DIGITRUBBER - Data mining and AI for optimized cross-process control

DIGITRUBBER is a collaborative project of seven institutes from all over Lower Saxony that are advancing digitalization of rubber as part of the MaterialDigital innovation platform. Throughout Germany, digitization of a wide variety of materials is being researched by 13 collaborative projects under the overall organization of MaterialDigital.

The goal of the DIGITRUBBER collaborative project is to develop a more sustainable production of extruded rubber components by combining new measurement technology approaches, classical modeling and machine learning. Online monitoring of the extrusion process is used for this purpose. The necessary data from the process are continuously recorded. If a deviation of the target values is detected, the controlled variables will be automatically adjusted by an artificial intelligence (AI) to be developed.

Based on their technical orientation, joint partners are responsible for either material development and characterization, measured value recording or data processing. The ITA is responsible for data mining, meaning automatic recognitions of dependencies in a large amount of data and for the AI to regulate the process. For this purpose, a data mining algorithm has to be implemented in order to describe the immanent correlations in a qualifying and quantifying way and to predict resulting effects on the quality of the final product. Therefore, the algorithm to be developed will be based of the recorded and processed measurement results, which will be stored in the project's internal process database.

The AI is trained to identify process and material parameter deviations and ensure the control of the plant in order to guarantee constant high quality of the final product. In addition, an AI uses the process-immanent correlations identified by data mining to implement a control system for online optimization the entire process chain.