Tire Pressure Monitoring System Using Feature Fusion and Family of Lazy Classifiers

Abstract

The tire pressure monitoring system (TPMS) is crucial for road safety, fuel efficiency, and vehicle performance. This study focuses on nitrogen-filled pneumatic tires due to their uniform pressure management and thermal stability advantages over air-filled tires. Using machine learning, the research analyzes TPMS data to enhance understanding of tire behavior and vehicle safety. It employs various feature extraction methods and lazy-based classifiers to analyze vibration signals collected under idle, high-speed, normal, and puncture conditions using MEMS accelerometers. The study examines autoregressive moving average (ARMA), histogram, and statistical features individually and in combinations (statistical-histogram, histogram-ARMA, statistical-ARMA, and statistical-histogram-ARMA) to improve predictive accuracy. By integrating these features, the study aims to optimize predictive modeling of TPMS. Empirically, the research achieved 97.92% accuracy using the local weighted learning (LWL) algorithm, demonstrating the effectiveness of combined statistical, histogram, and ARMA features in enhancing TPMS predictive capabilities.