Quantitative analysis of driving behavior from the frequency domain perspective is a new fuel consumption analysis method that can take into account the impact of fluctuation details that may be ignored in time domain analysis on fuel consumption. Compared with Fourier transform, wavelet transform can handle non-stationary signals well and obtain frequency domain information at each time point. This research uses wavelet transform to process the vehicle's time domain information (speed) to obtain its frequency domain information, and find the relationship between fuel consumption and frequency domain information. The data used in this study is divided into two categories, one is public data, which has high data accuracy, but lacks fuel consumption data, and is estimated through classic fuel consumption models. The other is collected by OBD, the accuracy of the data is low, but relatively accurate fuel consumption data can be obtained. This study proposed a Volatility Factor based on wavelet transform and verified it using data collected by OBD. Experiments show that there is a positive correlation between the newly developed Volatility Factor and fuel consumption, and this indicator can reflect changes in driving behavior during driving.

CICTP