Description
Rubber extrusion processes involve a high number of control parameters and batch-dependent material fluctuations, leading to lengthy and error-prone testing cycles for new products. To address this challenge and enable a more sustainable production, an artificial neural network-based data mining algorithm is developed that identifies correlations between process parameters and product characteristics. This approach allows for consideration of a larger quantity of non-linear correlations that improve with each new application, enabling more precise temperature control of rubber production. Real-world manufacturing data are used to validate the model, which incorporate batch- and recipe-dependent material variations. The results provide new insights into the complex relationships between process parameters and product characteristics, highlighting the potential of data mining algorithms to drive sustainable production in the rubber extrusion industry. This paper presents the iterative development of the data mining algorithm for rubber extrusion lines and demonstrates its practical application in the industry.
