Research On Association Rules Mining And Classification Model Of Digital NOTAM

Aeronautical information is an important part of modern aviation systems engineering. It is very important to the specification for civil aviation safety and the improvement of economic efficiency. With the development of computer-based navigation systems and other new navigation technology, aviation safety's dependence on aeronautical information become higher and higher. System Wide Information Management(SWIM) application system has been included in the twelfth five year science and technology development plan. In the establishment of a network-centric information service mode process, research on the service mode of the underlying data should be considered first. For the aeronautical information, study on its digital characteristics and classification model is the cornerstone of the relevant work.Since NOTAM is an important part of aeronautical information, in this paper, we do some research on NOTAM data. Firstly, NOTAM data is modeled according to its characteristics. Regular expressions are used to extract key information and establish the corresponding XML Schema model. Thus, the text of the NOTAM is transformed into a structured, standardized XML document. Secondly, an improved Apriori algorithm is proposed to do association rules mining in NOTAM data, which combines transact compression and candidate item compression. Constraint-method of before and after pieces of rules is used to limit the generation of redundant rules. Experiments show that the algorithm can accurately dig out association rules between the effective time of the NOTAM and incidents involved in the NOTAM, providing the basis for predicting the effective time. Also, the algorithm assists air control and the dispatch release decisions making process. Thirdly, we proposed a support vector machine model based on hybrid kernel function, and AFSA is used for support vector machine model parameter optimization. To improve the performance of optimization algorithms, artificial fish algorithm's step and vision are adjusted dynamically and its behavioral strategies are optimized in traditional AFSA. In the experiments, the IAFSA-SVM algorithm based on hybrid kernel function is compared with the support vector machine algorithm based on a single kernel and PSO-SVM algorithm to verify the validity of the improved algorithm. Further, the algorithm is applied to establish the classification model of NOTAM data, and achieved good classification results.