The probability of an emergency assessment is one of the most important problems in the aviation sector. Timely identification of possible equipment failure can be a solution of this problem, that saves people’s lives. The article considers existing methods for risk assessment. The comparison of these methods is carried out. As a result of this paper, the artificial neural network was designed. The set of vital input parameters was determined during the neural network development. All the input variables are divided into groups. The most promising and accurate approach for probability of an emergency assessing the was determined, which forms the basis of the mechanism for engineering accidents.