Adaptive Multi-scale Prognostics and Health Management for Smart Manufacturing Systems

Area of Science:

• Manufacturing Systems Engineering
• Reliability Engineering
• Artificial Intelligence in Manufacturing

Background:

• Prognostics and Health Management (PHM) data is underutilized in higher-level manufacturing decisions.
• Existing methods struggle with the complexity of large-scale manufacturing systems.
• Hierarchical integration of PHM information is needed for effective decision-making.

Purpose of the Study:

• To introduce the Adaptive Multi-scale Prognostics and Health Management (AM-PHM) methodology.
• To demonstrate how AM-PHM integrates PHM data across manufacturing hierarchies.
• To enable actionable prognostic and diagnostic intelligence for improved decision-making.

Main Methods:

• Development of the AM-PHM methodology.
• Leveraging hierarchical relationships across component, machine, work cell, and assembly line levels.
• Application of a hierarchical Markov Decision Process (MDP) approach to manage system complexity and optimize policies.

Main Results:

• AM-PHM enables actionable intelligence across the manufacturing hierarchy.
• Decisions are informed by the current and projected health state of the system.
• The hierarchical MDP approach effectively addresses system complexity.

Conclusions:

• AM-PHM provides a structured approach to integrate and utilize PHM information.
• The methodology enhances decision-making by considering system-wide health.
• Simulated results demonstrate the effectiveness of AM-PHM in industry-inspired scenarios.