人工神经网络在火焰图像比色测温法中的应用

APPLICATION OF ARTIFICIAL NEURAL NETWORK TO FLAME COLORIMETRIC TEMPERATURE MEASUREMENT METHOD

首先分析了基于火焰辐射图像的比色测温法在误差校正方面存在的问题 ,然后利用神经网络的BP算法在函数逼近方面的优良性质 ,提出了基于图像的神经网络逼近映射关系的比色测温法。该法现已用于实践 :将训练结束的神经网络融入到测量系统 ,并在某流化床锅炉火检系统中进行了温度测量试验 。

Firstly problems of error correction of colorimetric temperature measurement method based on flame radiation image are analyzed.Then with the advantage of neural network BP arithmetic in function approach,the corresponding measurement method is proposed.This method has already applied in engineering practice.The neural network of training results is combined with the measurement system,and a temperature measurement test on one FCB boiler is carried out,and satisfied results are achieved.