Title: Adaptive neuro fuzzy inference system in modelling/detecting cracks and porosity using liquid penetrant test
Authors: Bharat Mehta; Raman Bedi
Addresses: Department of Mechanical Engineering, National Institute of Technology, Jalandhar-144011, India ' Department of Mechanical Engineering, National Institute of Technology, Jalandhar-144011, India
Abstract: This research paper is concerned with a successfully developed adaptive neuro-fuzzy inference system (ANFIS) for detection of indication size (of the penetrant) for liquid penetrant test. The evaluation of the neuro-fuzzy model has been comprehensive; it has been performed using a database of indication size containing 252 valid data points in a structural steel sample and 468 valid data points in high-alloy steel sample. The surface discontinuities are artificially drilled holes varying in diameter and depth from 0.5 to 1.75 mm, respectively, with an increment of 0.25 mm between each hole. The test is conducted at varying time intervals from 2 to 30 min for structural steel samples and 2 to 60 min for high-alloy steel samples, respectively, and the results are obtained till 1/100th of an mm. The ANFIS modelling accuracy is very high, with R² values reaching about 0.9924 for the test set when 83:17 ratios for train set:test set are taken. Use of this technique can be useful as this would help with the correct detection of discontinuities and in deciding whether to reject or accept a sample.
Keywords: adaptive neuro fuzzy inference systems; ANFIS; dye penetrant test; non-destructive evaluation; NDE; pin holes; subtractive clustering; crack modelling; crack detection; cracks; porosity; liquid penetrant test; structural steel; high-alloy steel; surface discontinuities.
International Journal of Experimental Design and Process Optimisation, 2016 Vol.5 No.1/2, pp.117 - 132
Available online: 25 Nov 2016 *Full-text access for editors Access for subscribers Free access Comment on this article