An Artificial Neural Network Approach for Predicting Pavement Distress: A Case Study Toward Sustainable Road Maintenance

Yogi Oktopianto
Antonius
Rochim, Abdul

Advance Sustainable Science, Engineering and Technology (ASSET)
Vol. 7 No. 3 (2025): May - July

The Surface Distress Index (SDI) is a crucial parameter to consider when determining road conditions as part of an effective maintenance strategy. This study aims to develop an SDI prediction model using road surface distress data to enhance maintenance planning. The developed Artificial Neural Network (ANN) model resulted in an optimal structure with two hidden layers comprising 6 neurons and 4 neurons, respectively. The model was trained using two years of surface distress data collected from 40 road sections managed by the city’s road maintenance division. Variables used included Composition, Condition, Depression, Patches, Damage types, Crack Area, and Crack Width. The results demonstrated high accuracy in predicting SDI, with model performance achieving an R² of 0.87. This model can be applied to optimize the efficiency of road maintenance strategies.