A Unified Multitask Learning Framework for Automated Health Monitoring of Concrete Structures

Structural health monitoring (SHM) plays a pivotal role in maintaining the integrity and safety of critical concrete infrastructure, including bridges, tunnels, and high‐rise buildings. The growing complexity of modern structures and increasing safety regulations necessitate advanced, automated monitoring systems capable of detecting, localizing, and assessing damage with high precision. Traditional inspection methods, which rely heavily on manual labor and contact‐based sensors, are not only time‐consuming and costly but also prone to subjective errors, underscoring the urgent need for more reliable and scalable solutions. Despite recent advancements in deep learning for SHM, existing approaches often treat damage detection, localization, and severity assessment as isolated tasks, resulting in redundant computations and suboptimal performance. Moreover, the deployment of such models in real‐world scenarios is hindered by challenges such as limited labeled data, environmental variability, and the high computational demands of complex architectures. To address these limitations, this study introduces a novel multitask learning (MTL) framework that unifies these tasks within a single, efficient model. The proposed framework integrates lightweight deep learning architectures, EfficientNet, ShuffleNetV2, and MobileNetV3, with attention mechanisms to enable simultaneous and synergistic learning across tasks while minimizing computational overhead. Extensive experiments on benchmark datasets demonstrate the framework’s superior performance, with MobileNetV3 achieving 98.2% accuracy in damage detection, precise localization (RMSE ≤ 0.10), and robust severity estimation ( R 2 ≥ 0.98). The MTL approach reduces computational costs by 25% compared to single‐task models. By overcoming key challenges in data efficiency, model scalability, and task integration, this research significantly advances the field of automated SHM. The proposed framework not only enhances the accuracy and reliability of structural assessments but also provides a practical, cost‐effective solution for large‐scale infrastructure monitoring. Its contributions pave the way for smarter, data‐driven maintenance strategies, ultimately promoting safer and more sustainable built environments.