Novel traffic oscillation analysis method for continuous flow based on wavelet transform

Traffic oscillation is a prevalent phenomenon on roadways that significantly impacts driving safety, traffic efficiency and energy consumption. However, the analysis of traffic oscillations has typically been limited to the spatiotemporal dimension due to data quality constraints. This paper applies the wavelet transform to convert the spatiotemporal trajectories of vehicles into an energy function with frequency as the variable. Based on this approach, analytical formulas are established to calculate the traffic oscillation period, wave velocity, amplitude, intensity and frequency. This method extends traditional traffic parameters in the spatiotemporal dimension (density, speed, traffic volume) to new parameters in the frequency dimension (oscillation period, wave velocity, amplitude, intensity, frequency), enabling intuitive observation of the essential characteristics of traffic oscillation waves. Finally, the proposed method is analysed and validated using continuous-flow trajectory data. The results indicate that the proposed method can accurately describe the implicit characteristics of traffic oscillations. There are differences in the period of speed oscillation of different vehicles, but they roughly conform to the linear law. The wave speed is relatively stable in car-following scenarios, whereas its distribution becomes uneven in free-flow traffic. Time-frequency analysis is of great significance for modelling and mitigating traffic oscillations.