BP神经网络预测降落伞绸透气性的研究

A study of prediction of air permeability of the parachute fabric by BP neural network

透气性是降落伞绸的重要性能指标。降落伞绸织物的纱线线密度和织物密度、厚度、面密度及织物组织等结构参数与其透气量存在非线性的关系,由于织物结构各参数间的非线性关系及多个影响因素共同对织物的透气性能的作用,织物结构参数与其透气性能之间很难用传统数学、力学的方法来描述。人工神经网络能够处理复杂的非线性关系。本课题研究了利用人工神经网络来预测降落伞绸织物的透气性能,以期代替传统的测试方法,从而在新品种设计时对织物的参数设定提供指导。

The air permeability is an important parameter describing the fabric function of parachute. It is influenced by multiple factors of the fabric attribute such as fabric density and thickness, threads per unit length, the fabric area density, and weave structure, etc. The relationship between the air permeability and these factors is nonlinear. The nonlinear relationship is difficulty to build by traditional mathematical and mechanical methods. By taking advantage of BP neural network method that can handle the complicated non-linearity, this paper studies the prediction of air permeability based on fabric attribute, trying to introduce a new method that advances the traditional ones for parachute＇s fabric design.