A fingerprint approach for computationally efficient digital twins of precision manufacturing processes

Over the past decade, digital twin (DT) technology has seen significant advancements, enabling it to effectively address complex manufacturing challenges. However, the effectiveness of data-driven DTs is hindered by the sheer volume of data generated by sensors and devices which limits the traditional computational resources causing latency issues. To address this challenge, there is an urgent need to optimize data processing and ensure real-time responsiveness in DT applications. This paper presents a novel fingerprint-driven approach aimed at building computationally efficient DTs for smart manufacturing. When embedded within the DT framework, process and product fingerprints capture the essential manufacturing parameters while reducing the overall computational overhead of the DT. A case study on machining Inconel 718 for a high-fatigue application demonstrates the practical application of the proposed approach in ensuring computational efficiency. When compared to conventional machine learning (ML) methods, the fingerprint approach exhibited substantially superior performance in terms of responsiveness (20 times less latency) and prediction accuracy (3% more accurate).