Achieving a balance high productivity and maximum geometric precision remains challenging in laser micro-drilling. Specifically, the precise control of inlet and outlet diameters is crucial for the quality of the machining results. This paper presents a highly efficient closed-loop approach to multi-criteria optimization of a single-pulse laser drilling process on thin metal foils.

Temporally shaped laser pulses in the microsecond range were utilized for this process. This technical approach utilizes pulse shapes comprising of multiple plateaus, each with power levels ranging from 0 to 1,300 W and plateau durations exceeding 0.5 µs.

The vast parameter space of temporally shaped pulses limits the effectiveness of conventional, labor-intensive trial-and-error methods or traditional Design of Experiments are ineffective. Thus, a Bayesian Optimization (BO) framework was implemented. To ensure precise evaluation of the drilled holes, an AI-based image analysis was used to automatically measure and extract geometric features, such as inlet and outlet diameters. Optimization objectives included achieving specific target diameters, maximizing reproducibility, and minimizing pulse energy.

The results demonstrate that the BO algorithm identifies optimal pulse parameter configurations in fewer than 50 iterations. Employing this methodology, reproducible micro-holes with diameters in the double-digit micrometer range and standard deviations of less than 1 µm were successfully realized. The findings of this research substantiate the efficacy of integrating temporal pulse shaping with AI-driven optimization. This approach provides a robust and scalable framework for high-precision industrial laser applications with extreme throughput requirements.