In recent years, the consumer electronics manufacturing (CEM) has increasingly recognized the role of the Internet of Things (IoT) in improving production efficiency and reducing costs. Manufacturing factories have chosen to deploy servers at the edge of network to cache services, reducing energy consumption during data transmission. However, due to the dynamic nature of edge networks and the unpredictability of service requests, obtaining the optimal caching strategy for IoT devices remains a significant challenge. In this paper, we employ digital twin (DT) to formulate dynamic digital models of IoT devices and edge servers for enhancing the caching management efficiency in manufacturing factories. Additionally, we propose a service caching scheme based on deep reinforcement learning (DRL) enabled by DT, named SCRD, to obtain the optimal caching strategy for IoT devices. Firstly, the service caching problem is formulated as a Markov decision process (MDP), which is then solved using a multi-agent algorithm based on the Double Dueling Deep Q-Network (D3QN). Finally, experimental results demonstrate the proposed scheme is more effective than the baseline schemes.