In order to meet the arrival of large unmanned aerial vehicles and reduce the probability of accidents, the main roles and responsibilities and scenarios in the operation process were analyzed, and the control feedback structure in the operation process was constructed by using the System Theory Process Analysis Method (STPA) to identify the risk factors that lead to accidents. Based on the correlation between factors, the Bayesian network (BN) was constructed, and the GeNIe software was used to carry out forward causal inference on the risk probability, and the key factors were determined through reverse reasoning, sensitivity analysis and impact intensity analysis. The results show that control failure is the most critical factor leading to accidents. Navigation system failure, bad weather, and battery failure are highly sensitive factors, and the analysis results can provide a basis for the prevention and control of large-scale unmanned aircraft operation risks.