Classification and Prediction of Overloaded Trucks Passing the Southbound Lane of North-Luzon Expressway Using Artificial Neural Network

The rapid industrialization and development of the Philippines equate to an increased volume of vehicles. Overloaded vehicles, notably trucks and trailers, cause heavy traffic, accidents, and highway damage. To mitigate this problem associated with economy, highway safety, and the integrity of highways, this research proposed using machine learning models to gain information on the truck's patterns and predict overloading with high accuracy. In this study, the researcher used the High-Speed Weigh-In-Motion (HS-WIM) seven-day dataset provided by the North Luzon Expressway corporation to analyze and gain insights into the truck's pattern and behavior utilizing unsupervised learning technique (clustering) Principal Component Analysis and t-distributed stochastic neighborhood embedding, and k-means to cluster the trucks. Using an artificial neural network, the researcher developed several models and tested them on the unseen dataset to predict overloading. An average of 89% accuracy was obtained, which was higher than the 85% accuracy recorded during the 2019 calibration of the HS-WIM. This research provides valuable insights for local chief executives, decision-makers, and policymakers to formulate updated road safety and management guidelines and reasonably utilize resources and potential to develop and manufacture a low-cost sensor-based system that can predict an overloaded truck on our highways.