Accelerating ultrashort pulse laser micromachining process comprehensive optimization using a machine learning cycle design strategy integrated with a physical model

The demand for industrial development toward advanced and precision manufacturing has sparked interest in ultrafast laser-based micromachining methods, particularly emerging advanced machining methods, such as laser-induced plasma micromachining (LIPMM). The main challenge in laser micromachining is finding the optimal process in a large process space to achieve a comprehensive improvement in processing efficiency and quality as approaches that rely on trial-and-error are impractical. In this work, we combined data-driven machine learning and physical model into a cycle design strategy, in order to efficient capture the comprehensive optimization process of LIPMM with high material removal rate and high microgroove depth. Based on the small sample dataset and additional physical variables provided by the physical model, the optimal process in the whole process space can be obtained using only four design cycles and dozens of data groups, and the material removal rate and microgroove depth of which are improved comprehensively compared with the original data. The design strategy integrated with physical model presented here could be applied in a wide range of fields, and thus shows the promise of accelerating the development of laser micromachining processes.