Aiming at the characters and problems of the soft sensor, a soft sensor modelling method for the soft sensor regression problem based on the ensemble learning algorithm is proposed to improve the accuracy of the soft sensor. According to the shortages of soft sensor update in practical application, an incremental learning idea is added to the proposed ensemble algorithm for soft sensor modelling. The method is used to establish the soft sensor model of sizing in sizing production. The product data is used to test the model. The results show that the proposed soft sensor model can improve the prediction accuracy and realize online update better.