A practical prediction method for grinding accuracy based on multi-source data fusion in manufacturing

The quality of workpieces is affected by many factors, such as machine tool errors, and their machining accuracy needs to be improved. Therefore, an accuracy prediction method based on the attention convolutional long short-term memory neural network (ACLSTM) is proposed in this paper. According to an analysis of the operational data of certain equipment, such as the temperature, the current and the rotational speed of each motion axis of the machine tool, this method completes the prediction of the workpiece grinding accuracy. The experimental results show that the ACLSTM method is able to quickly and accurately predict the actual workpiece size after processing. The result of the proposed method was compared with other conventional regression prediction methods, and the performance of ACLSTM is significantly better than other methods, which can be practically applied to the workpiece size prediction in industrial processing to further control processing quality.