Logistics equipment condition monitoring and prediction based on digital twin and machine learning

Abstract

The use of digital twins is becoming a foundation for automation that will improve how industries handle and organize data about both virtual and physical things. It enables analyzing industrial data more seamlessly by merging the Internet of Things (IoT) with Artificial Intelligence (AI) to make sense of it. The growth of online retailers has made it harder for logistics professionals to keep people safe, make sure products are of superior quality, and run smoothly. The most recent advancements of digital twin (DT) have made it easier to create predictive maintenance. Using DT makes it easier to accurately assess equipment status and detect problems before they occur, thereby making the system more reliable. This shift from reactive to preventive operations makes maintenance plans more efficient, reduces disruptions, and boosts the company's profits and competitive advantage. Nevertheless, the research and implementation of Digital Twin (DT) for Predictive Maintenance remains developing, likely due to the incomplete exploration of the function and significance of machine learning (ML) within this context by academics and industry alike. In this paper, a digital twin solution in which the logistics equipment is monitored and continuously maintained through the application of ML algorithms in an IOT environment is proposed. The logistics 2.0-enabled system generates virtual replicas of physical logistics assets, such as forklifts, conveyor belts, automated guided vehicles, cranes, and warehousing robotic, and the digital twins are synchronized in real-time via IoT sensor networks. For anomaly detection, Remaining Unit Life (RUL) prediction, and failure classification, this study utilized Isolation Forest (iForest), Autoencoders, Long Short-Term Memory (LSTM) networks, and Random Forest (RF) machine learning models. The architecture consists of three layers, each of which is connected to the other named the physical layer, including heterogeneous IoT sensors (vibration, temperature, acoustic, and current/voltage, GPS, and load sensors); the digital twin layer, enabling real-time synchronization and simulation; and the ML layer running predictive maintenance and optimization models. Results from the application of the proposed system show a 30-50% reduction in the equipment downtime, 20-40% diminishment in the maintenance cost, longer lifespan for the equipment, and better operational safety.