A novel method of aviation safety prediction based on Lstm-Rbf model

Abstract

The aviation industry has great economic value. The traditional aviation safety management strategy is mainly “post emergency”, which has been difficult to meet the current needs of intelligent and refined safety management. Therefore, this paper proposes a new aviation safety prediction method based on LSTMRBF (Long and Short Term Memory-Radial Basis Function) neural network model. Firstly, based on SHEL (software, hardware, environment, liveware) accident causativeity theory, an aviation causativeity identification model was established to optimize the key causative events. Secondly, based on LSTM network, the temporal analysis model of causative events was established to predict the change trend of each causative event. Finally, the predicted value of the causative event was substituted into the trained RBF fitting model to obtain the results. For this purpose, we employ survey data collected from a transport aircraft. Result shows that the RMSE (Root Mean Square Error) and MAPE(Mean Absolute Percentage Error) are 6.191 and 4.4321(15% lower than the existing prediction model of temporal series prediction method), which proves the effectiveness of the proposed method.