| The coal acts an important role in our country's energy structure. In the utility boilers using coal as the input fuel, the uneven combustion, the bias of the flame center, the rushing of the flame to the walls etc., which result from the unreasonable organization of combustion condition, are the important reason leading to the boiler combustion operation failure such as fouling on the furnace wall, the leakage of the water-wall tubes, the flame extinction and the explosion of the furnaces. The image processing techniques had a successful application in the combustion diagnosis of utility boilers, so it is much meaningful to apply the technique into the boiler furnace's monitoring systems to show the flame's temperature distribution and optimize the combustion control in furnace in order to improve the boiler's economical efficiency and security. In this paper, the combustion diagnosis and control based on the radiative image processing techniques in a 300 MW twin-furnace once-through boiler had been done.Because of large scale and intensive pulsation of combustion flame in furnace, it is difficult to realize the 3-D temperature visualization in a utility boiler. Because The number of 3-D temperature values of a twin-furnace boiler is twice as those of a single-furnace boiler and the two furnaces is linked each other, it is more difficult to reconstruct the 3-D temperature distribution in the twin-furnace boiler. Based on the research findings of the research group led by Prof. Zhou Huai-chun, Huazhong University of Science & Technology, this difficult problem can be solved effectively: at first, the radiative imaging information could be obtained using the DRESOR method; second, the linear relationship of the 3-D temperature distribution and the flame radiative temperature image could be constructed using the radiative temperature imaging model; in the end, an improved Tikhonov regularization method was used to reconstruct the 3-D temperature distribution in the twin-furnace boiler. The simulation study showed that the reconstruction result was good even with the reconstruction error up to 0.11.The hardware and software design of a monitoring system in the twin-furnace boiler are introduced in detail. Through the system, the 3-D temperature distribution in the furnaces could be visualized in industrial computer based on the BCB and OpenGL hybrid programming and could be visualized in the utility unit's DCS control system based on the fieldbus for the first time. The refreshment ratio of the visualization is less than 5 seconds. It helps power plant operators to see the combustion condition such as the position of the flame center, the value of the average temperature and the bias of the flame center, et al more directly and to enforce the combustion adjustment more effectively.The multiple layer flame emissivity and characteristic parameters of the 3-D temperature distribution (include the average temperature, the temperature center coordinates of each cross section in furnace) had been measured under different combustion condition. The measurment results showed: the each layer flame emissivity and each layer average temperature is proportional to the unit load; under varied layer fuel quantity, the layer flame emissivity and layer average temperature increase at the areas where the coal density increase; when the oxygen contents decrease, the flame emissivity of the lower two layers increase and the flame emissivity of the higher two layers decrease a little, the radiative heat exchange of the flame center areas increase and the average temperature in full furnace increase at the same time. Based on the typical measurement samples, the dynamic RBF neural network model which using the coal consumption, primary air quantity and second air quantity of each burners as the input variables and the the characteristic parameters of temperature distribution as the output can be constructed. The dynamic model has a good adaptability and overcome the disadvantages of current such models.The Multiple Radiant Energy Signals (MRES) which reflect the combustion intensity of different areas in furnace are obtained by the computer image processing techniques through several Charge Coupled Devices (CCD) cameras installed in the boiler furnaces. In this paper, the idea of using the MRES as the combustion detailed diagnosis tool was put forward. By the analysis of a flame-extinction incident it could be shown that MRES reflects the incident more quickly and comprehensively than other parameters of the boiler, so MRES can play an important role in the combustion diagnosis of large scale coal-fired boilers. The similar incident has not happened after taking the corresponding protective action.A new algorithm of fuse MRES into Integral Radiant Energy Signal (IRES) had been put forward. IRES can maintain the maximum correlation with the unit load. The correlation coefficient between them is more than 0.92 and the IRES varies 30s earlier than the unit load. So IRES processed by this algorithm is suitable to been introduced into the unit's control system. A new control method of tuning the combustion cascade control system's PID parameters using the RBF neural network is put forward. The result of the simulation research shows that the cascade control system based on the new method can respond the step input of the main steam and overcome the fuel disturbance more quickly and more effectively than the traditional combustion cascade control system. It is featured by strong robustness and self-adaptability.The principle of air-coal ratio optimization control were introduced in this paper. The experiment reflecting the associative relationship of oxygen content and IRES in the boiler was done and the results were applied into the scheme design of air-coal ratio optimization control circuit. The optimization experiments under the 260MW stable load and from 295MW to 280MW varied load conditions showed that the maximum IRES can be obtained automatically and the boiler efficiency increased 0.11 and 0.35 percentage respectively.
|
PDF Full Text Request