Condition Monitoring of Robust Damage of Cantilever Shaft Using Experimental and Adaptive Neuro-fuzzy Inference System (ANFIS). (2016)

• Record Type: Journal Article
• Authors: Das, Sandeep; Nayak, Biswajit; Sarangi, Saroj Kumar; Biswal, Dillip Kumar
• Is Part Of: Procedia engineering. Volume 144(2016)
• Page Start: 328
• Page End: 335
• Publication Date: 2016
• Subjects: Shaft -- Natural Frequency -- ANFIS -- Crack Location -- Crack Depth.; Engineering -- Congresses; Engineering -- Periodicals; Engineering; Conference proceedings; Periodicals; 620.005
• DOI: 10.1016/j.proeng.2016.05.140
• Languages: English
• ISSNs: 1877-7058

Abstract

Abstract: Presently most of the failures encountered by machines are due to material fatigue. Therefore crack detection and localization is the main topic of discussion for various researchers across the globe. The dynamic behaviour of a whole structure is affected due to the presence of a crack as the stiffness of that structural element is altered. The cracks in the structure change the frequencies, amplitudes of free vibration and dynamic stability areas to an inevitable extent. In this work the effect of an open crack on the modal parameters of the cantilever shaft subjected to free vibration is analyzed. The results obtained from the experimental approach have been verified with the results obtained from finite element analysis using ANSIS 13. A methodology has been developed to predict fatigue crack propagation life of mild steel shaft. It has been assessed by adopting Adaptive Neuro-fuzzy Inference System (ANFIS), a novel soft-computing approach, suitable for non-linear, noisy and complex problems like fatigue. The proposed hybrid neuro-fuzzy system combines the learning capabilities of neural networks with fuzzy inference system for nonlinear function approximation. A single-output Sugeno-type Fuzzy Inference System (FIS) using grid partitioning has been modelled in this work. After comparing the output, it has been found that the proposed model has proved its efficiency quite satisfactorily.