The Method of Soft Sensor Modeling for Fly Ash Carbon Content Based on ARMA Deviation Prediction

The carbon content of fly ash is an important parameter in the process of boiler combustion. Aiming at the existing problems of fly ash detection, the soft measurement model was established based on PSO-SVM, and the method of deviation correction based on ARMA model was put forward on this basis, the soft sensing model was calibrated by the values which were obtained by off-line analysis at intervals. The 600 MW supercritical sliding pressure boiler was taken for research objects, the auxiliary variables were selected and the data which collected by DCS were simulated. The result shows that the prediction model for the carbon content of fly ash based on PSO-SVM is good in effect of fitting, and introducing the correction module is helpful to improve the prediction accuracy.