Constructing Dynamic Master Logic Models as Knowledge Graphs for Complex System Diagnostics Using Retrieval-Augmented Large Language Models
A framework for automated construction of Dynamic Master Logic (DML) models from system descriptions and their representation as Knowledge Graphs (KG-DML) is presented, using Retrieval-Augmented Generation and Large Language Models. The framework extends automated KG-DML construction and evaluation to larger and more complex systems. Model construction proceeds across the DML hierarchy using targeted retrieval while preserving functional dependencies and explicit logical relationships. The resulting KG-DML supports diagnostic reasoning, safety assessment, upward failure propagation, and downward dependency tracing. A multi-level validation methodology evaluates layer-specific precision and recall, logical gate consistency, and overall structural integrity. Application to the Low-Pressure Coolant Injection system is mentioned.
A research paper proposes a framework for automatically constructing Dynamic Master Logic (DML) models from system descriptions and representing them as Knowledge Graphs (KG-DML), using Retrieval-Augmented Generation and Large Language Models. The approach extends prior work on small-scale systems to larger and more complex systems. The construction process uses targeted retrieval across the DML hierarchy while preserving functional dependencies and logical relationships. The resulting KG-DML supports diagnostic reasoning, safety assessment, upward failure propagation, and downward dependency tracing. Validation includes layer-specific precision and recall, logical gate consistency, and structural integrity. The framework is applied to the Low-Pressure Coolant Injection system.
The framework combines retrieval-augmented generation with large language models to automate the construction of hierarchical DML models as knowledge graphs. Targeted retrieval is used at each level of the DML hierarchy to preserve functional dependencies and explicit logical relationships. The resulting KG-DML enables both upward failure propagation and downward dependency tracing, which are critical for diagnostic reasoning and safety assessment. The multi-level validation methodology evaluates layer-specific precision and recall, logical gate consistency, and overall structural integrity, providing a rigorous evaluation approach for complex system models.
Automated construction of DML models as knowledge graphs could reduce reliance on expert interpretation of technical documentation, improving scalability for complex systems. This approach may be particularly relevant in industries such as nuclear power, aerospace, and other safety-critical domains where DML models are used for diagnostics and safety assessment. The ability to automatically generate and validate such models could accelerate system analysis and reduce costs.
The framework could lower the cost and time required to build DML models for complex systems, enabling more frequent updates and broader coverage of system documentation. This may create opportunities for software tools or services that automate DML construction and validation, particularly in safety-critical industries.
Next observable signals include publication of detailed experimental results on the Low-Pressure Coolant Injection system, including precision and recall metrics for each DML layer. Further work may extend the framework to other complex systems or integrate with real-time diagnostic tools. Adoption by industry practitioners could be indicated by citations or collaborations with engineering organizations.