K-means clustering analysis and evaluation for internet of acoustic environment characteristics

In order to improve the validity of acoustic environment characteristic analysis of university libraries, an evaluation study method of acoustic environment characteristic cluster analysis of university libraries was proposed based on the modified Cop-Kmeans algorithm scheme. Firstly, Fraunhofer theory was used for establishing the sound diffusion proper quantification method of Schroeder diffusion to obtain diffusion coefficient with 1/3 octave frequency band, as well as adopting normalized method to eliminate the edge diffraction scattering effect produced under the extreme dimensions. Secondly, against adverse effect that the personal property of pairing constraint might have on the precision of cluster results, the paper improved the algorithm further based on the first solution. Two distribution rules were formulated respectively to find out sample sets with direct or indirect constrained relationship and distribute some or multiple samples with highest accuracy of distribution in priority, which enhanced the cluster quality to some extent. At last, the validity of the proposed algorithm was verified through simulation experiments.