear transfer functions can approximate any nonlinear functions with any precision, we proposed a BP neural network model to predict all pattern making-related body dimensions by inputting few key human body dimensions. Sixty students in the northeast of China were measured for collecting a learning data to train the proposed model, and eleven of the sixty subjects' body dimensions data were applied to test the accuracy of the model. The results show that the prediction accuracies of linear regression model and BP neural network model have little difference. As the traditional linear model can be well applied in pattern making, the BP neural network model also can be well used for pattern making. Moreover, if increasing the number of learning samples, the precision of the proposed model is further improved.
展开▼