Abstract: In order to effectively improve the machining accuracy of CNC machine tools, a research on thermal error compensation of CNC machine tools based on data twin virtual scene mapping is proposed. Build a digital twin model by placing temperature sensors in key parts of the machine tool to collect real-time operational data and synchronously map it to the digital twin model. Construct a neural network model using collected data to predict thermal errors in CNC machine tools. Use the predicted thermal error value of the neural network model as the input of the PID controller, and set the target error value according to the preset machining accuracy requirements. The PID controller calculates the thermal error compensation amount in real time based on parameters such as current error and error change rate, and feeds it back to the control system of the CNC machine tool to complete the thermal error compensation of the machine tool. The experimental results show that the thermal error compensation method for CNC machine tools based on data twin virtual scene mapping has high compensation rate and stability, providing strong technical support for high-precision CNC machining.