Studies On Data Warehousing And Data Mining Technologies In The Air Traffic Flow Management

With the development of China's economy, air transportation has increased greatly and the most urgent problem in civil aviation field is the inconsistency of the limited air traffic capacity and the unlimited air traffic demands. This dissertation focuses on the data warehousing and data mining technologies in the air traffic flow management. The work includes the following aspects. Research of the data warehouse modeling technology. This dissertation has summarized the traditional three-stage methods on relational database designing and developing according to the practical Data Warehousing applications, and has brought forward a new integrated method based on it, namely the four-stage standardized model, which includes Conceptual Modeling, Logical Modeling, Object Modeling and Physical Modeling. Considering that the three-stage method is unable to completely reflect the subject-oriented features of the data warehouse and describe its metadata models, this new method is obviously the reinforcement and improvement to the three-stage method, in that it displays the whole structure of the data warehouse more clearly and adapts to the modularization trend of the data warehouse modeling . Based on this new method, the air traffic flow data warehouse is redesigned and rebuilt, which is effective in practical application. Research of the implementation techniques of air traffic flow data warehouse. Based on the four-stage data warehouse modeling technology, the air traffic flow data warehouse is carefully designed and realized. The flow data warehouse is composed of four parts: Data modeling management module, in which the XML data warehouse builder is developed; Data warehouse core module, in which three main units are constructed, including data buffer area, data ware house core storage area and data warehouse management; Data display module, by which the integrated data in the warehouse can be presented to the end users. Through the performance comparing analysis, the data warehouse system possessed higher performance than the database system in the query of systhetical data. Research of the forecast methods of air traffic flow. This dissertation uses the statistical data in the fact tables of air traffic flow data warehouse to analyze the daily flux forecast problems via the data mining technologies. After careful consideration, the dissertation has built a new forecast model for the time series of air traffic flux and provided the data processing methods based on the combination of neutral network and autoregressison. This model is applied to the flux forecast at Da wangzhuang guidance station in Beijing air traffic control region, and the results demonstrate that the new method fits the actual flux very well, improves the generation ability of the neutral network and reduces the forecast errors.