Research On Online Multi-objective Combustion Optimization System Of350MW W-flame Boiler

W-flame combustion technology is widely used for low volatile anthracite, butthis type of design is a work in progress, with a variety of burning issues includingpoor burn-out characteristics, high slagging tendency and NOxemissions. In order toimprove the economic and environmental performance of W-flame boiler, a350MWW-flame boiler is taken as an example for study:Based on others’ research on coal quality soft-sensing models, a model withimprovements has been proposed. The improved model reduces a layer of loopiteration to increase stability. Compared results of the coal quality soft-sensing modelwith the coal quality data from power plant, the relative error is less than5%. So it fitsonsite coal quality monitoring requirements, and is able to provide references forboiler operation.Factor analysis method is used to take air distribution of a W-flame boiler foranalysis. Two independent factors that characterized air distribution will be extractedout from variables such as coal feed quantity and primary air of each mill, booster air,and secondary air damper opening.These factors don’t exist linear relationship, andcan reflect more than99%information of original variables.Taking the results of factor analysis as a pretreatment of boiler performanceprediction model based on least square support vector machine, and compared it withone without pretreatment, the results show that prediction accuracy of two types issimilar, but the former increases the modeling speed by15%~20%. Taking intoaccount the variation of equipment characteristics and coal quality during boileroperation process, sample database needs to be updated.There presents an improvedmethod to update sample data, and contrasts with other common methods, theimproved method has some advantages.On the basis of the boiler performance prediction model with pretreatment,genetic algorithm is made full use of for multi-objective optimization of combustionand a multi-objective combustion optimization system of W-flame boiler is proposed.The system connects to PI real-time database as data source for analysis by LAN inpower plant. And it provides two main functions, which are performance monitoringand combustion optimization guidance. The result of field operation shows, afteroptimization boiler efficiency increases0.41%, and NOxconcentration reduces59.59mg/Nm3.