神经网络在橡胶硫化温度预测中的应用

本文针对难以通过在线测温实现等效硫化控制的情况，在确定硫化条件的直接测温实验基础上，提出了用神经网络模型预测实际生产过程中硫化制品内部温度的方法．在该方法中将所测温度值作为样本提供给神经网络学习，温度采集时刻及易测外部温度作为网络输入，输出值则是在硫化时制品难测温点处不同时刻的内温．神经网络通过学习取得了良好的效果，网络输出的温度值将成为等效硫化计算和硫化质量智能控制的有价值的依据．

NEURAL NETWORKS APPLIED TO FORECASTING TEMPERATURE IN EQUIVALENT VULCANIZATION OF RUBBER

The ncrharm thermometry is a difficult problem in equivalent vulcanization when thick products are produced. Based on the experiment in which the requirement of vulcanization is decided by directly measuring the temperature, this paper proPOses a method to construct a model ofneural networks. In the proPOsed method, the collecting temperature time and the outside temperature are provided to neural networks as the input patterns while the inside temperature of the product in different spots are provided as the 0utput patterns. Application samples have proved that the proPOsed method is both effective and practical. The output temperature of the neural networks can be used as an valuable evidence in the calculati0n of equivalent vulcanization as well as in the intelligent control of curing quality.