Mission Reliability Evaluation for Heterogeneous Unmanned Swarm Systems Based on Conditional Evidential Networks

Heterogeneous unmanned swarm systems (HUSSs) have gained extensive deployment owing to their superior collaborative capabilities in novel battlefields. However, their heterogeneity and complexity also pose significant challenges for ensuring mission success, making the accurate evaluation of mission reliability of great importance. Addressing limitations of traditional approaches in handling system heterogeneity, multi-layered structures, and epistemic uncertainties, this study proposes a conditional evidential networks (CEN)-based approach to evaluating mission reliability in HUSSs, to reduce assessment inaccuracy caused by epistemic uncertainty. This study constructs a multi-layered collaborative capability evaluation framework for HUSSs, which comprises the basic capability layer, the collaborative interaction layer, and the mission effectiveness layer. By means of indicator mapping and weight propagation, this framework enables quantitative evaluation of capabilities. Subsequently, elaborating the approach to evaluating state credibility between underlying capability indicators and higher-level capabilities. Building upon this foundation, the integration of CEN with quantitative weighting enables effective evaluation of mission reliability for HUSSs. Finally, the effectiveness of the approach is validated using a heterogeneous unmanned aerial vehicle (UAV) swarm as the research object. The mission reliability of the HUSSs is calculated to be 0.5079.