Exhaust heat loss is the largest item in the heat loss of boiler units

which largely depends on the exhaust temperature.Therefore

the establishment of boiler exhaust gas temperature model has important significance to improving the boiler efficiency.On the basis of the combustion adjustment test data of a660 MW unit boiler

the support vector machine(SVM)method was employed to build up a boiler exhaust temperature model.Moreover

the improved chaotic particle swarm algorithm(CPSO)was applied to optimize the related parameters for the support vector machine

which greatly improved the accuracy of the model.The result shows that

compared with the genetic algorithm

the PSO algorithm has no crossover and mutation

it is simple and has high generalization ability and strong practicability.Therefore

the exhaust temperature model based on PSO algorithm and SVM can predict the change of the exhaust temperature precisely

and provide theoretical guidance for the boiler operation efficiently.