This paper explores the rapidly growing global market for data integration in the context of Industry 4.0. It highlights the challenges of applying AI within the Cross-industry Standard Process for Data Mining (CRISP-DM), particularly in the plastic waste sorting sector, where heterogeneous data formats and different machines pose integration challenges. The research proposes a robust data infrastructure that uses semantic data integration to facilitate the aggregation and transformation of heterogeneous time series data into a unified format for AI training. The developed data infrastructure has been applied for the data integration in the application domain of a waste sorting process. The results show that the proposed approach successfully integrates data from multiple sources, confirming the feasibility of adaptable and scalable solutions in complex industrial environments such as the presented plastics sorting plant.
Franke et al. (Thu,) studied this question.