基于改进光流法的锅炉火焰动态特征提取与分析

Feature Extraction and Analysis of Boiler Flame Dynamic Characteristics Based on Improved Optical Flow Method

锅炉燃烧过程具有连续性与随机性,火焰静态特征仅能表征燃烧过程的部分信息,无法反映燃烧随时间变化的动态过程,不利于长期监测炉内燃烧状态。针对这一问题,提出了一种基于金字塔分层改进的Lucas-Kanade光流法,提取炉内火焰动态特征,用于描述炉内燃烧过程的变化。为了验证火焰动态特征监测炉内燃烧状态的灵敏性和有效性,采用互信息法建立动态特征与锅炉负荷、NO_(x)排放量间的相关性模型。结果表明,所提动态特征能准确跟随火焰的运动轨迹,反映当前运行工况、燃烧状态的变化。采用改进后光流法提取的火焰动态特征与锅炉负荷、NO_(x)排放量间具有显著相关性,能够为锅炉燃烧状态实时监测提供一个更有效的特征。

Boiler combustion process is continuous and random.The flame static characteristic can only partly describe the combustion process but cannot reflect the dynamic process of combustion with time,which is not conducive to long-term combustion state monitoring.To solve this problem,this paper proposed an improved Lucas-Kanade optical flow method based on a pyramidal hierarchical to extract the dynamic characteristic of flame and describe the change of combustion process in the boiler.To verify the sensitivity and effectiveness of the flame dynamic characteristic,the correlation model between the dynamic characteristic and the boiler load and NO_(x) is established based on mutual information method.The results show that the proposed dynamic characteristic can accurately track the trajectory of the flame and reflect the changes of instant operation condition and combustion state.There is a significant correlation between the flame dynamic characteristic and the boiler load and NO_(x).The dynamic characteristic is a more effective feature for real-time monitoring of boiler combustion state.