MSc and PhD in Sound and Vibration: Machine Fault Detection


Project Title

Modelling and analysis of 3D vibration data pattern recognition for fault detection in rotating machinery using statistical-analysis-based method


Supervisors

Profesor Madya Ts Dr Azma Putra

Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka

Email: azma.putra@utem.edu.my


Ts Dr Mohd Irman Ramli

Faculty of Mechanical and Manufacturing Engineering Technology, Universiti Teknikal Malaysia Melaka

Email: irman@utem.edu.my


ABSTRACT

Vibration analysis is a well-established technique in predictive maintenance to detect the root cause of faults in rotating machinery. The vibration signal has different pattern for each machinery fault. Mass unbalance and bearing defects for example, have completely different features, both in time and frequency domain. Unfortunately, to interpret the complicated signal features to draw a valid conclusion about the fault requires a strong background knowledge of vibration and signal as well as experience of industrial exposure. Thus to enable vibration analysis to be conveniently performed by non-expert vibration analyst, the requirement of knowledge-based expertise has to be reduced

Several automate machine learning methods are also available, but these are still in 2D graph representation of signal, required numerous amounts of signal data for good validation and are currently limited to bearing defect.

This research projects proposes an alternative method by utilizing the statistical-analysis-based approach to present the vibration signals in the form of 3D graphical data pattern to conveniently recognize the fault based on the uniqueness of the generated shape. The know I-kaz model is modified into three new models to suit their usage for different type of faults, namely mass unbalance, parallel misalignment, angular misalignment, bearing defects, belt drive and pump cavitation.

The fault will be simulated using Machinery Fault Simulator and the signals will be recorded and simulated using the modified I-kaz models. The generated data pattern will be analyzed and the difference between each fault will be discussed. It is expected that each fault in rotating machinery will have a unique 3D pattern that can be easily recognized and is distinct with other fault. The modified I- kaz model will also be used to simulate signals containing multiple faults to test the robustness of the model to detect the machinery faults.


Requirements:

  1. Malaysian or Non-Malaysian
  2. Bachelor Degree in Mechanical Engineering or related field
  3. CGPA > 3.0
  4. Good writing & speaking in English
  5. Highly motivated and hard working
  6. Must be registered as MSc or PhD student in UTeM (once agreed to be involved in the project)


Benefits:

Monthly allowance up to RM1800


If you are interested, please fill in the FORM here.