Research On The Warning And Predicted Method Of Derived Events By Flight Delay

With the fast development of civil aviation, the problem of fights delay has become increasinglyprominent, moreover, the derived events caused by fights delay cannot be ignored as well. If fightsdelay and derived events can be warned accurately, a large amount of economic loss of airlines can bereduced. With the forecasting analysis of flights delay, as well as crowd characteristic informationprovided by video surveillance system, the main purpose of the thesis is to build a warning model toforecast the passenger mass incidents caused by flights delay, so that the staff can have betteremergency disposal in a shorter time.The thesis firstly introduced methods of flights delay forecast and crowd characteristic extraction.In the forecast the weighted Markov prediction algorithm and Bayesian statistics were used to obtainthe amount of departure fights delay and average delay time; In crowd characteristic extraction, themethod combining SA-AdaBoostSVM face detection algorithm and texture analysis was used toextract the crowd density characteristics, meanwhile, the algorithm of spatial clustering was used toextract the crowd appearance characteristics. Then the feasibility of every algorithm was verified, sothat the features were correctly obtained. Then sensitive features were chosen to build the warningindex system, and a warning magnitude was constructed by means of literature and research, thus acomprehensive evaluation model of derived events caused by fights delay was built up, the accuracyof which was tested by a series of experiment. At last, the thesis summarized the content and researchachievement, and proposed some problems under further exploratory research.