A method for quantitative identification of magnetic flux leakage of fatigue cracks in ferromagnetic components

• Published in: Insight (Northampton) Vol. 62; no. 7; pp. 428 - 432
• Main Authors: Hong, Li, Cai, Jianxian, Wu, Yanxiong, Yao, Zhenjing, Qiu, Zhongchao, Teng, Yuntian
• Format: Journal Article
• Language: English
• Published: The British Institute of Non-Destructive Testing 01.07.2020
• Subjects: Fatigue Cracks; Magnetic Flux Leakage Testing; PSO-LSSVM; Quantitative Identification
• ISSN: 1354-2575; 1754-4904
• DOI: 10.1784/insi.2020.62.7.428

A modelling method integrating principal component analysis (PCA) and the least-squares support vector machine with particle swarm optimisation (PSO-LSSVM) is proposed to address the difficulties in quantitative identification of fatigue cracks. The widths and depths of fatigue cracks are quantitatively identified by establishing a non-linear mapping relationship between these dimensions and magnetic flux leakage (MFL) signals. A series of fatigue crack samples are prepared through the fatigue tensile test for the MFL detection system. A sample library is established through MFL experimentation to verify the feasibility of the method for quantitative identification of fatigue cracks based on PSO-LSSVM. The results indicate that the method is in a position to effectively and quantitatively identify the widths and depths of fatigue cracks less than 1 mm in size with a maximum error of approximately 0.3 mm.