AI-Driven Digital Twins: A Theoretical Framework for Predictive Decision-Making in Manufacturing and Supply Chains

Abstract

Manufacturing supply chains are increasingly faced with volatility such as demands, supply chain disruptions, and equipment breakdowns. Although organizations have modernized their approaches to digitalization and analytics, decision-making remains more reactionary and descriptive, making it more difficult to proactively detect and avoid obstacles. Digital twins are a dynamic virtual replica of physical assets and processes, but currently these twins are generally limited to monitoring and offline analytics, with less incorporation of more sophisticated artificial intelligence (AI) functions. This paper builds upon theoretical foundations concerning artificial intelligence-based digital twins functioning as predictive decision engines within manufacturing and supply chain domains. To commence, we gather a synopsis on theoretical foundations concerning digital twins, artificial intelligence/ machine-learning analytics, and predictive decision-making, and put forth a strategic two-tiered architecture concerning physical domains, information/data domains, artificial intelligence/modeling, and human-AI interaction. From this strategic architecture, we put forth a decision-centric theory concerning correlations between information value, digital twins’ accuracy, artificial intelligence’ capability to make predictions on manufacturing processes and domains to arrive at higher-quality decisions and multidimensional manufacturing outcomes such as efficiency, resilience, and sustainability. Using hypothetical examples concerning predictive maintenance, quality prediction, demand projection, and comprehensive end-to-end risk management, we formulate theoretical propositions concerning this innovative technology theory to serve as guidelines within future studies. This paper adds by shedding light on how artificial intelligence-driven digital twins are a socio-technical capability within organizations to enable proactivity within manufacturing domains to transition such organizations’ processes and operations to more proactive and futuristic, and ultimately much more proactive manufacturing and supply chain management.