Mass Estimation of Heavy-Duty Vehicle Using Machine Learning with Simulation-Generated Data

In this study, a large amount of synthetic data was generated using data obtained from heavy-duty vehicles and a validated vehicle simulation model, which produced different driving cycles and trailer masses. A machine learning model and deep neural network models were then used to estimate the total mass of the vehicle with the generated synthetic data. In addition to signals such as vehicle speed, acceleration, torque, and gear shifts, the study recorded various signals from the power unit and driver inputs to produce more detailed outputs for statistical purposes. After performing a consistency analysis of the data, various machine learning algorithms were compared to determine the most suitable model, and the results of the model with the lowest error were evaluated and compared with previous study. This approach aims to reduce costs and minimize error risks by estimating the vehicle's mass without the use of sensors, thereby reducing the potential for sensor manipulation.