Study Of Flame Inside The Furnace Detection Algorithm

With the great development of our industry, the safe running of the furnaces in a big enterprise such as a power plant is getting more and more attention. Owing to unstable combustion or improper operation, a part of or all burners would go out. This moment, if the coal powder is fed continuously to the burner, the accumulation of unburned coal power will result in an explosion, which poses a grave threat to the safety and lifespan of furnaces and other relevant equipment. Therefore, the safe running for furnaces has much respect to lives and property, attracting an exclusive attention of all large enterprises. Now, the main power set in our electrical industry utilizes mainly the furnaces inside which four burners are configured at four corners tangent to a circle, so it is important to detect the state not only of the flame of the whole furnace but also of each single burner inside the furnace.Based on the flame detection method using digital image processing, this dissertation analyzes firstly current flame detection methods, demonstrates their advantage and disadvantage after test on the flame image data supplied by Nanchang Guodian Longyuan Combustion Company. Then, the frame difference (FD) method is proposed and applied to flame images of both the coal burners and the oil burners. Experimental results show that the FD method is simple and effective. Additionally, the fuzzy C-means clustering (FCM) method is analyzed and proposed to be applied to detect flame images. Experimental results also show its effectiveness and simplicity. Finally, the theory of neural network (NN) is briefly introduced, and the NN based flame detection method is utilized. The dissertation gives a detail training procedure and realization code, with test results on current flame image data, which also show the NN based method can effectively detect the state of flames.In the last part of the dissertation, the hardware and software realization principles are introduced. The software development environment is Visual C++6.0 under Windows XP. The Code Composer Studio (6000) is utilized to develop DSP programs and to realize the communication between DSP and the development environment.