论文部分内容阅读
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 ,取得了满意的结果
Firstly, the problems existing in the error correction of the colorimetric temperature measurement method based on the flame radiation image are analyzed. Based on the good properties of the BP algorithm in the approximation of the function, a colorimetric test based on the neural network approximation mapping is proposed Warm method. The law is now used in practice: the neural network training into the end of the measurement system, and in a fluidized-bed boiler fire detection system for temperature measurement test, and achieved satisfactory results