On the basis of the furnace temperature characteristics like nonlinear and coupling

a soft measurement dynamic model for furnace temperature based on the least squares support vector machine was proposed.By using the alternative ideas and matrix theory

the offline model was corrected and there's no need to retrain the model during online prediction.Therefore

this method not only improves the prediction accuracy

but also reduces the computation complexity.Using this model to carry out online prediction for temperature at the furnace arch

the online prediction results agrees well with the actual measured data

the relative prediction error is below 0.02%

indicating this prediction model has good prediction effect

which provides basis for prediction of temperature at platen superheater inlet and has great significance in heating surface safety.