Development of a Pavement Condition Prediction Model Using Machine Learning
This project was part of the Bachelor’s Final Year Project that I supervised. The students involved in this research were Tan, and Yufei. Together, we developed a Pavement Condition Prediction Model that leverages machine learning techniques.
Road networks are an essential part of modern society. Due to environmental factors and traffic loading, roads experience constant deterioration which causes many defects such as ruts, cracks and potholes. Therefore, road condition monitoring and maintenance is very important. But modern methods of road monitoring are very costly and time consuming with road condition reports being made once or twice every two years. This situation has encouraged more research on new road monitoring strategies. Among the proposed solutions, it has been suggested that machine learning or deep learning models can be used to predict road deterioration.
This project aims to create and validate a machine learning or deep learning model for predicting road roughness deterioration measured in the International Roughness index (IRI). After conducting literature review, we determined several factors that lead to road roughness deterioration. Among these factors are Rainfall Volume, Daily Temperature and Traffic Loading. After collecting available data, raw data were analysed and categorized into a complete dataset for model training.
Next, our research into machine learning and deep learning models have led us to choosing the Multilayer Perceptron and Random Forest Regression. These models were chosen for their ability to handle complex, non-linear relationships in the data. We prepared and tuned these models by optimizing their parameters. Both models performed reasonably but the accuracy of models were hindered by the quality of data used.
This project demonstrates the potential of machine learning for predicting road roughness deterioration. Future work should focus on improving data quality and explore more advanced models. Nevertheless, this project serves as an important step towards using data-driven techniques to support proactive road monitoring and maintenance.
Photos from the FYP poster presentation
from left to right, Tan, Yihai, Yufei and Me