ANFIS Model for Prediction of Critical Strain Without Back Calculation of Layer Moduli

Existing roads often feature pavements with uneven characteristics. The thickness of each layer and the materials used can vary significantly throughout the road, creating inconsistencies that complicate the assessment of overall structural health. Traditionally, the back-calculation method was used to evaluate the structural health of pavements, but this approach is cumbersome and time-consuming. This research proposes a novel prediction model that simplifies the process by directly evaluating critical strains within the pavement. The study applies prediction models based on the Adaptive Neuro-Fuzzy Inference System (ANFIS) to estimate critical strains. These models utilize deflection bowl parameters, including the surface curvature index (SCI) and base curvature index (BCI), derived from falling weight deflectometer (FWD) deflection data, pavement surface temperature, and existing layer thickness data. The proposed ANFIS model demonstrates a high coefficient of determination (R²) of 0.96 and a very low root mean square error (RMSE) of 0.04, indicating excellent predictive accuracy. By eliminating the need for time-intensive back-calculation methods, this model provides a more straightforward and efficient approach to assessing pavement conditions.