Evaluation method for driver comfort under multi axis coherent vibration of seats.

Abstract

Ergonomics increasingly emphasises that seat design should align with the driver's physiological needs to enhance comfort and health. This study uses deep learning to evaluate the impact of seat multi-axis coherent vibration on driver comfort. Through road tests, the multi-axis vibration signals were collected from the seat backrest, cushion and floor, simultaneously collecting subjective evaluation data. The consistency between subjective and objective data was verified using Stevens' power law, with the R² exceeding 70%, indicating subjective evaluations can reflect driver comfort. Furthermore, a deep learning model integrating multimodal coherent features was used for quantitative evaluation. The results show that the method accurately captures frequency characteristics affecting comfort, with the metrics R², RMSE and MAE being 0.931, 0.096 and 0.071, respectively. This is comparable to the evaluations of ergonomics experts. The proposed method provides a promising solution for driver comfort evaluation. It is significant for enhancing driver health, comfort and safety.