Abstract:  In order to improve the usability and precision of the model for predicting the viscosity of mixed rubber in mixer‚a new predicting model using the nonlinear regression algorithm of Support Vector Machine （SVM） was pro-posed based on a single-mixer model and the prediction of the measured data was then carried out by the proposed model．The results show that the new model not only simplifies the establishing process of the predicting model but al -so improves the precision and speed of prediction．It is also shown that‚compared with the results obtained by mult- iregression‚the new model is more accurate and effective when applied to predict the on-line quality indexes of the mixed rubber．

Key words:  mixed rubber, viscosity estimation, support vector machine, regression