Abstract
Quality prediction and condition monitoring are crucial for realizing zero-defect intelligent manufacturing. Surface roughness is an important technical indicator to measure the surface quality of parts. In the cutting process, tool wear is directly related to the dynamic change of the surface roughness and the machining efficiency. To enhance data utilization and reduce computational costs, this paper develops a novel two-task simultaneous monitoring method for surface roughness and tool wear. First, the enhancement layer corresponding to each sub-task in the broad learning system is replaced with a reservoir with echo state characteristics, and through information sharing between sub-tasks and the capture of their respective dynamic characteristics, a broad echo state two-task learning system with incremental learning is constructed. Then, an end-face milling machining experiment was conducted on a vertical machining center, and the feasibility of the developed method for two-task simultaneous monitoring of surface roughness and tool wear was verified through feature extraction and nonlinear dimensionality reduction of the monitoring signals during the machining process. The experimental results indicate that the developed two-task simultaneous monitoring method is superior to other two-task learning methods in terms of comprehensive performance, and it lays a solid foundation for the high-quality and efficient machining of CNC machine tools.