Study Of Clustering Method In Airport Noise Prediction

With the rapid development of civil aviation, a large number of airports are being built,rebuilt or expanded across the country. The throughput and scale of airport are correspondingly to expand. Meanwhile, the airport land and urban land become much closer,and the disputes caused by the airport noise become more and more. It is necessary to predict airport noise scientifically and reasonably. However, using the traditional noise prediction model to predict the noise will spent a lot of time. Therefore, to mine noise distribution model for a specific type of flight incident is a great significance for prediction of airport noise.This paper studies the theory and method of airport noise prediction and data mining,and introduces clustering technology in detail, and especially gives a detailed introduction about hierarchical clustering algorithm, partition clustering algorithm, density clustering algorithm, grid clustering algorithms. At the same time, Aiming at the problem faced by the airport noise prediction, this paper puts forward a method of the airport noise prediction based on data mining.This paper first constructs the airport noise prediction model based on data mining,and analyses the airport noise data mining problem. On this basis, according to the characteristics of airport noise data, this paper presents a fast hierarchical clustering algorithm based on representative point. This algorithm improves the traditional condensed hierarchical clustering algorithm by clustering representative point method and dichotomy strategy. Meanwhile, a clustering results evaluation method which combines clustering representative point and the definition of clustering algorithm similarity is proposed in this paper.Noisy data for experimental verification and analysis is selected by the proposed clustering algorithm and evaluation methodology based on analyzing the influence factor of the airport noise. Experimental results show that the proposed clustering algorithm could effectively explore the distribute pattern of noise around the airport, and the accuracy rate of noise forecast based on the explored pattern could be up to 93%.