Integration of Digital Twin and Machine Learning for Optimization Maintenance Predictive on Systems Production Manufacturing

Abstract

Maintenance predictive (Predictive Maintenance/PdM) becomes element strategic in industry modern manufacturing because capable prevent downtime planned and improved efficiency line production. In line with progress Industry 4.0, the integration of Digital Twin (DT) and Machine Learning (ML) offers diagnostic and predictive capabilities failure machine based on real-time data, so that decision maintenance can done in a way proactive. Research This aim study in a way comprehensive effectiveness and challenges implementation DT–ML integration in PdM through study literature systematic in publication 2020–2025 period. Data analyzed use approach synthesis thematic for identify pattern algorithmic, benefits operational, obstacles implementation, as well as trend study contemporary. Study results show that DT combination as virtual representation of assets and ML as machine prediction capable increase accuracy detection anomalies, projecting remaining useful life (RUL), reducing downtime, and improving reliability manufacturing. However, the obstacles technical in the form of interoperability system, sensor data quality, requirements security, and operator readiness are still limit adoption scale big. This study give implications scientific and practical that success PdM DT–ML based defined No only by accuracy algorithm, but by architecture a modular, secure, interoperable, and user-oriented system man.