ESC 2003 - University of Sheffield

Report
Robust Fault analysis Technique for
Permanent Magnet DC Motor In safety
Critical Applications
Wathiq Abed
Supervisor - Sanjay Sharma
University of Plymouth
UKACC PhD Presentation Showcase
Introduction
 Condition monitoring and fault diagnosis of the electrical motor
motor are thus necessary to optimise maintenance and improve
reliability levels in electrical power system especially for critical
applications. The isolation of the fault in time ensures that integrity
of the power system and performance of the overall system are un
affected.
 Artificial Intelligence has been introduced into the fault diagnosis
process for condition monitoring
 Background and motivation for research
 Bearing faults are common faults in electric motors and
represent about 40% to 50% of all motor faults .
 Condition monitoring and fault diagnosis includes data
acquisition, signal processing and fault identification
UKACC PhD Presentation Showcase
Slide 2
 Research methodology
 Experimental set up to collect data under normal and abnormal
operating conditions
 Discrete wavelet transform for feature extraction
 To avoid feature redundancy that effect on fault classification
accuracy OFNDA have been applied as features reduction tool
 Wavelet activated neural network nonlinear model
 Online real time fault classification using Dynamic neural network
(DNN)
 Contribution to knowledge
 A novel approach for identifying rolling element bearing defects in
brushless DC motors under stationary and non-stationary operating
conditions with different severities of fault
UKACC PhD Presentation Showcase
Slide 3
Fault diagnosis Results
DNN for fault classification
Inner race re
Outer race re
Ball defect re
DNN performance re
UKACC PhD Presentation Showcase
Slide 4
Conclusion
 Dynamic neural networks are more versatile and provide the
capability to learn the dynamics of complicated nonlinear
systems which conventional static neural networks cannot
model
 Orthogonal fuzzy neighbourhood discriminative analysis were
applied to obtain the best features for fault classification
UKACC PhD Presentation Showcase
Slide 5

similar documents