Prediction Model Of Flights Alternate Probability Distribution Based On Observational Learning

The flight alternate is an important part of ensuring flight safety. At Present, there exists a problem that gates are inadequate for parking sometimes, which prevents aircraft landing and affects the flight safety. Flights alternate probability distribution is a vital evidence of reserving sufficient gates and the reference for airports and airline scheduling related resources. Therefore, it is necessary to build a probability distribution prediction model scientifically.Because the flights alternate data is small sample. The thought of loosening control condition is adopted for SSSP. And the prediction model of probability distribution based on observational learning is proposed. In training step, revised piecewise cubic spline interpolation functions are viewed as base learners of observational learning. In observing step, the mechanism of mixed virtual sample generation is used for learning each other and improves its generalization ability and precision. Then the process of training-observingtraining is repeated until the learners achieving agreement and generating the final distribution function.To make full use of the history data, a nonlinear programming model used for improving belief matrix parameter is established. The model takes full use of the implied relations between attributes in old data and improves the prediction accuracy and generalization ability of OLA. The simulated annealing is applied to solve the nonlinear programming model. The suitable belief matrix parameter improves the algorithm accuracy.Flights alternate probability prediction model is established,the Bayesian model and the observational learning model are applied to flights alternate model. Experiment based on flight operational data shows that the method of observational learning is better than the Bayesian probability prediction model. The method of observational learning is also used to predict the specified flights probability distribution function to provide support for related department decision.