BP神经网络在塑身内衣压力预测中的应用

Application of BP neural network in prediction of shapewear pressure

为预测人体在穿着塑身内衣时产生的压力,提出一种基于神经网络的预测方法。首先测试在4种不同姿势下26名女大学生穿着同一品牌的塑身内衣时,人体20个部位产生的压力值;然后用2种不同的BP神经网络工具箱函数,建立压力预测模型,并分析压力值与人体5个测量值之间的关系,最后检验并比较2种工具箱函数以及不同输入因素对压力的预测效果。结果表明:该预测方法无需复杂的建模计算过程,可直接利用人体的5个测量值较为准确地预测塑身内衣对人体产生的压力值; newff函数的预测效果优于feedforwardnet函数;对2种站姿时的预测精度优于2种坐姿;增加站姿的压力值作为预测输入因素,可准确预测其他3种姿势下的压力值,其预测精度达到82%以上。

In order to predict underwear pressure when wearing shapewear,a prediction method based on neural network was proposed.A total of 26 female college students wore the shapewear of the same brand,and were asked to perform four poses.The pressure values at 20 body points of each subject were measured,respectively.Two different BP neural network toolbox functions were used to establish the pressure prediction models.The relationships between the pressure values and five body measurements were analyzed.The prediction results were tested and compared while using the two different toolbox functions and different input values.The results show that the method based on neural network does not need complicated calculation process.By this method,five body measurement values can be directly used to predict the pressure value of shapewear.The prediction accuracy of the newff function is superior to that of the feedforwardnet function.The accuracy of the prediction of pressure in the two standing poses is better than that in the two sitting poses.When taking the pressure values of standing pose and five body measurements as input values,the pressure values of the other three poses can be predicted with the accuracy over 82%.