| Sustainable development of energy utilization needs clean combustion technology. According to the advantages of circulating fluidized bed (CFB) technology, such as high efficiency, low pollution, and wide adaptability to coal, more and more researchers pay attention to it, and it becomes a relatively mature technology among clean combustion technologies at present. Stable and uniform combustion fame in furnaces is a basic requirement of boiler normal operation in power plants, so reliable and effective combustion detection technologies are very important to the safe and economic operation of utility boilers.In order to optimize the combustion status in boilers, it is necessary to detect combustion process exactly and establish combustion diagnosis and optimizing operation system based on it. Flame temperature and bed height of CFB boilers are two important parameters for boilers' continuous operation. Based on visual radiation image processing, a set of online combustion detection system in CFB boilers was designed in this paper, including the measurement of flame temperature and emissivity. At the same time, a simulating study of bed-height detection in CFB boilers was carried out based on image processing and BP neural network technologies, and a prediction model of heat absorption rate in CFB boilers was proposed.Firstly, we have developed a set of online combustion detection system in a CFB boiler. The mainly work includes hardware selection, software design and system debugging. Through this system, flame temperature and emissivity could be calculated by the flame image detectors mounted on the walls and displayed on monitors as visualization ways. The detection results could provide good reference for combustion adjustment of the boiler.With the better knowledge to the working principle of color CCD (Charge-Couple Device) cameras and image processing technology, a simple measurement method of temperature and emissivity for coal-fired flames from visible radiation image was presented in the paper. With the assumption of gray radiation for coal-fired flames, the relations between the temperature and the ratio of any two of the three primary colors (red and green are selected in this paper) would be got by the black-body calibration, and the three primary colors (red is selected in the paper) captured from a black-body also could be obtained by the same method. So the temperature of the coal-fired flames could be calculated by the ratio of two of the three primary colors, furthermore flame emissivity would be got. Comparing with the traditional colorimetric method, it is not necessary to know the spectroscopic characteristics of the CCD cameras and complex system structures for this method. Experiments conducted on a coal-fired circulating fluidized bed boiler with a steam capacity of 480 t/h shows that the results by this method were in accord with the actual operation condition of the CFB boiler. The average temperature measured by this method was validated by that got by a thermocouple, and the agreement of them was good with the largest difference less than 10%. So the method could meet the accuracy demands of industrial utility.After the analysis of influences between bed height and flame emissivity in CFB boilers, a detection method of CFB bed height from flame emissivity was described in the paper. As we know, it is a nonlinear relationship between bed height and flame emissivity in CFB boilers, and this relationship could not be described by mathematical model. By use the good ability of BP neural networks which can solve nonlinear problems, simulating researches were carried out under the four conditions to verify the feasibility and convergence of this method. Firstly, the flame emissivity were got by flame image detectors mounted at the different height of a CFB boiler, and then the bed height of the CFB boiler could be obtained by the BP neural network. The simulating results show that the bed height detected by BP neural network was good with the actual values if the optical depth was not great. When the optical depth was great to a certain extent, the detected bed height was very bad. Even so, the method presented in the paper is a new thought for the detection of CFB boilers' bed height, and has some research values.Finally, combined the achievements of previous studies about heat transfer in CFB boilers and image processing technology, an online prediction model of CFB boiler heat absorption rate was established. From this model, the boiler heat absorption could be detected by flame temperature and emissivity calculated by flame image detectors. Also, the predicted heat absorption rate was prior to that got by unit's parameters. The experiments of main-steam temperature control based on radiation energy signal were conducted on a 480t/h CFB boiler. On the other hand the experimental results proved that the predicted boiler heat absorption rate has great foresight, and the control strategy based on this prediction model would improve the response speed and adjusting quality of unit's control system.
|
PDF Full Text Request