The maintenance and diagnosis of failures in mechanical machines, especially in induction motors, represent critical and costly challenges. Rotor failures, such as breaks or fractures, are especially problematic and can affect operational safety and efficiency. To address these challenges, intelligent fault diagnosis has become vital, employing machine learning techniques such as KNN, SVM and decision trees to prevent failures in real time. Motor current signature analysis (MCSA) stands out as a non-intrusive diagnostic technique, complemented by signal processing such as FFT and wavelet. Alternatives to MCSA include vibration signal analysis, with accelerometers capturing data and classification techniques identifying rotor and bearing faults. In this paper, the CRISP-DM model was applied, including data preprocessing, Fourier analysis and Hamming window for current and vibration signals. Machine learning models such as Random Forest and SVM were trained and evaluated, reaching an average accuracy of 90% when combining current and vibration data.