Abstract: A new algorithm based on information fusion and soft-sensing technique for modeling the measurement of the carbon content in fly ash for power plant is proposed.Firstly,adaptive weighted fusion and least square support vector machine(LSSVM) algorithms are designed.Secondly,models based on three nonlinear testing functions are constructed by the BP neural network,LSSVM and LSSVM based on adaptive weighted fusion algorithms,and then their accuracies are compared respectively.Finally,the LSSVM algorithms based on adaptive weighted fusion are applied to modeling of the carbon content in fly ash for power plant,and the results are given.