Voice-Enabled Structural Health Monitoring via Large Language Models—A Cantilever Beam Case Study

Abstract

Structural Health Monitoring (SHM) is critical in ensuring the safety and longevity of infrastructure such as bridges, pipelines, and buildings. Traditional SHM systems rely on sensor networks to capture vibrational and visual data, which are then analyzed using signal processing, statistical models, and machine learning to detect damage and predict failures. However, these systems often require expert interpretation and lack intuitive interfaces for real-time insights. Recent advancements in artificial intelligence, particularly Large Language Models (LLMs), present new opportunities to enhance SHM workflows through natural language understanding and data-driven reasoning. In this paper, we introduce a novel LLM-powered voice agent that enables users to interact conversationally with structural sensor data. By integrating LLMs with time-series SHM data, our system can respond to voice queries, summarize experiment outcomes, detect anomalies, and generate visualizations on demand. We validate our approach using experimental data from a cantilever beam subjected to multiple damage scenarios. These results highlight the potential of LLM-driven conversational interfaces to make SHM systems more accessible, interpretable, and actionable for both experts and non-specialists.