In the background of Industry 4.0, sustainable advanced manufacturing has become a pivotal challenge, necessitating the development of technologies that minimize resource consumption and environmental impact while maintaining high-quality production. Grinding technology plays a crucial role in this domain, providing solutions for processing hard and brittle materials with microscale accuracy, particularly when the grinding process is classified as precision machining. This study addresses the gaps in sustainable grinding by presenting a comprehensive review of lubrication strategies using an improved latent Dirichlet allocation (LDA) model integrated with Principal component analysis (PCA). The proposed model systematically identifies keyword distributions and topic clusters related to lubricant selection, providing insights into the evolving trends and practices in sustainable grinding. The findings enhance understanding of the main themes and future perspectives of sustainable grinding, bridging the gap between theoretical research and practical implementation in sustainable manufacturing.

To address the complicated modeling process and inadequate explainability of current methods for improving the contour accuracy of ultraprecision machining (UPM), this study presented an Internet of Things (IoT)–based contour error compensation (CEC) framework. To achieve a convincing and real-time compensation solution, a hybrid mechanism-data-driven CEC model was created that integrated the 1DCNN-BiLSTM-attention model for predicting the axis actual positions, contour error estimation, and bidirectional compensation algorithms. Bayesian hyperparameter optimization and sensitivity analysis were used in the proposed models to improve the prediction accuracy of the actual position of each axis, with high-quality training datasets from well-designed experiments. Finally, validating the system on a three-axis ultraprecision milling machine demonstrated its superior performance. This study first demonstrated the feasibility of a deep learning approach for improving UPM accuracy, which will assist in accelerating digitalization and intellectualization for UPM.