AI-ENABLED DIGITAL TWIN FRAMEWORK FOR PREDICTIVE MAINTENANCE AND ENERGY OPTIMIZATION IN INDUSTRIAL SYSTEMS

Abstract

Industrial organizations face growing pressure to improve asset reliability and energy efficiency at the same time, yet there is limited quantitative evidence on how AI enabled digital twins contribute to these outcomes. This study therefore examines how AI enabled digital twin capability relates to predictive maintenance effectiveness and energy optimization performance in real industrial enterprise environments. A quantitative, cross sectional, case-based survey design was applied to 220 professionals from discrete manufacturing, process industries and energy intensive utilities using digital twin and AI driven maintenance and energy management. Key latent variables were AI enabled digital twin capability, predictive maintenance effectiveness, energy optimization performance, AI analytics maturity and organizational readiness, measured with multi-item Likert scales showing strong internal consistency. The analysis plan combined descriptive statistics, correlation analysis and multiple regression with mediation and moderation tests. Results show moderately high perceived capability levels, with mean scores of 3.67 for digital twin capability, 3.79 for predictive maintenance and 3.61 for energy optimization on a five-point scale, and strong positive associations among the constructs. Digital twin capability significantly predicts predictive maintenance effectiveness (β = 0.58, R² = 0.53) and energy optimization performance (direct β = 0.29 within a model R² = 0.56), with predictive maintenance providing a significant partial mediation and AI analytics maturity and organizational readiness acting as important enablers. These findings indicate that AI enabled digital twins, supported by mature analytics and organizational readiness, act as strategic levers to advance predictive maintenance and industrial energy efficiency in Industry 4.0 enterprises.