Bayesian Modelling and Uncertainty Analysis for Wire Rope Defect Signal Recognition

Steel wire rope is usually operating under the sophisticated servicing conditions in various practical application fields, which is playing a significant role in promoting the fast development of domestic and global economy. However, the environmental interferences and three-dimensional spiral geometric structures make the extract defect inspection and health monitoring of wire rope full of challenges and difficulties, which lead to the service site and related mechanical objects always in a dangerous state. Thus, a novel Bayesian modelling and uncertainty analysis method based on stochastic process characterization and abnormal sequence analysis are proposed here for wire rope defect recognition. First, the wire rope testing conditions and raw datasets acquired through non-destructive testing (NDT) method are described. Then, these time series signals are preliminarily analyzed both in time and frequency domain. Second, the principles of the new Bayesian abnormal inspection and Markov chain Monte Carlo chain (MCMC) model are introduced, and the time series recognition from the perspective of abnormal sequence inspection and statistics are both tested and compared. It is found that the new Bayesian modeling method is more effective in abnormal wire rope defect signal recognition than common methods. Finally, the defect recognition performance through the stochastic subspace system and the uncertainty analysis is evaluated and the feasibility of the proposed method is validated. Additionally, the recognition features, application promises and future work are discussed.