Design of Models and Algorithms for Decision Support in Aviation Safety Tasks
ABSTRACT The aim of the study was to analyze effective models and algorithms of decision support that contribute to enhancing aviation safety. The study employed decision-making task formalization, probabilistic modeling, machine learning, multi-criteria analysis, and data analysis methods for processing heterogeneous information, including technical parameters, video streams, and behavioral characteristics. The research identified key features of decision support models and algorithms that improve aviation safety through real-time monitoring and adaptive response. Machine learning methods achieved up to 92% accuracy with a response time of 80 milliseconds, while multi-criteria analysis methods, including the analytic hierarchy process, reached 88% accuracy. The integration of probabilistic models with adaptive algorithms enabled consideration of operational environment variability and timely risk assessment. In 2024, 45% of aviation incidents were associated with crew error, 24% with technical malfunctions, and 13% with weather conditions. Intelligent decision support systems could potentially prevent 75% of incidents related to procedural violations. Neural networks were most suitable for behavioral analysis, decision trees for access control, and Bayesian networks for assessing technical failures. The findings contribute to the development of intelligent data analysis methods and decision support algorithms for aviation safety.