"How to leverage Agentic AI and Knowledge Graphs to enhance Overall Equipment Efficiency (OEE)"

Presentation Abstract

Overall Equipment Effectiveness (OEE) is a critical metric for evaluating the efficiency and productivity of industrial equipment and plants. It provides a holistic view of operational effectiveness, encompassing three core components: Availability, Performance, and Quality. Identifying the root causes of poor OEE requires analyzing diverse plant data, including predictive maintenance records, work orders, timesheets, root cause analyses (RCAs), and event logs. However, manually sifting through these large, often unstructured datasets is a complex, time-consuming, and potentially error-prone process, hindering timely identification and resolution of OEE bottlenecks. Furthermore, the interplay of multiple factors can obscure the true source of underperformance.

 This research proposes a novel approach to OEE optimization by developing AI agents based on Large Language Models (LLMs) that leverage a comprehensive knowledge graph. This knowledge graph represents a digital twin of the plant, including equipment specifications, performance history, failure records, and work order details. The AI agents intelligently traverse this knowledge graph to investigate instances of poor OEE and pinpoint their underlying causes.

 We present a case study showcasing how the AI agent successfully identified work order execution delays and unresolved RCAs as significant contributing factors to decreased plant OEE for an Oil and Gas industry. By intelligently leveraging LLMs and knowledge graphs, we can move beyond reactive problem-solving in industrial settings and proactively identify and address the root causes of efficiency losses. Future work will focus on expanding the scope of the knowledge graph, refining the LLM prompts, and developing more sophisticated agent navigation strategies.

Speaker Bio

Akhilesh Jain is a seasoned AI and data science professional with a robust interdisciplinary background spanning chemical engineering, computer science, and machine learning. Currently serving as a Product Manager for Advanced AI and Analytics at Baker Hughes, Akhilesh brings over a decade of experience in solving complex industrial challenges across energy, chemicals, and fertilizers industries.

 Previously, he led high-impact data science teams at SparkCognition (now Avathon), where he applied cutting-edge AI techniques to drive innovation and operational efficiency for Oil & Gas (BP, Marathon). Akhilesh holds a PhD in Chemical Engineering from The University of Texas at Austin, where his research focused on developing software for computational nanoimprint lithography and fluid dynamics.

 Akhilesh has authored multiple peer-reviewed publications and contributed to open-source projects, including time-series forecasting and anomaly detection. His leadership roles in academic organizations—such as NSF ERC NASCENT Student Leadership Council—reflect his commitment to community engagement and mentorship. Akhilesh combines technical depth with strategic vision to deliver scalable, data-driven solutions. He is passionate about bridging domain expertise in Chemical Engineering with AI to create meaningful impact.