Research And Prodiction On Soundscape In Underground Shopping Streets

Currently, a great number of population and shortage of land are main causes of impeding the rapid development of urbanization in China, while the development and utilizing of underground shopping streets can mitigate that problem. During the design and usage of the underground shopping streets, the sound environment quality is more and more valued by users. How to make the designers and managers improve the sound environment of underground shopping streets effectively is the focus of this paper which is concerned with the analysis of subjective loudness and acoustic comfort evaluation. Subjective loudness and acoustic comfort evaluation are belong to soundscape research field, involving some factors such as sound, the users, space and environment, thus,it is particularly important to clarify the relationships between those factors and the subjective loudness and acoustic comfort evaluation.This paper chose five typical underground shopping streets in Harbin as the survey sites, and the questionnaire surveys that conclude the relationships of the sounds, the users, the environment and the space were send in these case study sites. Ultimately more than 2,800 copies of valid questionnaire surveys were turned back. What's more, temperature, relative humidity, luminance, sound pressure level and users'density have also been investigated. Thus more comprehensive material of the sounds in the underground shopping streets, the users, the space and environment were grasped.Sounds and the sound sources are not only the objects of the research of the underground shopping street soundscape , but also an important part of the soundscape. Under the surveys it found that in the complex sound environment, measured LAeq has a liner correlation with the subjective loudness, and the relationship between measured LAeq and acoustic comfort evaluation changes in parabolic trend. The study also found that the characteristics of each independent source is also different. For examples, background music and PA system can increase the acoustic comfort, while music from shops and venders' shouting will reduce the acoustic comfort. Futhermore, this paper analysed acoustic comfort of the individual sound sources from three aspects: the subjective loudness, intelligibility and sense of rhythm respectively.Users are the subject of the research of the underground shopping streets soundscape. In this paper, the relationships of social characteristics– users'gender, age, income, education and occupation– are analysed and found that income, education and occupation have correlation with subjective loudness or acoustic comfort evaluation, in which income has the most effect on acoustic comfort evaluation. And then, the behavioural characteristics such as the purpose of the users, frequency, time and the number of companions are different, in this paper, the correlation between those behavioural characteristics and subjective loudness and acoustic comfort evaluation is also concerned.As a bridge between the users and the sound, the space and environment is the carrier of the soundscape research. In space aspect, it found that space factors such as the type of space, functionality, feelings of reverberation and users'density effect subjective loudness and acoustic comfort evaluation. In the environmental aspect, humidity and illumination have correlation with subjective loudness or acoustic comfort evaluation, conversely, the temperature has nearly no effect on them. In this paper, the correlation between those factors and subjective loudness or acoustic comfort is also discussed.To facilitate the designers to make a suitable sound environment of underground shopping streets during design period, this paper set up three types of models for predicting subjective loudness and acoustic comfort evaluation based on BP neural network, which combined the effect of sounds, users, space and environment to subjective loudness and acoustic comfort evaluation. They are general models, individual models and group models respectively. In these three models, the general models have the best generality, but the least predictive effect; individual models do well in predicting, but limitations; group models are the most useful for the architectures, and have moderate forecasting effect. To compare with the BP network model, the models based on ordinal logistic regression (OLR) are also set up, but they are less effective than BP neural network.Under the background developing of low carbon economy and building a harmonious society, underground shopping streets soundscape research and predictive results will provide a new design view to the improvement of the underground shopping streets.