神经网络技术在预测织物透湿性能上的应用

Artificial Neural Network Technology Applied on Forecasting Fabric Moisure Permeability

用蒸发法测试了几种织物的透湿性能,应用神经网络对织物结构参数与透湿性能之间的关系进行了深入研究。利用MATLAB建立神经网络模型,选择最佳网络参数并对网络进行训练。通过应用实例提出了如何处理试验数据的方法,并通过建立大量的BP网络进行比较,筛选出最具应用价值的网络。用该网络对未经训练的试样进行相对透湿率预测,发现预测结果与实测值非常接近。说明用神经网络对各种织物进行透湿性预测是可行的。

Several fabrics moisture permeability were tested through evaporation method, relation between fabric stitch parameters and moisture permeability was researched in detail through applying artificial neural network technology. Artificial neural network modal was established by using MATLAB, the best network parameter was chose and the network was trained. Dealing method of test data was proposed by applied example, and establishing a number of BP network to screen the applied value network. Relative moisture permeability of non-train assay was forecasted by the network, the test shows that forecasted result is similar to practice result, so it is feasible for artificial neural network to forecasting fabric moisture permeability.