Study On NO_x Emissions And Efficiency Of Industrial Boiler: Mixed Modeling, Prediction, And Optimization

Although the multiform heating modes will make great process in the future, at present, most of the heat energy in China is consumed by small-sized and medium-sized industrial boilers using fossil fuels. As the main heat source of production and life, the industrial boilers are widely used in many industries such as metallurgy, mechanism, chemistry, textile, energy, and so on, and will not be replaced by the utility boilers in one or two decades. However, two problems - high pollution of flue gas emissions and low efficiency - exist universally. So, it is very significant to reduce flue gas emissions and improve the efficiency of the industrial boilers by adjusting the operation parameters without reconstruction or installing new instruments.In order to meet the requirement of high efficiency and low emissions in boiler operations, a mixed model is developed based on experimental data and combined with Back Propagation Neural Network (BPNN) and boiler thermodynamic calculation. This model uses the adjustable operation parameters of industrial boiler as the input, uses NOX emissions and efficiency of industrial boiler as the output, and achieves prediction of NOX emissions and calculation of efficiency successfully. In the model, a new method of calculating fuel consumption, one of difficult problems in this area, is proposed combined with grate structure and rotation speed of grate motor. All of them above are innovation in my thesis.Based on the above model, the test data of industrial boiler are divided into two parts, including training samples and test samples used to train and test the BPNN, which has been improved by regularization that modifies the performance function. After BPNN is trained and tested well, a optimization modeling is proposed, whose object function is the efficiency and whose independent variables are rotation speed of air blower and rotation speed of grate motor. Then, the optimized efficiencies are obtained by using genetic algorithm and compared with those by using exhaustive method. Results show that, the generation of BPNN is improved greatly, the accurate prediction is achieved by using the regularization method, and the boiler efficiency on basis of the modeling is very near that on basis of direct calculation of experimental data. Feasibility of building a mixed model combining NOX emissions with boiler efficiency is proved. Then, genetic algorithm whose results are completely consistent with thoseby exhaustive method, is proved to be feasible in the case of the modeling, which optimized rotation speed of air blower and rotation speed of grate motor and increased the boiler efficiency. Based on the work above, a valuable thermodynamic calculation software is developed by using VB 6.0, which is applied to superheated steam chain grate boiler and saturated steam chain grate boiler, in order to help calculating boiler efficiency in the modeling and show the change of parameters of all heating surfaces after optimization.