Fig. 9: Results of Bayesian Optimization Task.
From: Physics-Informed Bayesian learning of electrohydrodynamic polymer jet printing dynamics
Performing Bayesian Optimization to find the minimum lag-distance (\({L}_{j}\)) by fitting a Gaussian Process Model to lag distance (\({L}_{j}\)) observation data obtained from the computer vision metrology module of the GPJet framework (the distance between the red arrow and the red dashed line) for specific speed ratios (\({U}_{c}/{V}_{{jm}}\)). a–c Iterations of the Bayesian optimization algorithm until it meets termination criteria. In every case, the observation point chosen at each iteration is denoted with a black dashed line box pointed by a black arrow. d For speed ratios less than one \(({U}_{c}/{V}_{{jm}} < 1)\) the process is unstable, no straight line is formed, instead the translated coiling, alternating loops, W patterns and meanders patterns are formed (depicted with a black line), for \({U}_{c}/{V}_{{jm}}=0.23,{U}_{c}/{V}_{{jm}}=0.48,\,{U}_{c}/{V}_{{jm}}=0.64,{U}_{c}/{V}_{{jm}}=0.83\) respectively. Therefore, no lag distance (\({L}_{j}\)) observation data can be obtained from the computer vision metrology module of the GPJet framework. Filled contours (shading) represent uncertainty bounds (95% confidence intervals (CIs)). The jet profile images in a, b and c are images in grayscale (0–255) with the 0 value and the 255 value in the color bar representing the black jet profile, and the white background, respectively.
