AI-Assisted Structural Maintenance Framework for Serviceability Assessment Using LVDT-Based Displacement Monitoring and Machine Learning
Please login to view abstract download link
Serviceability-related deterioration, particularly excessive deflection, is one of the earliest and most operationally significant indicators of performance degradation in reinforced concrete structures. While displacement monitoring using linear variable differential transformers (LVDTs) provides direct and high-resolution measurements, raw sensor telemetry alone is often insufficient to support reliable and timely maintenance decisions due to noise, drift accumulation, and the absence of predictive context. This work proposes an AI-assisted structural maintenance framework that integrates continuous LVDT-based deflection monitoring with machine-learning behavioural models to support serviceability-driven decision-making in reinforced concrete beams. Mid-span deflection is monitored in real time and evaluated against learned structural response patterns obtained from a supervised learning model trained on historical displacement data under normal operating conditions. An AI interpretation layer acts as a virtual maintenance assistant, continuously comparing measured and predicted responses, quantifying confidence levels, identifying progressive stiffness loss, and assessing proximity to serviceability limit states. The framework is demonstrated on a reinforced concrete beam subjected to sustained loading, where displacement data are streamed continuously and analysed in parallel by the predictive model. Results indicate that the proposed sensing–learning integration reduces displacement signal variance by approximately 35–45% relative to raw LVDT measurements and enables earlier identification of abnormal deflection trends. Compared with conventional threshold-based monitoring, the framework reduces maintenance decision latency by approximately 40% and decreases false alerts caused by transient sensor noise by nearly 30%. These improvements enhance both the reliability and interpretability of serviceability assessments, demonstrating the potential of AI-assisted displacement monitoring for predictive maintenance of reinforced concrete structures.
