Abstract
<jats:p>Abstract. This paper presents an intelligent method for the adaptive assurance of dependability in unmanned aerial vehicle (UAV) control systems based on state evolution prediction within a receding horizon framework. The proposed method integrates a nonlinear dynamic model, predictive trajectory estimation, probabilistic dependability assessment, risk-based interpretation of performance degradation, and optimization-based control synthesis under safe operating region constraints. Dependability assurance is formalized through an integrated criterion that accounts for both probabilistic indicators and dynamic stability conditions of the system. Numerical simulations were conducted for three degradation scenarios: turbulence, partial engine efficiency loss (20–30%), and attitude sensor drift. Experimental results show that the average dependability level of the proposed method reaches 0.92–0.95, compared to 0.78–0.82 for reactive control. The frequency of violations of the safe operating region was reduced by more than two times, the average safe operation time increased by 25–30%, and peak risk values decreased from 0.3–0.35 to below 0.1. The results confirm the effectiveness of transitioning from a reactive to a predictive-adaptive dependability assurance model for autonomous UAVs. Key words: adaptive control, dependability, receding horizon, safe operating region, state prediction, stochastic invariance, unmanned aerial vehicles.</jats:p>