Research On Clothing Pressure Prediction Based On Improved BP Neural Network

In the field of clothing ergonomics,the problem of clothing pressure prediction is related to clothing comfort and human health,and it is a hot topic for clothing experts and scholars.Its purpose is to establish the relationship between clothing pressure and human body characteristics by analyzing the factors affecting clothing pressure,so as to timely adjust the comfort of clothing to ensure that it does not affect human health.Due to the successful application of BP neural network algorithm in various fields in recent years,but less application research in the field of clothing,therefore using BP neural network to predict clothing pressure is a worthy research direction.Clothing pressure is often affected by the physical characteristics of the individual.Due to different personal physical characteristics,the data collected are also different,huge and complicated,which leads to the complicated relationship between clothing pressure and physical characteristics of the human body.This paper proposes an improvement new algorithm combining the BP neural network and PCA solves the difficulties in the clothing pressure prediction process.The research in this paper is mainly:Firstly,for the problem of clothing pressure prediction,an improved BP neural network is proposed:additional momentum-elastic gradient descent-adaptive learning rate method.This algorithm is improved to solve the problems existing in the traditional BP neural network algorithm,such as the selection of learning step,the size of weight,the difficulty in determining the direction and the difficulty in controlling the learning rate.And simulation experiments verified that the improved BP neural network compared with the traditional BP neural network,the average error rate is reduced by 5 percentage points.Secondly,in order to reduce the network size of the improved BP neural network,improving the computational efficiency and prediction accuracy,this paper proposes an improved PCA-BP neural network model.The algorithm firstly conducts PCA treatment on the input sample,reduces the dimension of the input sample,and then uses the improved BP neural network to predict the clothing p ressure.Experimental results show that the proposed PCA-BP neural network algorithm in both scale and computing time has made very good improvement.At the same time,the convergence of the network has also been Improv ed,The clothing pressure compared with the measured values and predicted values of pure Improv ed BP neural network algorithm,the average relative error is decreased by 4.35%,the average absolute error is decreased by 0<3 1,the running time is reduced by 10.407 seconds.