Machine Learning-Based Prediction of Heavy Truck Tire and Brake Drum Temperatures from Vehicle Parameters

Abstract

Heavy trucks' overheated tires and brake drums present a significant risk of blowouts and failures, which can lead to severe traffic accidents. Traditional temperature monitoring systems typically rely on thermal imaging; however, these systems can be compromised by partial occlusions or subpar image quality. This study introduces a machine learning-based model to predict tire, sidewall, and brake drum temperatures using only vehicle parameters, thereby eliminating reliance on complete thermal imaging. A Random Forest approach has been developed, utilizing vehicle load, size, and environmental factors to estimate temperatures when direct thermal data is unavailable. The proposed model achieves high predictive accuracy across all three targets, with R² values of 0.974 (MSE = 17.318) for overall tire temperature, 0.967 (MSE = 24.839) for brake drum temperature, and 0.976 (MSE = 13.939) for sidewall temperature. This research enhances real-time overheating detection by enabling temperature prediction based solely on vehicle parameters. It significantly improves safety monitoring for heavy trucks and provides an adaptive solution for accident prevention and intelligent traffic management.