This project develops a predictive maintenance system for Bay 3030 in Purdue’s Dudley Smart Factory. Using IoT sensors, Kepware connectivity, and AWS cloud services, we collect real-time production data and visualize key performance metrics through an interactive QuickSight dashboard. An AI assistant, TROY, analyzes trends, explains anomalies, forecasts failures, and supports troubleshooting. Together, these tools provide a unified, data-driven solution that improves machine reliability, reduces downtime, and enhances operational efficiency.
Modern manufacturing facilities face significant challenges in maintaining high levels of efficiency, product quality, and equipment reliability. Unplanned downtime due to equipment failure remains one of the largest sources of production loss, often leading to increased costs, missed deadlines, and compromised customer satisfaction. Traditional maintenance strategies—such as scheduled or reactive repairs—are either too rigid, resulting in unnecessary downtime, or too delayed, occurring only after failures have already disrupted operations.
In Bay 3030 of the Dudley Smart Lab, the skateboard production line exemplifies these challenges. The facility relies on a variety of machines and processes, each with its own operating conditions and failure modes. Currently, there is limited visibility into real-time machine health, leaving operators without the predictive insights necessary to proactively address potential issues before they escalate. This lack of foresight not only increases the risk of equipment breakdown but also limits opportunities to optimize production flow, energy use, and material efficiency.
To address this problem, there is a need for an integrated solution that continuously monitors production through IoT sensors, captures and organizes machine data, and applies advanced analytics to predict failures and optimize operations. By combining industrial connectivity (Kepware), scalable cloud infrastructure (AWS), and artificial intelligence (LLMs), the Dudley Smart Lab can move from reactive maintenance toward predictive and prescriptive decision-making. This will reduce downtime, extend equipment life, and enhance the overall resilience and competitiveness of the skateboard production system.