Research on data-driven flight safety risk assessment and time series forecasting method

The paper presents a method for delineating flight safety risk events using flight data mining. The approach employs the entropy weight method to determine the weight coefficients of risk events, thereby enabling the calculation of a flight safety risk index through their weighted inner product. Moreover, the study utilizes the time series algorithm, known as Prophet, to conduct a comprehensive analysis of flight safety risk trends. Initially, this method extracts key risk factors that significantly impact flight safety from extensive QAR (Quick Access Recorder) data acquired from actual airline operations. Subsequently, a statistical analysis is conducted on historical flight segments to identify safety risk events, while the entropy weight method assigns weights to each event. This step contributes to the development of a comprehensive evaluation model for flight safety risk analysis. The method also exhibits exceptional performance in practical flight evaluation and trend analysis. It effectively predicts the trend of the flight safety risk index and provides a comprehensive reflection of the overall flight safety level of the fleet.