Ride comfort analysis of transient real world road signals through accurate frequency component identification

This paper presents a ride-comfort strategy for non-stationary road excitations. Because real road inputs are transient, accurate identification of their frequency, amplitude, and damping components is essential for subsequent comfort assessment. The proposed methodology performs short-time analysis by combining autoregressive spectral estimation and power spectral density with a nonlinear error-minimization procedure to estimate the frequency components of the road signal, so precise comfort analysis can be executed. The approach is first validated on a controlled synthetic signal built from 15 sinusoids with predefined frequency, amplitude and damping values in different windows over a 10 second signal. In the synthetic validation case, the method recovered the imposed components with frequency errors below 0.5%, amplitude errors below 8%, damping errors within 5%, a mean absolute window-wise VDV error of 1.7%, and a maximum absolute VDV discrepancy of 5.8% across the nine analysis windows. These low errors indicate that the reconstructed signal preserves the event severity, supporting practical comfort comparison between road excitations. The coherence between original and reconstructed signals supports the reconstruction accuracy, with a minimum mean coherence of 0.91, while time-frequency trends are visualized with Campbell diagrams to track natural frequencies emerging within windows. Then, the proven methodology is applied to three non-stationary real world road measurements obtained from experiments. Based on the identified components, ride-comfort metrics are computed following ISO 2631-1. The resulting reconstruction-based representation of transient components enables comfort studies for virtual development of automotive systems under realistic time-varying excitations.