Hybrid Modeling Optimization of Thermal Efficiency and NOx Emission of Utility Boiler

Increasing thermal efficiency of utility boilers and reducing pollutant emission are both important for saving energy and protecting environment.The artificial intelligence method is widely used in boiler combustion optimization.Based on the test data of boiler combustion adjustment in a 300 MW power plant,a model of the boiler combustion system was established by using BP neural network targeting at boiler efficiency and NOx emission.The present model was optimized with the genetic algorithms to improve the training and prediction accuracy.The average prediction error declines from 0.22% to 0.06% for boiler efficiency and from 3.5% to 0.15% for NOx emission concentration.And based on global optimization with genetic algorithm method,the multi-objective optimization was transformed to a single objective optimization with weight coefficients.The results show that the present method can optimize the boiler efficiency and NOx emission according to practical requirements,and the optimization level depends on each weight coefficient.Thus,the corresponding boiler operation control parameters could be obtained,which can provide guidance for power plant boiler optimal operation.