To solve the problems of great investment and low accuracy in measuring the oxygen content in flue gas of utility boilers

the soft-sensor model of secondary variables and oxygen content in flue gas was established by selecting reasonable secondary variables and introducing the support vector machine(SVM)

on the basis of studying the flue gas oxygen content theory. The genetic algorithm(GA) was used to optimize the penalty coefficient and kernel function parameter

and then the GA-SVM soft measurement model of oxygen content in flue gas was built up by applying the optimal value. Compared with the parameter optimization results of particle swarm optimization algorithm and grid method

the accuracy and generalization of the model was tested. The results show that

the genetic algorithm was easier to find the global optimal solution than the particle swarm optimization algorithm and the grid search method. The soft-sensor GA-SVM model had an error within ±0.2% and relative error within ±4%

indicating its measurement result is more accurate and it can meet the prediction requirements for oxygen content in the flue gas under conditions with different loads and operation periods.