The Analysis Of Boiler Combustion Stability Based On Flame Images

Boiler combustion stability determines whether the thermal power units can run economically and safely.Because of the complicated combustion condition in the boiler,the widely used detection measures is not able to detect the combustion stability in real time.The flame image can show the combustion condition inside the boiler directly.It has great potential for analyzing the combustion stability.In this paper,the relationship between the characteristics of flame image and the stability of pulverized coal combustion in coal-fired boiler is analyzed.The fluctuation index of the combustion zone,the similarity of the edge.contours of the pulverized coal and the difference of the contour length of the pulverized coal are proposed to measure the combustion stability.By combining these three characters with K-means algorithm and RBF neural network the stability of pulverized coal combustion inside the boiler can be recognized.The combustion of pulverized coal in the flame image forms three combustion zones in the image from entering the boiler to completing combustion.The fluctuation index can reflect the change of the position of the combustion zone.Based on this,the fluctuation of the combustion areas over time can be analyzed.According to the coincidence area of different combustion zones,weighted sum can be obtained as the fluctuation index of the combustion zone.Fluctuation of the contours of the pulverized coal in the image can also reflect combustion stability.So this paper looks for the edge contour of coal powder,by compare the contours of two frame,the similarity of the edge contours of the pulverized coal and the difference of the contour length of the pulverized coal can be calculated.Due to the complex combustion conditions inside the boiler,the extracted characteristic values fluctuate greatly.So in this paper,the mean,variance,maximum and minimum difference of three characters in a specific time period are taken as the input of the RBF neural network to judge the combustion stability in the current time period.The experimental results show that the method has good effect on the judgment of boiler combustion stability.