Tino Band, Benedikt Karrasch, Markus Patzold, Chia-Mei Lin, Ralph Gottschalg, Kai Kaufmann
{"title":"Simulation-Trained Neural Networks for Automatable Crack Detection in Magnetic Field Images","authors":"Tino Band, Benedikt Karrasch, Markus Patzold, Chia-Mei Lin, Ralph Gottschalg, Kai Kaufmann","doi":"10.1007/s10921-023-01034-9","DOIUrl":null,"url":null,"abstract":"<div><p>Magnetic field measurements play a vital role in various industries, particularly in the detection of cracks using magnetic field images, also known as magnetic field leakage testing. This paper presents an approach to automate the extraction of crack signals in magnetic field imaging by using neural networks. The proposed method relies on simulation-based training using the lightweight Python library Magpylib to calculate the three-dimensional static magnetic field of permanent magnets with surface defects. This approach has numerous advantages. It allows control of training data set variance by tuning simulation input parameters such as sample magnetization, measurement parameters, and defect properties to cover a wide range of cracks in size and position. Starting data acquisition before system operation allows investigating potential changes in sample shape or measurement parameters. Importantly, simulation-based data generation eliminates the need for physical measurements, leading to significant time savings. The study presents and discusses results obtained on two different ferromagnetic samples with surface cracks, a hollow cylinder and a steel sheet.</p></div>","PeriodicalId":655,"journal":{"name":"Journal of Nondestructive Evaluation","volume":null,"pages":null},"PeriodicalIF":2.6000,"publicationDate":"2024-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Nondestructive Evaluation","FirstCategoryId":"88","ListUrlMain":"https://link.springer.com/article/10.1007/s10921-023-01034-9","RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MATERIALS SCIENCE, CHARACTERIZATION & TESTING","Score":null,"Total":0}
引用次数: 0
Abstract
Magnetic field measurements play a vital role in various industries, particularly in the detection of cracks using magnetic field images, also known as magnetic field leakage testing. This paper presents an approach to automate the extraction of crack signals in magnetic field imaging by using neural networks. The proposed method relies on simulation-based training using the lightweight Python library Magpylib to calculate the three-dimensional static magnetic field of permanent magnets with surface defects. This approach has numerous advantages. It allows control of training data set variance by tuning simulation input parameters such as sample magnetization, measurement parameters, and defect properties to cover a wide range of cracks in size and position. Starting data acquisition before system operation allows investigating potential changes in sample shape or measurement parameters. Importantly, simulation-based data generation eliminates the need for physical measurements, leading to significant time savings. The study presents and discusses results obtained on two different ferromagnetic samples with surface cracks, a hollow cylinder and a steel sheet.
期刊介绍:
Journal of Nondestructive Evaluation provides a forum for the broad range of scientific and engineering activities involved in developing a quantitative nondestructive evaluation (NDE) capability. This interdisciplinary journal publishes papers on the development of new equipment, analyses, and approaches to nondestructive measurements.