Data Mining In The Mixed Coal Combustion Nox Run Optimization

As one of the main atmospheric pollutants, the nitrogen oxide (NOx) affecting human ecological environment seriously. The NOx coal-fired which power plants discharged is about40%of the total emissions NOx, and become the main source of NOx emissions.Blended coals combustion are mostly use in coal-fire power plants in our country, it makes the emissions of nitrogen oxides more complex to the single coal combust. Although many scholars home and abroad have research the emission of nitrogen oxides. It is mainly aimed at single coal emissions, the research of the mixed coal relatively few. And in the past, most of the emissions studies of nitrogen oxides during blended coal combustion are experimental researches, this method is not only waste time and power plant. A large number of emissions data of nitrogen oxides is produced in power plant, emission characteristics of nitrogen oxides during blended coals combustion were implied in this data, emission data of nitrogen oxides from the power plant are carry on data mining to find the messages hidden in the data.To begin with, emission sample data of nitrogen oxides during blended coals combustion were carried on correlation analysis and feature selection in this paper. It is found that the relationship between the emissions of NOx and CO is negative relationship, and positive relationship between emission of NOx and boiler efficiency, by the correlation coefficient of positive and negative in the related analysis. Because multiple factors influence emissions of NOx, coefficient of partial correlation was to chosen to study the relationship of positive and negative between emission of NOx and influence factors, the result shows that the relationship between emission of NOx and export air temperature of coal mill is positive, the relationship between emission of NOx and the proportion of the bottom quantity of heat and the total's in burner is positive. In the feature selection, the main influence factor of the emission of NOx was to be found out by value of importance index. The results showed that the largest influence for emissions of NOx is types of coal.Next, the influence factors of emission of NOx during blended coals combustion were carried on factor analysis, with10extraction factors instead of57original influence factors, and the build the emission linear prediction model of NOx during blended coals combustion, but the forecast of results is not an ideal value.Then, due to the limitation of the linear mode, emission nonlinear prediction model of NOx during blended coals combustion was created. Before created the mode, division of the simple data had been done by using the cluster analysis and BP neural network predictive model for every division simple data had been created. The result showed that the emissions prediction model of NOx during blended coals combustion is provided with good predictability.Finally, in the optimization efficient and lower NOx-emission association rules model for discrete samples was created, the result showed that the efficient and lower NOx-emission combustion adjustment program included application of constricted air distribution mode, the discontinuation upper coal mill, the favorable over fire air.