The electric propulsion system serves as the power source for multi-rotor drones, helping them complete various maneuvering actions. The reliability of this system directly affects whether the drone can successfully complete its mission. The multi-rotor drone propulsion system is a k-out-of-n system with functional dependence (FDEP). With the insufficient basis for selecting k-values, the problem of incalculable reliability caused by computational space explosion due to voting gates, and the uncertain impact of functional dependence on system reliability, we propose a reliability evaluation method based on controllability theory and BN (Bayesian network) reconstruction. The drone is dynamically modeled, and a control model is built, and k-values are selected through different failure combination controllability evaluations. We model the system with BN, use functional dependent components as BN node inputs, and reconstruct BN via an adder model to solve the problem of exponential growth in the conditional probability table. This paper analyzes system reliability, safety, and the impact of FDEP on the system, and conducts component importance analysis. The result provides important reference for the reliability, safety assessment, and dynamic maintenance processes of multi-rotor drone.

As civil aircraft components become more autonomous in China, airworthiness certification is facing the problem of insufficient failure data, leading to challenges in terms of uncertainty. Traditional probabilistic models struggle to effectively integrate multisource uncertainties, limiting their application in data‐scarce scenarios. This study therefore proposes a framework based on evidential networks for quantifying uncertainty risk, including: (1) classifying the failure rates of bottom events into handbook data, flight operation data, and expert experience; (2) quantifying uncertainties through log‐normal distributions and evidence combination rules; (3) converting fault trees into evidential networks and propagating uncertainties through logical gates. In the analysis of the flap control system, the probability intervals for catastrophic failure condition uncommanded motion of flaps are [3.449 × 10 ⁻¹¹ and 4.811 × 10 ⁻⁹ ]. After the improved pignistic decision probability transformation, the point estimate is 1.638 × 10 ⁻⁹ . For the eVTOL aircraft power system, the failure probability intervals of the catastrophic failure condition battery thermal runaway causes uncontrolled fire are [1.604 × 10 ⁻¹⁰ and 2.287 × 10 ⁻⁹ ], with a point estimate of 9.227 × 10 ⁻¹⁰ . The result shows that by accommodating heterogeneous data sources and transparently propagating uncertainties, this method demonstrates significant applicability in both traditional aircraft components and new eVTOL aircraft systems, highlighting its effectiveness. It provides an extensible framework for safety assessment and airworthiness certification.