Automated Failure Mode & Effects Analysis (FMEA) Prioritization via Hierarchical Bayesian Networks and Dynamic HyperScore Adaptation in Silicon Carbide (SiC) Power Semiconductor Manufacturing

**Abstract:** This paper introduces a novel methodology for enhancing Failure Mode and Effects Analysis (FMEA) in SiC power semiconductor manufacturing by leveraging Hierarchical Bayesian Networks (HBNs) and a dynamically adapting HyperScore system. Current FMEA processes are often reliant on subjective risk prioritization, resulting in inefficient resource allocation and delayed mitigation efforts. Our system automates risk ranking by synthesizing historical process data, simulation results, and expert knowledge within an HBN framework. Furthermore, introducing a dynamically adjusted HyperScore based on real-time process monitoring and data streams ensures continuous FMEA refinement and proactive failure mitigation, leading to a 15-20% yield improvement and significant reduction in costly debug cycles within 3-5 years.