Abstract:In order to reduce the collision risk of urban logistics unmanned aerial vehicle(UAV) and identify the key risk factors, a Bayesian network model is constructed for risk assessment. On the basis of collecting 7 real UAV collision accident cases and literature research, 15 main risk factors are extracted, and the risk set is divided into three levels: low, medium and high through questionnaire survey to collect and evaluate the risk probability data. Then the risk factors are hierarchically divided by using the interpretative structural modeling(ISM) to construct the Bayesian network model. According to the constructed Bayesian network, the collected data is imported into GeNIe, and the parameter learning of Bayesian network model is carried out to obtain the probability distribution of different levels of UAV collision risk. Finally, reverse reasoning, sensitivity analysis and impact intensity analysis are carried out on the Bayesian network model to clarify the key and sensitive factors that lead to UAV collision accidents, and propose risk prevention and control suggestions based on the analysis results.