China is exploring and implementing policies to promote the recycling of electric vehicle batteries. Few studies have investigated the coevolution between residents and recyclers in the recycling system. This study employs agent-based modeling and network-based evolutionary game theory to develop a model based on a two-layer heterogeneous social network. This model captures the complex interactions between residents and recyclers, analyzing the impact of various policies on these coevolutionary systems. Our results reveal that resident-oriented subsidies can increase the formal recycling strategy diffusion rate to 99.8 %, while indirectly raising the proportion of formal recyclers to 85.1 %. The evolutionary patterns of formal recycling strategy in both resident and recycler networks follow similar trends, requiring prolonged social interaction and iteration to reach equilibrium. Additionally, the marginal diminishing effect of subsidies highlights the potential for policy redundancy compared to regulatory measures. Policy effects are transmitted across agent types through coevolution and information dissemination, although information distortion may impair transmission efficacy. Moreover, formal recycling strategies receive a greater boost in the quantity subsidy model, while capacity subsidies incentivize higher levels of recycling diffusion in the system. This research contributes to a deeper understanding of the policy-driven co-evolutionary dynamics in the electric vehicle battery recycling system.