The Analysis Of Boiler Combustion Stability Based On Flame Images

This paper mainly analyzes the flame image generated by pulverized coal combustion process in the boiler,so as to extract the characteristics that can reflect the combustion state of the boiler,and distinguish the combustion stability of the boiler.This paper mainly studies the extraction methods of characteristic parameters such as black dragon length,eccentricity distance,fractal dimension and burning area,and takes the characteristics such as black dragon length,eccentricity distance,fractal dimension and burning area parameters as the characteristic indexes of boiler combustion stability discrimination.The length of black dragon in the unburned area in the flame image reflects the movement of pulverized coal inside the furnace from the nozzle to the beginning of combustion,and can reflect the stable combustion state of the boiler.The region and direction of black dragon length detection were determined according to the incidence direction of pulverized coal airflow.The difference algorithm was applied to search the pixel points with the maximum gray gradient change in the detection area,namely the black dragon edge points.The length of black dragon on the flame image was determined by the coordinates of the edge points.The flame produced by pulverized coal combustion in the furnace is a random natural phenomenon without regular morphological changes,and it has certain self-similarity.Fractal theory is used to describe those phenomena that have no specific length,but have fine structure,which can explain the furnace flame.In this paper,based on the fractal feature of the flame image segmentation method,the burning area of the flame image is segmented from its background,and the burning area of the segmented image and the fractal dimension of the original flame image are extracted as the characteristic parameters to judge the flame combustion stability.The combustion fluctuation of primary air and pulverized coal in the furnace will directly affect the flame,and the degree of flame fluctuation can reflect the stability of combustion in the furnace.The eccentricity distance is an embodiment of the degree of flame fluctuation.The traditional measurement of eccentricity distance has strict requirements for the reference point.With the operation of the unit,the reference point will be offset,resulting in errors.In this paper,the distance between the center of mass in the complete combustion zone and the center of mass in the combustion zone is extracted as the eccentricity distance,which avoids the difficulty of selecting the reference point and takes the eccentricity distance as the characteristic parameter to judge the flame combustion stability.In view of the complexity of combustion condition in furnace,this paper adopts the method of limit learning machine to judge the combustion stability of flame image.In this paper,the length,burning area,fractal dimension and eccentricity of the black dragon extracted from the flame image are selected as the characteristic indexes,and the mean value,maximal-small value and variance of these four characteristic quantities in a specific period are extracted as the input vectors of the discriminant model.In order to improve the recognition rate of the model,cuckoo algorithm was used to optimize the model.The experimental results show that this method is effective in judging the combustion stability of boiler.