The Optimization of Equipment System-ofsystems Resilience Protection based on Reinforcement Learning

The military data of the 2022 Russia-Ukraine conflict shows that some equipment will always be damaged or recovered in the confrontation of equipment system-of-systems (ESoS), and resilience is a key factor for that. Although equipment system-of-systems resilience (ESoSR) has important value for effectively completing tasks and optimizing its architecture, studies on the optimization of ESoSR, which still not enough, especially when in the face of continuous disturbances. Therefore, this paper proposes an optimization method for resilience based on reinforcement learning from the perspective of protection. First, the ESoS combat network is modeled based operation loop. Secondly, the capability of the ESoS is quantified by the effectiveness of operation loops, and based on this, a new ESoSR comprehensive model is constructed considering continuous disturbance. Then, we present an optimization strategy for ESoSR based on reinforcement learning to determine the protection sequence of equipment. Finally, a military example is used to illustrate the reliability and effectiveness of the resilience model and protection strategy. The results provide useful insights for operational guidance and the design of a more resilient ESoS.