Research Of The Fault Diagnosis Technology About Cage Asynchronous Traction Motor Of Electric Locomotive

At present, AC motor drive system constituted by AC-DC-AC voltage type converter and AC cage asynchronous traction motor has been the most popular electrical drive technique used in electric locomotive in the world, cage asynchronous traction motor is one of the most important parts, great loss will be brought out in case of fault. Therefore, to adapt advanced fault detection and diagnosis equipment become very vital. This paper makes research in fault reasons and diagnosis methods of traction motor on the basis of former students. Several methods have been applied to diagnosis of traction motor, which provided realizable scheme for diagnosis application.This paper has studied the detection and diagnosis technology for failure of cage asynchronous traction motor with broken rotor bars and rolling bearing. As to the broken rotor bars in motor, this paper uses continuous wavelet transform to analyze load oscillation under steady running state. Therefore, current signal interfered by load oscillation has been removed. Furthermore, a kind of new test technology combining Hilbert transform demodulation technique and spectrum zooming method on Chirp Z transform has been put forward for current signal without load oscillation, and the technology has achieved the goal of extraction of fault features and fault identification. Concerning the rolling bearing fault in motor, this paper presents a method combining wavelet package, envelope demodulation technology and fuzzy mathematics to extract fault features, which can reflect condition of bearing fully. The paper also uses the improved back-propagation neural network to realize earlier fault's classification and identification and achieve the goal of fault diagnosis.This paper brings forward the hardware and software design of fault diagnosis system based on the research of cage asynchronous traction motor of DJ₁ AC electric locomotive, The validity of the above mentioned methods has been analyzed by fault data and the simulation supports it.