Detection of tire types using tire noise from passing vehicles

Abstract

Winter tire is important and helpful for road users or automobile drivers to obviate serious traffic accidents. They also help road administrators to prevent many slip traffic accidents by such vehicles especially from the expressways, particularly in a snowy area. This paper is concerned with the reliable detection of tire types using only tire noise from passing vehicles. In practice, the tire noise emitted from moving vehicles varies momentarily depending on several mechanisms, such as road surface properties, tire tread patterns, and so on. As a result, it may be possible to passively and easily detect the tire type. For example, the least signal differences between winter and summer tires. To detect tire noise from running vehicles at 30, 40, 50 and 60 km/h on average, only when road surfaces were dry or wet state, we used a commercially available microphone as an acoustic sensor, which enabled us to easily reduce cost and size in a practical system for detecting tire types. We propose simple detection methods based on the cumulative distribution function of the power spectrum and the autocorrelation function of the tire noise signals to extract the signal features in the frequency domain and the time domain, respectively. Experimental results obtained from recorded signals in the snowy area demonstrated that the proposed method achieves high classification accuracy.