基于BP网络的旋流燃烧器火焰燃烧状态识别

Swirl Burner Flame Recognition Based on Backpropagation Neural Network

通过对所采集的电站锅炉旋流燃烧器火焰图像与直流燃烧器火焰图像的对比分析,在原有直流燃烧器火焰图像判据的基础上,讨论了旋流燃烧器火焰图像特征量的提取与特征区的选定,运用现代人工神经BP网络智能理论,设计并训练了BP网络实现旋流燃烧器燃烧状态实时判断的功能。通过实验证明在旋流燃烧器火焰图像上提取的特征量和燃烧状态之间存在映射关系,通过所建立的BP网络模型可以对旋流燃烧器的火焰燃烧状态进行实时判断。

Based on the original flame image criterion for parallel flow burner, comparing and analyzing the flame image of swirl burner and tangential burner using for power plant boiler, character values are distilled from and character sections are chosen in the flame image of swirl burner. The theory of artificial neural network is applied in this article. Back propagation （BP） Neural Network is designed and trained to discern combustion state of swirl burner at the present time.