In this paper, a predictive maintenance strategy is developed to enhance the fault identification of induction machines using the analysis of stator current. Induction machines (IMs) are robust and easy to maintain. However, a minor fault in a single rotor bar can cause it to break, subsequently causing a large adjacent bar to break due to the vibrations created by the imbalance in the rotor. It is, therefore, essential to detect any minor failures early to avoid further damage and reduce maintenance costs. This paper proposes a new approach to detecting incipient defects, using the cyclo-stationarity of the electrical signal rather than the vibration signal. This approach combines Discrete Wavelet Transform (DWT) with the new Residual Electrical Signal Extraction (RESE) method and fuzzy logic to enable the identification, localization, and classification of rotor faults. The results revealed that these newly developed strategies can be used to identify and then classify the level of cracking in a single rotor bar, even under different load levels.