Design of an intralayer data-based controller for laser powder bed fusion processing

This paper proposes a data-based approach to reduce defect-prone process conditions during the additive manufacturing of Ti-6Al-4V alloy parts using the laser powder bed fusion technique. The focus is on controlling melt pool emissions intensity used as a temperature-related proxy generated as the laser interacts with the alloy powder to avoid processing conditions which generate defects such as lack-of-fusion (LOF) or keyhole porosity. A data-based controller is developed using photodiode signals from the melt pool emissions to predict future melt pool data and then update the control input accordingly. The prediction is achieved by estimating the Markov parameters and constructing a finite impulse response (FIR) model of the process. This controller, termed the data-based predictive FIR-based proportional controller, adjusts the laser energy input by applying a proportional correction to deviations of the melt pool emissions from their setpoint, thereby mitigating under- or overheating of the melt pool. The developed approach is investigated for the printing of spinal cage medical devices and evaluated through a closed-loop MATLAB simulation to assess its potential for intralayer control. A threshold-based anomaly detection approach is then employed to quantitatively evaluate the controller’s effectiveness in reducing defect-prone signal excursions. Results demonstrate a 99.98% reduction in the number of samples exceeding the defined thresholds within the studied layer after applying the controller, indicating the potential of the controller to reduce defect-prone process conditions associated with LOF and keyhole porosity if successfully deployed on the L-PBF system. Furthermore, closed-loop stability and sensitivity to control parameter variations are analysed.