Digital Twin-Based Production Sustainability Index for Dynamic Manufacturing Sustainability Assessment: A Fuzzy MCDM and LSTM-PSO Framework

Abstract

Sustainable manufacturing has become a key challenge for industrial competitiveness and environmental stewardship. However, many sustainability assessment frameworks remain static and do not adequately represent dynamic changes in production systems. This article proposes a Digital Twin-Based Production Sustainability Index (DTPSI) that integrates digital-twin technology, Multi-Criteria Decision-Making (MCDM), and machine learning for dynamic sustainability assessment. The framework combines environmental, economic, and social dimensions in a composite index driven by Industry 4.0 technologies for continuous data collection and evaluation. The fuzzy Analytic Hierarchy Process (AHP) determines indicator weights while accounting for uncertainty in expert judgment. Long Short-Term Memory (LSTM) neural networks optimized using Particle Swarm Optimization (PSO) are used to forecast future sustainability performance. The methodology is demonstrated in a precision-manufacturing case study, where the reported results show a 46% increase in the composite sustainability index, a 26.5% reduction in energy consumption per unit, and a 32% reduction in CO₂ emissions over the study period. The DTPSI framework provides a decision-support approach for production planning and process improvement and illustrates how Industry 4.0 technologies can support the human-centric and sustainable objectives associated with Industry 5.0.