In response to the complex working conditions of modern presses, which make it difficult to trace and decouple machining accuracy, and traditional methods are difficult to achieve accurate prediction of machining accuracy, this paper proposes a machining accuracy prediction method based on multi-scale attention convolutional neural networks. Taking the dead center position data of the press as the object, a method combined multi-scale analysis and convolutional neural network is proposed to extract the features. At the same time, the importance of the features is weighted using the Squeeze Extraction Network, achieving accurate feature extraction. The case study shows that this method can accurately predict the precision state and be applied to the actual machining prediction, and the defect rate of the product is effectively reduced.