Estimating Traffic Load Based on Multi-Layer Pavement Strain Monitored by Distributed Optical Fiber Sensors

Traffic load is a primary factor contributing to the onset of diseases in pavement structures, necessitating its rigorous monitoring and management. Current calibration techniques for commercial weigh-in-motion (WIM) systems fall short in ensuring high precision and reliability of identification results. This study introduces an innovative approach that employs dynamic ensemble surrogate models and multi-layer pavement strain monitored by distributed optical fiber sensors (DOFS) for traffic load estimation. Firstly, the multi-layer pavement strain from roads based on field tests and finite element simulations is collected, and the input characteristics of multi-layer strain associated with traffic loads are delineated. Subsequently, a high-performance ensemble surrogate model that maps strain characteristics to traffic loads are established. Finally, a Bayesian-based dynamic sampling approach aimed at enhancing the model's accuracy from a sample modeling is introduced. The results indicate that the total weighing errors is below the permissible limit for the long-term pavement performance at a 95% confidence level, and superior to WIM and portable vehicle weighing systems. In summary, the proposed technique utilizes DOFS to accurate WIM for pavement monitoring applications.