Considering factors as the pollutant formation

pollution discharge punishment and pollutant control devices

the coal blending method for a 300 MW unit was optimized.On the basis of the commonly used coals of this unit

the support vector machine(SVM)was applied to establish the pollutant formation prediction model for coal-fired boilers.Moreover

the above build model was trained and verified using the boiler actual operation data.Then

combing with the coal rank

the genetic algorithm was employed to optimized the coal blending scheme

taking the lowest operation cost as the goal.The results show that

this model has high prediction precision.After the optimization

the ammonia consumption

NOxemission punishment

CaCO3 consumption and water consumption for desulfurization all decreased

which means both the desulfurization cost and denitration cost are reduced.