Proposed method integrates image features with neural networks to identify flow patterns in aviation engines, suggesting improved monitoring techniques.
In response to the limitations regarding applicable pipe diameters and media in the study of two-phase flow in aviation engine scavenge pipe, a flow pattern identification method that integrates image features with neural networks has been proposed, offering an innovative solution for the analysis of intricate oil–gas flow states. A small-diameter oil–gas two-phase flow experimental system was developed, utilizing high-speed imaging technology to capture 1104 sets of images representing four distinct flow patterns. Following the feature selection criteria delineated in this study, 12 multidimensional feature parameters encompassing color, texture, and shape were extracted through various image processing techniques. Comparative validation revealed that the Elman neural network significantly outperforms the back propagation neural network in flow pattern identification, achieving an impressive accuracy of 97.06%. This research elucidates the mechanisms by which oil–gas media influence flow morphology through image analysis, effectively addressing the adaptability limitations of conventional water–gas models. The intelligent flow pattern identification framework constructed herein provides substantial data support for the structural optimization of aviation engine scavenge pipes, and its non-intrusive analytical characteristics hold considerable practical value in monitoring the flow states of lubrication systems.
No takes yet. Share an insight, caveat, or question.
Lyu et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: