Thinking Like an Expert: Aligning LLM Thought Processes for Automated Safety Modeling of High-Altitude Solar Drones
Open Access
Abstract
As the application of high-altitude solar drones expands, ensuring their safety is paramount. Traditional safety modeling, which relies on manual expert analysis, struggles to keep pace with rapid development cycles. While Large Language Models (LLMs) offer a path to automation, state-of-the-art reasoning frameworks like Graph of Thoughts (GoT) are too generic, lacking the domain-specific knowledge required for effective application. To address this gap, we introduce K-EGoT, a framework that grounds LLM reasoning in a verifiable, domain-specific knowledge base. Our method introduces a “Safety Rationale”—a mandatory, auditable link between LLM-generated model extensions and expert-curated safety principles. We then train a specialized model using a novel “thought process alignment” strategy, applying Direct Preference Optimization (DPO) to the quality of these rationales to ensure the model’sreasoning aligns with expert logic. On a high-fidelity dataset for the flight control–energy coupling problem, our 7B K-EGoT model achieved a Safety Extension Score (SES) of 92.7, significantly outperforming the 84.7 score from standard GoT prompting. Our work delivers a reliable and auditable solution for automated safety modeling for this critical class of drones.Keywords
Funding Information
- National Key R&D Program (#2022YFB4501904)
This publication has 16 references indexed in Scilit:
- Robust Spatial-Temporal Autoencoder for Unsupervised Anomaly Detection of Unmanned Aerial Vehicle With Flight DataIEEE Transactions on Instrumentation and Measurement, 2024
- Deeply Integrated Autoencoder-Based Anomaly Detection and Critical Parameter Identification for Unmanned Aerial Vehicle ActuatorsIEEE Sensors Journal, 2024
- Data-driven multivariate regression-based anomaly detection and recovery of unmanned aerial vehicle flight dataJournal of Computational Design and Engineering, 2024
- Spatio-temporal correlation-based multiple regression for anomaly detection and recovery of unmanned aerial vehicle flight dataAdvanced Engineering Informatics, 2024
- Deep learning-based anomaly detection for individual drone vehicles performing swarm missionsExpert Systems with Applications, 2023
- Multivariate Regression-Based Fault Detection and Recovery of UAV Flight DataIEEE Transactions on Instrumentation and Measurement, 2019
- A STAMP-based ontology approach to support safety and security analysesJournal of Information Security and Applications, 2019
- A profile and tool for modelling safety information with design information in SysMLSoftware and Systems Modeling, 2014
- Overview of Model-Based Engineering2013
- Functional hazard analysis for highly integrated aerospace systems1998