Characterization of wear-debris group in on-line visual ferrographic images

Abstract

Wear-debris characterization using ferrography, which is of importance in machine condition monitoring and fault diagnosis, remains a challenge. The newly developed on-line visual ferrograph can provide digital wear-debris images and perform on-line analysis during wear monitoring. In this article, image projection transformation was utilized for extracting the overall characteristics of the wear-debris chains according to the fundamental feature of wear-debris arraying along the horizontal direction. Moreover, Full Binary Tree Based Image Division was also proposed to analyse the regional features on different scales in on-line visual ferrographic images. Several descriptive parameters including thinning ratio, chain length and chain width were proposed. In the experiments, four types of images with different wear-debris groups were compared and a group of time-sequence on-line visual ferrographic images of an inline four-cylinder gasoline engine was studied. It is found that more comprehensive wear information can be acquired through the proposed multi-parameter description method. Meanwhile, the image projection transformation method extracts macrocharacteristics of images rapidly and efficiently and the thinning ratio measures wear-debris width in an intuitive way.

Wear particle classification using genetic programming evolved features

“…In previous research, noise of wear particle images is removed using a morphological filtering method. Corrosion expansion, being the major operation mode of morphological filtering, is widely used in the denoising of ferrographic images . Different structural elements need to be chosen to process different images, which makes the morphological filtering inaccurate and inefficient.…”

“…Genetic programming uses the concept of evolutionary algorithms to evolve a population of computer programs presented as expression trees, which is dynamically self‐adaptive to denote individuals and more suitable for expressing complex problems . Hence, GP can be adapted to accommodate a wide range of problems in science and engineering fields, such as physics, helicopter hovering control, decision making, and pattern recognition . Genetic programming is applied for the first time to generate new features for roller bearing fault classification .…”

Correction strategies of debris concentration for engine wear monitoring via online visual ferrograph

“…deposition, filtering, decomposition and leakage) that affect the concentration of debris in an engine; this adoption blurred the influence of those factors on the debris concentration. Wu et al 2,3 have determined and analysed the relationships among the concentration of ferromagnetic particles in an engine lubrication system, the rate of generation of wear particles and the ferrograph image of the debris. Lu et al 4 adopted a transfer function to describe the dynamic variation of the debris concentration in a lubrication system.…”

“…Wu et al. 2,3 have determined and analysed the relationships among the concentration of ferromagnetic particles in an engine lubrication system, the rate of generation of wear particles and the ferrograph image of the debris. Lu et al.…”

A microfluidic device for three-dimensional wear debris imaging in online condition monitoring

“…Similar to the analytical ferrograph, 24 wear debris attracted by the magnetic force always form into agglomerated chains. Although efforts have been made to separate particles in these chains, 25 it is still difficult to extract features of individual particles. In contrast, the LNF system, developed by Lockheed Martin Tactical Defence Systems in cooperation with the Naval Research Laboratory, employs a flow cell without using magnetism to acquire wear debris in motion.…”