Considering the perception of combined railway noise and vibration as a multidimensional phenomenon

Classical models for predicting annoyance due to combined noise and vibration have typically taken the form of a summation of the magnitudes of the individual stimuli. In this paper, the perception of combined railway noise and vibration will be investigated as a complex multidimensional phenomenon. In a laboratory study, 30 subjects were exposed to 10 combined railway noise and vibration stimuli and asked to make paired comparison judgements of annoyance and judgements of pairwise dissimilarity. A multidimensional scaling analysis on the results revealed a perceptual space of 4 dimensions. Correlations between these perceptual dimensions and objective parameters of the noise and vibration signals were then tested in an attempt to find which, if any, noise and vibration parameters may be related to these perceptual dimensions. Once correlating parameters were found, a multiple regression model was developed to predict total annoyance due to combined railway noise and vibration as a function of these parameters. The final multidimensional regression model has a very strong and significant correlation coefficient (0.99, significant at the 0.001 level) and is able to accurately predict total annoyance as a function of spectral characteristics of the noise and vibration stimuli and the duration of the combined stimulus.