Visual identification of oscillatory two-phase flow with complex flow patterns

We present an approach based on computer vision and machine learning methods to identify two-phase flow with complex patterns in oscillatory conditions. A visualization experiment bench was designed, constructed, used simulate the actual reciprocating motion of cooling gallery inside piston low-speed diesel engines. The results our proposed show that feature vectors extracted from optical images provides a valuable reference for velocity flow. it is possible videos respect Reynolds numbers 10568 31704 using Bayesian Network classifier, best accuracy 94%. purposed this paper can not only be validating sources numerical simulation results, but also widely applied multiphase flow, which key area developed basic research heat transfer systems.