基于粗糙集理论的火焰图像处理与状态识别

针对全炉膛火焰检测的主要问题,提出了一种基于粗糙集理论的火焰状态识别方法。该方法利用粗糙集理论的分类原则,得到火焰燃烧的高温区域,结合其它火焰图像的特征量,用基于粗糙集的知识决策系统建立火焰状态识别规则,构建全炉膛燃烧状态识别的基本模型,判断燃烧状态。由于决策表的简单属性表达,有利于操作人员的理解并为调整锅炉燃烧提供操作指导。现场试验表明,火焰图像的特征量与燃烧的好坏有密切关系。通过属性集的定量分类实现输出量的定性识别,系统状态识别准确可靠。

Flame Image Processing and Status Discrimination Based on Rough Sets Theory

A flame status discrimination method based on rough sets theory has been presented in the light of the main problems concerning the flame detection of an entire furnace. The method makes use of the classification principle in the rough sets theory to identify the high temperature zone of flame combustion. In combination with other flame image characteristic values, a knowledge decision-making system was employed based on the rough sets theory to set up flame-status discrimination rules and establish a basic model for the discrimination of entire furnace combustion status to make a judgement of the latter. As the decision-making table is of a simple attribute expression, it lends itself to be easily understood by operators and provides operation guidance for boiler combustion regulation. The on-site tests indicate that the characteristic magnitudes of flame images have a close bearing on the quality of combustion process. The processing based on the rough sets theory can lead to a higher effectiveness and the quantitative classification by adopting attribute sets should help realize a qualitative discrimination of the output values, thus resulting in a precise and reliable discrimination of the system status.