Remaining Life Estimation of Motor Bearing

Tan, Emilyn Jiea Min (2019) Remaining Life Estimation of Motor Bearing. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Emilyn Tan Jiea Min.pdf Restricted to Registered users only Download (2MB)

Abstract

Motor bearing is a part of machine that enables it to perform actions. Motor bearing breakdown that occurs in machines can lead to many serious consequences and the worst case that can happen is fatal accident. Hence, with a prognostic model, it can estimate the remaining life of the motor bearing and this can reduce the possibility of unexpected machine breakdown and minimize the impact brought by this situation. In this dissertation, a prognostic model is to be developed using the vibrational signals obtained from a number of motor bearings. Different statistical features are tried and different feature extraction methods are done to study the accuracy through the result obtained. The prognostic model using basic time domain analysis is improved by using Wavelet Packet Transform and Wavelet Decomposition method to allow more features to be extracted. Training tool implemented is Neural Net Fitting Tool in MATLAB Neural Network Toolbox. The predicted remaining useful life using wavelet transform method is being compared with the basic time domain analysis method. Wavelet analysis enables more features to be extracted and hence, the model is able to predict the trend of most of the motor bearing testing sets. Thus, WPD can be proposed as one of the methods to develop a prognostic model for vibration equipment. However, due to the number of training sets is very limited, the accuracy is not as high as expected. Further improvements can be done by training larger number of sets to ensure that the model is capable to deal with more conditions happen within vibrational signal.