Risk Identification And Analysis Of Air Traffic Control Based On Text Data And Radiotelephony Data

With the increasing in air traffic flow,which will inevitably occur a lot of air traffic safety risk,how to mine safety information data quickly and efficiently from the ATC text data and radiotelephony data largely affected ATC Safety Risk Identification and Analysis.In this paper,the text data,especially the hazard control list,and the radiotelephony data in the air traffic control data are taken as the research background.The data mining technology and the speech recognition technology are used to deal with the two types of data,and establish the risk identification and analysis index system of ATC operation,combined with the control unit operating data for outlier analysis.The main work and innovation are as follows:(1)A LDA model is proposed based on the LDA model,and the LDA model is used as a solution to the problem of text classification,and good results are achieved in this experiment.The algorithm uses the LDA model to combine the Gibbs sampling to accurately identify the factors that affect the safety of the ATC.(2)A method of identifying the hazard source of air traffic control instruction readback based on Microsoft speech recognition is proposed.The recognition engine can identify the contents of the instruction more accurately with the self-built corpus.The method first translates the voice command signal from the controller and displays it on the screen.When the pilot recites,the instructor recognizes the instruction recited by the pilot,and also converts the speech into text information output on the screen.(3)An algorithm for outlier detection of text data and radiotelephony data is proposed.The algorithm can effectively solve the problem of outlier mining of text data and radiotelephony data.The algorithm first divides the attributes of the text data and radiotelephony data by using the information entropy,reduces the important attributes of some important attributes,preserves the important attributes which can reflect the whole information of the data,and makes use of the important attributes and their weights(contribution degree)combined with the weighted distance formula Detection of cluster points.The experimental results show that the algorithm is efficient for outlier detection of text data and radiotelephony data.