Neural network based incipient fault detection of induction motors

M. Rokonuzzaman, M.A. Rahman
Proceedings of IEEE/IAS International Conference on Industrial Automation and Control  
An incipi ent fault detection scheme of inductio n moto rs thro ugh the recognition of frequency spect ra of the stat or current has been developed in this thes is. It is based on the adap tive resonance th eory of neural networks. T his fault diagnosi s scheme is not only capab le of detecting a fau lt but also can repor t if it canno t ident ify a parti cular fault so that necessary preventive steps can be taken to update the underl ying neural network to adapt to this undetected fault .
more » ... tected fault . Moreover , it can update its elf to cope with t his dyn amic situation ret aining al ready acquired knowledge with out the need of ret rain ing with th e old patt ern s. A laboratory experimental set-up using a digital signal proc essing(DSP) technique has been employed to collect the frequency spect ra of the st ator curren t at different fault conditions. A wound -rotor ind uction motor has been used as the test motor to create different types of faults making unbala nce in the st ator and rotor cir cuits. A 24-bit high spee d nsp boa rd has been used with a personal computer to develop a reel-t ime interac tive software to collect the spectr a . A driver for the UP-plotter has also been developed to directly plot the frequency spectra of the stator cu rrent. Adapt ive resonance theory(ART ) based network is a recent additio n to the neural network family. A Dew software bas been su ccessfully developed and implemented in the laboratory experimen t using ART neu ral network.
doi:10.1109/iacc.1995.465853 fatcat:mhqqejvanbhddcorcbm5w7432e