Study on evaluation model of soundscape in urban park based on Radial Basis Function Neural Network: A case study of Shiba Park and Kamogawa Park, Japan

Abstract In order to explore whether artificial neural network can simulate and predict the subjective and objective data of soundscape in urban environment, this study developed the subjective and objective transformation model of soundscape in urban park based on radial basis function neural network. The test was conducted based on the soundscape survey data of Shiba Park and Kamogawa Park in Japan. The results showed that the subjective and objective evaluation model of soundscape constructed by radial basis function neural network could predict more accurate subjective evaluation value, the average prediction accuracy rate was 91.23%. In addition, the soundscape in the higher loudness, loudness level and sound pressure level, and the lower sharpness got a higher accuracy, which is beneficial to simulating the tourists’ psychological state of the soundscape. The study proved that the artificial neural network model can provide an effective method for further and more comprehensive acoustical environment research in the future.