基于神经网络的织物剪切性能的预测研究

Predictive Study on the Shearing Property of Worsted Fabrics Based on Neural Network

探讨织物剪切性能和结构参数之间的关系,在33种精纺毛型织物实验数据的基础上,利用BP神经网络方法,建立了精纺毛型织物剪切性能与织物结构参数之间的网络模型,并对模型进行验证和评估。结果表明可以用神经网络来预测织物剪切性能指标,限于样本量,预测平均误差率在15％左右。

A neural network method was used to discuss shearing property. Neural network models between the indexes of shearing property of worsted fabrics and fabric structure parameters were established based on the data of 33 kinds of worsted fabrics. Then the predictive accuracy of the models was validated. The result shows that the shearing property can be discussed by neural network. The accuracy can reach 85% in the case of the samples is not enough.