The continuous growth of network traffic and mobile device numbers poses significant challenges to traditional network architectures. Edge computing, by decentralizing resources from the cloud to the network edge, alleviates backhaul bandwidth pressure and enhances user Quality of Service. However, user service requests are still negatively impacted by limited cache space in hotspot areas. Efficiently utilizing idle cache resources and formulating effective caching strategies are therefore critical issues in edge caching. To address these challenges, this paper proposes a cooperative edge caching strategy based on Deep Q-Network (DQN). First, edge nodes are categorized into hotspot nodes and regular nodes based on their network traffic. Then, the K-means++ algorithm is used to cluster edge nodes based on their traffic data and distance, forming cooperative domains where nodes collectively provide cache space for requested content, reducing the latency associated with downloading content from cloud servers. We then analyze the download latency and caching costs associated with storing content on edge nodes and within cooperative caching domains, formulating a content caching problem. To minimize overall system costs, a DQN-based scheme is proposed to optimize caching decisions. Simulation results demonstrate that the proposed scheme can significantly improve system utility.