This study introduces a hybrid Genetic-Inductive Miner framework for object-centric process mining (OCPM) in high-speed manufacturing environments. Traditional case-centric methods often fail to capture multiobject interactions, particularly under conditions of concurrency, variability, and incomplete data. To overcome these limitations, the proposed framework leverages the OCEL (Object-Centric Event Log) standard, combining the structural accuracy of Inductive Miner with the optimization strength of Genetic Miner. Applied to a comprehensive dataset of 100,181 manufacturing events collected over 1,356 days from a dental tube production line, systematic data quality assessment identified 13,715 records with missing critical attributes. Following rigorous preprocessing, the hybrid approach was evaluated on a high-quality dataset of 86,466 complete events, achieving a superior F-Score of 0.923, balancing high fitness (0.938) and precision (0.908) with a discovery time of just 50.2 milliseconds. The framework enabled complete lifecycle traceability across all 11 equipment units and identified significant optimization opportunities, including critical bottlenecks with average delays up to 222.5 minutes. The results confirm the scalability and robustness of the framework for smart factory deployment, with an 86.3% data retention rate demonstrating excellent manufacturing data quality. Manufacturers are advised to adopt OCEL-compliant logging systems and start with pilot use cases in high-variance areas to gradually build internal capabilities. Future research should explore cross-industry applications of the hybrid miner, extend OCEL with sustainability indicators such as energy consumption and emissions, and investigate adaptive tuning strategies for real-time integration. This work offers a validated, scalable approach to process discovery and predictive analytics in complex industrial systems, enhancing operational visibility, efficiency, and data-driven decision-making.
Matonya et al. (Thu,) studied this question.