A Data-Driven Recipe Simulation for Synthetic Rubber Production

To manufacture synthetic rubber, rubber manufacturers require optimal recipes to ensure that it satisfies the required quality standards. Several experiments are required to create the optimal recipe, which adversely affects not only the cost and time required but also the health of workers. Suppose the experimental results can be predicted in advance at the recipe design stage before direct experimentation. In that case, the cost of the experiment can be reduced, and the workers’ health can be significantly less impacted. For this purpose, a method called the prediction walk model using a machine learning model was developed to generate the temperature trajectory in a kneading machine. A cross-updating method to predict the quality of the kneading operation is also proposed. From the results of the experiment, it was confirmed that the performance of the proposed models was superior to that of the existing prediction models.