With the rapid development of areas such as the Internet of Things (IoT) and the Internet of vehicles (IoV), the surge in traffic demand could cause network congestion and interfere with infrastructure planning. With the increasing popularity of 6G networks, wider network bandwidth also means smaller coverage of base stations (BSs). Therefore, an efficient wireless BS deployment method is urgently needed. This paper proposes a BS location optimization framework, trying to solve the problem of wireless BS location from a micro point of view, the problem is divided into traffic demand prediction and BS location optimization. Specifically, the framework is divided into physical layer, virtual layer and control layer. The physical layer contains all physical objects. In the virtual layer, we use real-time profiles based on digital twins (DT) to generate traffic. This is followed by incorporating the most important traffic logs into the forecast. In the control layer, we use the transformer architecture to predict long period network traffic and generate the traffic heat map. On the basis of forecast traffic, we design a BS location optimization module based on deep deterministic policy gradient (DDPG), and output a forward-looking BS location scheme. The experimental results show that the framework and algorithm designed in this paper can effectively improve the traffic coverage of BSs.