Cluster Ensembles Research Based On Validity Index And Application On The Evaluation Of Airport Noise

With the sustainable development of civil aviation, the airport noise pollution problem has beenincreasingly serious. It has a great theoretical and practical significance on how to evaluate the levelof airport noise and the impact on the surrounding environment. Based on the background of airportnoise, in this paper, we analysis and research some problems of clustering validity indices and clusterensembles, and apply related methods to the airport noise evaluation. Many good results have beachieved.Clustering validity indices are used to evaluation the quality of clustering results. In thispaper, the principles of clustering evaluation have been illuminated and analyzed, and the often-usedvalidity indices have been compared. A validity index based on the similarity of density within orbetween clusters has been proposed by referencing to the advantages of other validity indices. Bycomparing with other indices, the validity of the new index has been confirmed.Ensemble learning is a popular research direction of machine learning. Based on thecharacteristics that there is not the priori class information in the process of clustering, the concept ofbipartite graph has been imported to improve the cluster label matching method, and a weightcomputational method and decision-making method of cluster ensembles has been provided in thebasis of research results of clustering validity indices. By using the cluster label matching methodwhich can get the cluster label vectors with highest match accuracy and the weightcomputational method which can reflect the diversity between clustering results, the best number ofclustering and the best clustering result can be achieved.By analysing the characteristics and historical data of airport noise and reseaching the methodsof clustering validity evaluation and cluster ensembles, the cluster ensembles method based on thevalidty indics has been applied to the airport noise dataset. According to the cluster ensembles resultand interrelated regulation, the airport noise level evaluation criterion that not only corresponds to theresidents’ subjective feelings, but also meets the airports’ effective demand has been worked out. Thisairport noise level evaluation criterion can be used to reflect the real noise affect level, and toprovide the theory basis for the airport construction and interrelated regulation.