In order to improve the accuracy and adaptability of thermal error prediction for NC machine tools, a thermal error modeling and prediction method for NC machine tools based on sequence deep learning network was proposed.A sequence deep learning prediction network based on LSTM was presented. The dynamic data matrix containing historical sequence data was established as input of the model.The truncated training method was used to reduce the complexity of each parameter update in the deep prediction network. The timing feature of the temperature time sequence was automatically extracted by using the sequence deep learning network, and the complex mapping relationship between temperature sequence signal and thermal error was accurately characterized.Thermal error was accurately predicted through the Softmax output layer.The experimental results show that by this method,the problems of low accuracy and poor robustness of the traditional shallow method because it does not consider the influence of historical sequence data on the current output are solved, and the root mean square error of thermal error prediction is reduced to less than 2.5 μm.The proposed method is better than the traditional SVM and BP prediction methods.It provides reference for thermal error compensation of NC machine tools.