The damage to opposite wall caused by excessive ablation will severely affect the performance of the turbine blade in femtosecond laser drilling of film cooling holes. Real-time monitoring and control of drilling stage is regarded as one of the most effective methods for solving the excessive ablation problem. However, due to the complexity of femtosecond laser drilling process, it is difficult to model the relationship between drilling state and machining phenomenon physically. In this paper, the quantitative analysis of femtosecond laser drilling and an intelligent methodology for real-time monitoring and control of drilling stage were developed. The laser drilling was comprehensively analyzed and four drilling stage was firstly divided automatically based on the hole evolution process and the optical emission signal detected in drilling process. And the correlation between signal and drilling stages was established through the analysis of the optical emission signal. Furthermore, this paper proposed a data-driven method to build the classification model of different drilling stages. The periodicity and pulse characteristics of signal and corresponding time-domain features were extracted for further modelling. Finally, the classification model was established via AdaBoost method, which achieves a classification accuracy of 95.22 %. And the generalization performance for real-monitoring of drilling stage was further verified. This study presented a set of processes for real-time monitoring and process control in laser drilling process, which provides a new insight to solve the excessive ablation problem.
Read full abstract