Classification of the location of floor impact noise using AI model based on temporal, spectral, and spatial features in apartment building

In Korea, conflicts are occurring due to misidentification of the floor impact noise (upper floor or other floors). In this study, the characteristics of floor impact noise that can effectively classify the location of floor impact noise were investigated using AI. Floor impact noise was recorded on the upper 3 floors and the lower 3 floors based on the excitation floor, and temporal, spectral, and spatial features were extracted for each floor location of floor impact noise. Based on the extracted features, the location of floor impact noise was classified through CNN and ANN. The performance of the models through each feature was compared and analyzed with the accuracy values shown through the same data. Finally, this study presented the most effective features and AI models for classifying the location of floor impact noise.