Bayesian Network risk analysis of aircraft runway excursion during thunderstorm scenarios
A model integrating text mining with Bayesian networks was developed to quantify the risk of aircraft runway excursions during thunderstorm scenarios.Initially,136 official accident investigation reports were collected,and key paragraphs were extracted to build a corpus.An LDA topic model was then employed to mine the text and extract key words.Based on these findings,critical risks associated with airport environment,runway conditions,approach and landing procedures,and braking deceleration were identified.Aircraft operations,crew control,airport runways,and airport environment were identified as primary risk aspects,forming a risk analysis index system with 26 secondary indicators.Subsequently,these indicators were compared with historical accident statistics,and node priorities were determined using causal effect formulas.The K2 and EM algorithms were employed to learn the structure and parameters of the Bayesian network.Finally,the Bayesian network model was employed to analyze aircraft runway excursion accident data.It calculated occurrence probabilities of risk factors,inferred direct accident causes through reverse reasoning,and analyzed sensitivity between accidents and risk factors.Research findings indicate that aircraft are more prone to overrun the runway end during thunderstorm scenarios,with a probability of 74.27％,while the probability of runway excursions is 12.6％.Among the risk factors contributing to aircraft runway excursions,excessive float distance,excessive touchdown speed,improper use of reverse thrust,and excessive approach altitude have the highest occurrence probabilities and sensitivity to accident consequences.Regarding runway excursions,excessive touchdown speed,hard landings,and deviation from the runway centerline exhibit the highest occurrence probabilities and sensitivity.Regarding airport runway risks,factors such as friction coefficient and runway surface water area contribute to the degradation of runway usage parameters.These factors,along with risks associated with aircraft operation and crew control,are significant contributors to runway excursions.In the environmental aspect,tailwind,crosswind,and heavy rainfall are identified as major risk factors affecting both aircraft operation and runway performance.Based on the probability calculations from the Bayesian Network model,recommendations for risk prevention and control have been proposed for both crews and airports.
万方数据
维普
