Modelling of an elastomer profile extrusion process using support vector machines (SVM)

This article describes a methodology for semi-automatically classifying and modelling the best start-up curves of an elastomer extrusion process for the automobile industry. These starts are performed manually due to their difficulty and depend on the experience of the operator and on the type of product to be extruded. If the product is not extruded correctly, considerable time and raw material may be lost. The final objective of this study was to identify and model, using the database of manufacturing histories, the best start-up curves produced for each profile manufactured. With the new models obtained, we were able to automate the process and reduce the time used in these start-ups subsequently increasing production, improving quality, reducing defective material and stress on production staff. Initially, principal components analysis (PCA) was used to identify the start-ups that reached the stationary regime most quickly. After extracting the most significant variables from these start-up curves, a dynamic control model was developed using support vector machines (SVM) capable of predicting the velocity variables of the extruders.