Level set based image segmentation for oil film monitoring using ECT

Abstract

In air/oil lubrication system, the flow regime and oil film thickness are significant for lubrication efficiency and can only be monitored online. Electrical capacitance tomography (ECT) provides a promising strategy for monitoring air/oil two phase flow in the pipelines by reconstructing cross sectional oil distributions with good real-time performance. While the low spatial resolution of ECT reconstructions has seriously influenced the visualization of the interface of oil and air. To apply ECT to oil film monitoring and film thickness estimation, a modified level set-based image segmentation technique is introduced: a distance regularized level set evolution (DRLSE) formulation is adopted to accelerate the computation process while guaranteeing the stability of the evolution, and since the interface propagates in different directions (inwards or outwards) at different regions of the cross section, the image is divided into 4 regions, and contour evolutions are carried out in different regions respectively using local void fractions to determine the direction of evolution (inward or outward). Experiment is carried out using an 8-electrode ECT system installed on an air/oil lubrication facility, the results demonstrate that the proposed method is capable of visualizing the air/oil interface online.