Abstarct: Abstract Nowadays, many industries and factories use different types of electric machines to meet the needs of society. Therefore, condition monitoring and maintenance of these types of machines is necessary and unavoidable. Nevertheless, in the recent years, various invasive and non-invasive methods have been used to diagnose induction motor faults. The motor current signature analysis method has been widely used to detect multiple types of induction motor faults. According to the various advantages of motor current signature analysis, this method has been chosen to detect and distinguish types of faults. On the other hand, low-frequency load torque oscillations lead to a false warning of rotor asymmetry fault, it is also inevitable to separate indicators of load torque oscillations from fault characteristic components. In this thesis, the normalized frequency domain energy operator has been selected to detect and separate the low frequency load torque oscillations from the fault index. Another method called variational mode decomposition and rectification is also used to detect and identify the rotor asymmetry fault. Finally, by using the third order energy operator and a high-quality time-frequency method with high resolution, the asymmetry fault in induction motors has been identified. All the methods used in this thesis have been evaluated in two steady-state and transient mode. For steady state in all strategies, Fast Fourier transform is used and for transient mode, a new and high-resolution technique is used in each chapter. Finally, after applying the proposed methods on a wound rotor induction motor, acceptable results have been obtained in the both steady state and transient mode. The results obtained in the time-frequency analysis also have a narrow band in their output response.