Detecting and diagnosing faults in industrial systems is important for ensuring their reliable operation. Traditional onboard fault diagnosis methods often rely on numerous sensors and complex techniques, limiting operational capability. This work develops an onboard fault diagnostic system with lightweight decision tree-based (DT) feature analysis. The approach uses a DT-based model to extract relevant features and adapt them to specific faults. Additionally, it investigates the severity of stator winding insulation failures by combining deep learning with time–frequency-based features. The initial phase of stator winding insulation degradation, known as a stator turn-turn fault (STTF), is critical to monitor in industrial machines to prevent catastrophic failures. An experimental test-rig setup mimics various STTFs, simulating real-time industrial conditions, was developed to validate the approach. This work demonstrates the reliability and scalability of the model, making it suitable for industrial applications.