Air traffic system modeling approach based on OO, image-moment & self-adaptive clustering

Abstract

Much work was done to set up metrics subjected to air traffic system, because isolated and fragmented traffic data should be converted into meaningful information with which decision-makers could understand the situation easily and find the solutions automatically, intelligently and effectively [1]. However, there was little approach successfully integrating various kinds of metrics in a comprehensive and undistorted framework. That's why a traffic system modeling approach including object oriented, image-moment and self-adaptive clustering techniques is set up. It's designed for DMS aiming to generally and flexibly analyze the operating status of air traffic system with a unified benchmark and hierarchical characteristics system of the metrics. Multivariate data mining, pattern recognition and knowledge discovery on radar and flight plan data then could be accomplished based on such an approach to improve ATM system performance. Case analysis with real radar data of Guangzhou Area Control Center indicates that the approach is effective as it has been accepted by controllers getting better understanding of various kinds of metrics with more hierarchical and systematic modes.