Detection Of Combustion Stability Of The Medium Furnace Flame Based On Image Processing

The medium furnace as the primary heat transfer equipment has been widelyused in the oil pipeline of crude oil transportation. The stable combustion of themedium furnace flame is a basic requirement of the equipment safe running. If theflame is instable, it will decrease combustion efficiency, generate noise and exhaustgas and so on. If the flame combustion is eccentric, it will cause imbalance of thefurnace temperature field and affects the medium furnace circulatory system. If theflame combustion extinguishes, it can easily lead to security incidents and everyextinguished may cost a lot. In order to promptly and accurately determine theburning condition of the furnace flame, this article used digital image processingmethod to detect the flame state.This article had designed a set of simple and feasible medium furnace flamedetection systems. The video capture program, which acquires the flame images andtransmits them to the control room, run on ARM-based embedded Linux system.While the control room mainly completes detection of the flame conditions. First,preprocess of the image, including removing noise with fast median filter, doingcontrast stretching with the arc curve and segmenting image with marker-basedwatershed algorithm. Then, obtain the flame zone based on the statisticalcharacteristics of the furnace flame. Finally, determine the flame burning condition bydetecting the flame burning character value.The innovation of this article is that we proposed a new character value flamestabilization area rate and improved the original eccentric distance algorithm, for theinsufficient of the existing flame character values in the detection of the stablecombustion position of the flame. We proposed a new concept in the analysis of thestatistical properties of the flame: the flame combustion stability region and flamecombustion critical zone. According to the ratio of the two regions area, we can obtainthe flame stability area rate and according to the center point of the two regions wecan calculated the flame eccentric distance.After the experiment analysis, the main advantages of this article method in theflame detection compared to traditional methods lies in three aspects: Firstly, it canaccurately extract the flame burning area. Secondly, the character value of flamestabilization area rate strengthen the detection of the flame volatility. Thirdly, it is more effective to detect flame eccentric with improved eccentric distance. Moreover,the detection process no longer related with the furnace center point, which avoidedmarking process of the furnace center and simplified user’s operation.