Prediction and Evaluation of Park Sound Comfort Based on Back Propagation Neural Network

Abstract

Park soundscape construction is an important method to improve the quality of urban environments. At present, the studies on soundscape mostly focus on the evaluation of soundscape indexes and the accuracy analysis of model simulations of actual sites, but the research on small-scale soundscape characteristic spaces is inadequate. Based on a Back Propagation(BP) neural network model, we predict and evaluate the sound comfort in a park. The results show that: (1) The distribution trends of measured and predicted sound comfort values in different scenes (space type, plant type, functional area and sound source type) are relatively consistent. (2) A sound comfort of the park is space dependent. The sound landscape design of small-scale characteristic space is of great significance to improve the environmental quality. (3) The evaluation of emotional acoustics has obvious correlation with sound comfort. (4) The soundscape planning process based on BP neural network is clearly proposed. The research results are of great significance to promote soundscape evaluation and planning based on the evaluation results.