Design and evaluation of a piezoelectric pressure sensor for mass detection with COMSOL and machine learning modeling

• Sensor readings corrected for Arduino ADC range and sensor amplification behavior. • Residual plots added to identify prediction errors and improve model interpretability. • Dynamic sampling clarified with 1.6 kHz effective rate and real-world limitations stated. • Asymmetric load effects analyzed with updated data tables and sensor performance notes. The need to monitor and manage the impact of heavy vehicles on infrastructure has led to the development of novel sensor technologies. This paper presents a prototype piezoelectric-based pressure sensor designed to detect vehicle weight through real-time mass detection, potentially enhancing transportation planning and infrastructure management. A COMSOL Multiphysics model was employed to simulate sensor response under load, and a machine learning (ML) framework, optimized using the BFGS algorithm, was implemented for accurate weight estimation. The model achieved high predictive performance, with a Mean Absolute Error (MAE) of 0.0677 and a Root Mean Squared Error (RMSE) of 0.1207, demonstrating strong agreement between predicted and actual values. While the R2 score on synthetic data was 0.99, real-world testing confirmed the model’s robustness by handling minor deviations caused by environmental and operational factors. The prototype was tested with weights up to 70 kg, with planned future studies aimed at scaling the sensor array for heavy-duty vehicle applications. Operating with a sampling interval of 5 ms, the system theoretically supports weight detection for moving loads at various speeds. However, achieving consistent performance under real-world high-speed conditions may require enhancements, such as faster or parallel data acquisition methods. This work highlights advancements in sensor design and mass detection, with future efforts focused on full-scale deployment in urban infrastructure for real-time traffic monitoring and enforcement.