基于神经网络的加热器建模方法

Modeling method based on neural network for heater

基于带有回归单元的Elman神经网络对腈纶生产中的加热器进行建模 ,采用一种带惯性项的动态反向传播学习算法 ,克服了通常的动态BP算法振荡和收敛速度慢的弱点 .实例表明 ,利用该方法迭代后的学习结果最大相对误差为 0 .1%

Based on Elman neural network with recursive structure, the heater model of acrylic fibres production was set up. The dynamic back-propagation learning algorithm with ineria was adopted. The shortages of the back- propagation algorithm such as vibrating and low speed convergence were overcame. Practice shows that the method is better in performance.