An Intelligent Operation and Maintenance Framework Based on Belief Reliability for Unmanned Swarm Systems

With the widespread deployment of unmanned swarm systems in rescue, logistics, military and other fields, their scale and complexity are rapidly increasing. Intelligent operation and maintenance (I-O&M) based on deep reinforcement learning (DRL) has gradually become a key means to ensure the high availability and cost-effectiveness of unmanned swarm systems. The effectiveness of decision-making in I-O&M is inseparable from scientific assessment methods. However, the assessment indicators and methods in existing research lack universality and are difficult to provide reliable support for I-O&M in diverse environments. Therefore, this paper proposes an I-O&M framework for unmanned swarm systems based on belief reliability. The framework covers four common technical modules: unmanned swarm system modeling, multi-dimensional situation perception, intelligent decision-making, and belief reliability assessment. By integrating the belief reliability theory and multiple deep reinforcement learning algorithms, this framework can optimize the O&M strategy in a complex dynamic environment to reduce the O&M cost and improve the reliability level of the unmanned swarm systems, thereby realizing the autonomous O&M and dynamic management of the unmanned swarm systems.