Acoustic emission-based diagnostics and prognostics of slow rotating bearings using Bayesian techniques
Year: 2014
Type: dissertation
Abstract: Diagnostics and prognostics in rotating machinery is a subject of much on-going \nresearch. There are three approaches to diagnostics and prognostics. These include \nexperience-based approaches, data-driven techniques and model-based techniques. \nBayesian data-driven techniques are gaining widespread application in diagnostics \nand prognostics of mechanical and allied systems including slow rotating bearings, as \na result of their ability to handle the stochastic nature of the measured data well. The \naim of the study is to detect incipient damage of slow rotating bearings and develop \ndiagnostics which will be robust under changing operating conditions. Further it is \nrequired to explore and develop an optimal prognostic model for the prediction of \nremaining useful life (RUL) of slow rotating bearings. \nThis research develops a novel integrated nonlinear method for the effective feature \nextraction from acoustic emission (AE) signals and the construction of a degradation \nassessment index (DAI), which is subsequently used for the fault diagnostics of slow \nrotating bearings. A slow rotating bearing test rig was developed to measure AE data \nunder variable operational conditions. The proposed novel DAI obtained by the \nintegration of the PKPCA (polynomial kernel principal component analysis), a \nGaussian mixture model (GMM) and an exponentially weighted moving average \n(EWMA) is shown to be effective and suitable for monitoring the degradation of slow \nrotating bearings and is robust under variable operating conditions. Furthermore, this \nstudy integrates the novel DAI into alternative Bayesian methods for the prediction of \nRUL. The DAI is used as input in several Bayesian regression models such as the multi-layer perceptron (MLP), radial basis function (RBF), Bayesian linear regression \n(BLR), Gaussian mixture regression (GMR) and the Gaussian process regression \n(GPR) for RUL prediction. The combination of the DAI with the GPR model, \notherwise, known as the DAI-GPR gives the best prediction. The findings show that \nthe GPR model is suitable and effective in the prediction of RUL of slow rotating \nbearings and robust to varying operating conditions. Further, the models are also \nrobust when the training and tests sets are obtained from dependent and independent \nsamples. \nFinally, an optimal GPR for the prediction of RUL of slow rotating bearings based on \na DAI is developed. The model performance is evaluated for cases where the training \nand test samples from cross validation approach are dependent as well as when they \nare independent. The optimal GPR is obtained from the integration or combination of \nexisting simple mean and covariance functions in order to capture the observed trend \nof the bearing degradation as well as the irregularities in the data. The resulting \nintegrated GPR model provides an excellent fit to the data and improvements over the \nsimple GPR models that are based on simple mean and covariance functions. In \naddition, it achieves a near zero percentage error prediction of the RUL of slow \nrotating bearings when the training and test sets are from dependent samples but \nslightly different values when the estimation is based on independent samples. These \nfindings are robust under varying operating conditions such as loading and speed. The \nproposed methodology can be applied to nonlinear and non-stationary machine \nresponse signals and is useful for preventive machine maintenance purposes. \nKeywords: acoustic emission, Bayesian linear regression, Bayesian techniques, \ncovariance function, data-driven, degradation assessment index, diagnostics, \nexperience-based, exponentially weighted moving average, Gaussian mixture model, \nGaussian mixture regression, Gaussian process regression, integration, mean function, \nmodel-based, multi-layer perceptron, polynomial kernel principal component \nanalysis, prognostics, radial basis function, remaining useful life.
Author: Sylvester A. Aye
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
Funder: University of Pretoria
OpenAlex ID: W595174522