Rotor Broken Bar Fault Detection Of Asynchronous Motors Based On Music Algorithm

Motor fault diagnosis can evaluate the running condition of motor, by monitoring motor's related operation parameters, such as current, voltage, flux, speed, temperature, noise, vibration and partial discharge. If fault occurs to motors, motor fault diagnosis can identify the type of fault, location and severity. The emergency of this technology is a revolution in the field of on-line monitoring. It changes traditional after-fault maintenance method to a predictive maintenance method, which enhances the safety of motor machine and reduces unnecessary losses greatly.Motor Current Signature Analysis (MCSA) is a branch of motor fault diagnosis. MCSA can detect motor machines'fault through monitoring their current. When it comes to fault diagnosis on broken rotor bars of squirrel cage induction motor, MCSA through the motor machines'state current's spectrum analysis achieve fault components. Traditional spectrum analysis uses Fourier transform. But broken rotor bars of squirrel cage induction motor result in (1 ? 2 s )f0 fault current, which is very close to fundamental frequency component and small amplitude. To get enough resolution, Fourier transform needs sufficient sampling data, which means quite a long time. This will not only increase the computation, but also let it become much easier to be influenced by the fluctuations of load. As a result, Fourier transform can not apply to fault diagnosis on broken rotor bars of squirrel cage induction motor.Multiple Signal Classification (MUSIC) is a milestone in the history of spatial spectrum estimation algorithm. Its basic idea is to decompensate the signals'autocorrelation matrix eigenvalue, and then decompensate signals'space characteristics into the signal component corresponding to the signal subspace and phase quadrate signal component with the noise subspace, using these two spaces orthogonally to estimate the frequency of the signal. It is arbitrary in theory, high resolution; the simulation also shows that it has good noise immunity and can detect weak harmonic. This paper applies MUSIC algorithm on induction motor broken rotor bars fault diagnosis and achieve satisfactory result by using short amount of data to detect broken rotor bars fault.