BP Neural Network Forecast of Flight "Estimated Final Arrival Time" Based on Levenberg-Marquardt Algorithm Optimization

In response to the existing flight forecasting system and the loss of passengers caused by flight delays, a new indicator was proposed: EFAT. After using wavelet analysis method to extract outliers in data, this paper predicts flight arrival time based on BP neural network model optimized by Levenberg-Marquardt algorithm. Referring to flight data from Airsavvi official website, the RMSE of the model's prediction results and actual data is about 39 minutes, of which 97.28% of the prediction errors are within 26 minutes.