Comfort Assessment in Railway Vehicles by an Optimal Identification of Transfer Function

Abstract

Purpose: This research proposes a multi-input multi-output (MIMO) model, based on a data-driven approach, to assess passenger vibration comfort on rail vehicles. The MIMO model represents the mathematical relationship between input and output variables that contains modal properties and vibration transmission. An optimization algorithm estimates the frequency response functions at the interface seat-passenger of the trains. Methods: The MIMO model evaluated the frequency response functions along x-, y- and z- axis. Multivariate analysis considered the calculation of the partial coherence functions, the cross-correlation functions, and the Anderson-Darling nonparametric test. Results: This research considered the acceleration measurements on two Trains, one Tramway, and one Underground. This research developed accelerations analyses in the time domain; the transmissibility of accelerations in the frequency domain. The measuring points were on the seat base and the interface seat-passenger of the trains. The frequency response data, obtained experimentally according to x- y- and z- axis, generated a multi-input and multi-output frequency response model. This research is based on experimental investigations and statistical tests. The calculation of partial coherence evaluated the percentage of spectrum output due to a specific (conditional) input. The partial coherences assess the energy contribution of each input to the output. The cross-correlation test showed the phase between input and output accelerations. The Anderson- Darling test showed the provenance of the data sample from a population with a specific distribution. Anderson-Darling's nonparametric test attaches weight to tails. Conclusions: The seats of the trains were exposed to complex vibrations according to the three directions x, y, and z. The three inputs are linear accelerations along the vertical, lateral, and forward direction of the seat base. If the accelerations on the seat base represented the inputs to the MIMO model, the accelerations acquired at the passenger-seat interface represented the output of the MIMO model.