Background <p>Bearing faults, particularly outer race defects and angular misalignment across coupling, significantly impact the performance and reliability of rotating machinery, leading to&#xa0;increased vibration, instability, and potential system failure.</p> Methods <p>This study investigates the vibration characteristics of a rotor-bearing system with outer race faults and angular misalignment between the rotor and motor shaft across coupling, focusing on the sugar industry. An experimental setup simulates these defects using Electrode Discharge Machining (EDM) to create outer race defects and inserting thin strips beneath the bearing block to induce angular misalignment. A dynamic model, employing dimensional theory, is developed and validated through experimental data analysis. To enhance fault detection accuracy, a hybrid approach integrating expert knowledge with a 1D-CNN (one-dimensional Convolutional Neural Network) for automated feature extraction and failure mode classification is proposed.</p> Results <p>Angular misalignment exacerbates variable compliance excitation, leading to increased vibration acceleration and directional instability in the rotor-bearing system. The study reveals amplification of vibration spectra components due to angular misalignment, with a decrease in rotational frequency amplitude. The 1D-CNN-based detection method achieves a fault diagnosis accuracy of 98%, demonstrating its efficacy in real-time bearing health monitoring.</p> Conclusion <p>This research provides valuable insights into mitigating angular misalignment and outer race defects in rotating machinery, thereby enhancing operational stability and reliability. The findings advocate for the adoption of advanced monitoring techniques like 1D-CNNs for improved fault detection and maintenance strategies in industrial applications.</p>