Evaluation of electric vehicle sounds, new concepts regarding speed-dependency and prediction of the warning effect

Abstract

In today's urban environment inhabitants are permanently exposed to elevated noise levels, which are mostly dominated by traffic noise. The current electrification of vehicles might affect the traffic noise in cities. The aim of this work was to determine the pedestrian reaction, the annoyance and the warning effect of electric vehicle sounds. For this purpose the differences in the perceived annoyance, warning effect and detection time were investigated with perception studies. Furthermore the sound level of a full speed-scaling of an approaching vehicle starting from 0 km/h at the critical distance is nearly 10 dB below the level of a constant speed of 10 km/h. Therefore variants of electric vehicle sounds were generated, in which a constant level is used below 5 or 10 km/h. The results show that the change of the speed-scaling influences the detection time enormously. In this study, artificial neural network (ANN) is used as an indexing tool to imitate subjective perceptions, because in some further work the results of artificial neural networks show great correlation with the assessments of subjects in listening tests. Through the use of ANN, a flexible model can be developed which can predict the warning effect of future electric vehicle sounds.