Abstract
The THREAD Act (Transportation Recall Enhancement, Accountability and Documentation) mandated the use of a suitable tyre pressure monitoring system (TPMS) technology in all light motor vehicles under 5 tons. In the United States, as of 2008 and the European Union, as of November 1, 2012, all new passenger car models released must be equipped with a TPMS. This would alert drivers of under-inflation events. Many countries followed the adoption of TPMS into vehicles. The existing systems depend on pressure sensors strapped on the rim of the tyre. These sensors read the pressure information inside the tyre and transmit it to the receiver in the car. Some systems depend on wheel speed data from the wheel speed sensor. A difference in wheel speed would trigger an alarm based on the algorithm implemented. This paper proposes a new method to monitor tyre pressure by utilising the machine learning approach. Vertical vibrations are extracted from a wheel hub of a moving vehicle using an accelerometer and are classified using machine learning techniques. Statistical features are used to represent the data in the signal. The logistic model tree (LMT) was used as the classifier and attained an accuracy of 92.5% with 10 fold cross validation and 98.5 when tested.
Keywords: Accelerometer, logistic model tree (LMT), machine learning, pneumatic tyres, statistical features, tyre pressure monitoring system.
Graphical Abstract
6