A practical investigation to solving the inverse problem of crack identification through vibration measurements

Abstract

PurposeTo investigate the feasibility of using single/multi variable optimisation techniques with vibration measurements in solving the inverse crack identification problem.Design/methodology/approachThe finite element method is used to solve the forward crack problem with a special nodal crack force approach. The multi‐variable optimisation approach is reduced to a much more efficient single‐variable one by decoupling the physical variables in the problem.FindingsIt is shown that, for the crack identification problem, global optimisation algorithms perform much better than other algorithms relying heavily on objective function gradients. Simultaneous identification of crack size and location proved to be difficult. Decoupling of the physical variable is introduced and proved to provide efficient results with single‐variable optimisation algorithms.Research limitations/implicationsNeed for improving the reliability and accuracy of the procedure for smaller crack sizes. Need for developing and investigation more rigorous and robust multi‐variable optimisation algorithm.Practical implicationsAny information about approximate crack size and location provides significant aid in the maintenance and online monitoring of rotating equipment.Originality/valueThe paper offers practical approach and procedure for online monitoring and crack identification of slow rotating equipment.