This paper reports the features of a splashing drop on a solid surface and the temporal evolution, which are extracted through image-sequence classification using a highly interpretable feedforward neural network (FNN) with zero hidden layer. The image sequences used for training-validation and testing of the FNN show the early-stage deformation of milli-sized ethanol drops that impact a hydrophilic glass substrate with the Weber number ranges between 31-474 (splashing threshold about 173). Specific videographing conditions and digital image processing are performed to ensure the high similarity among the image sequences. As a result, the trained FNNs achieved a test accuracy higher than 96%. Remarkably, the feature extraction shows that the trained FNN identifies the temporal evolution of the ejected secondary droplets around the aerodynamically lifted lamella and the relatively high contour of the main body as the features of a splashing drop, while the relatively short and thick lamella as the feature of a nonsplashing drop. The physical interpretation for these features and their respective temporal evolution have been identified except for the difference in contour height of the main body between splashing and nonsplashing drops. The observation reported in this study is important for the development of a data-driven simulation for modeling the deformation of a splashing drop during the impact on a solid surface.
View on arXiv