In order to improve the accuracy of thermal error compensation and realize thermal error compensation group control, a group control method of thermal error compensation for workshop machine tools was proposed. By adopting distributed acquisition, compensation control and centralized modeling and calculation, the expansibility of the acquisition side was effectively improved and the high performance of the service side was fully utilized. Based on CAN bus, the machine tool state information acquisition platform was constructed. Combined with PSO-BP neural network, the thermal error modeling was studied, the thermal error compensation group control system was developed, and the machine tool state information acquisition monitoring and error compensation experiments were carried out. The experimental results show that the system has good expansibility and can adapt to the acquisition and interaction of multitype machine tools. By using the thermal error compensation group control model, the positioning accuracy of the machine tool is effectively improved. The model has good engineering application prospects