Intelligent sensor systems and low power predictive maintenance for automotive applications

The growing market for high-performance electric vehicles demands motors that deliver exceptional power density while maintaining high reliability. This dual requirement necessitates operating electric motors at their maximum thermal and mechanical limits, which in turn demands continuous state-of-health monitoring to prevent failures. This thesis addresses this challenge by targeting two of the most predominant failure modes: thermal demagnetization of rotor magnets and mechanical bearing degradation, a primary cause of mechanical breakdowns. To address the critical challenge of thermal monitoring, this work presents the iterative design and development of a novel rotor temperature sensor across three distinct iterations. Each version is built upon the lessons learned from its predecessor, allowing the exploration and testing of different technological approaches. The primary objective remains consistent throughout the iterations: to create a solution that reliably tracks rotor magnet temperatures while prioritizing a high ease of integration within the motor assembly. The design methodology focuses on progressively reducing the sensor's invasiveness, simplifying the installation process, and ensuring the final component remains low-cost and accessible for widespread adoption. To address mechanical reliability, a novel vibration sensor was designed for high integration, innovatively sharing mounting components with the motor's existing position sensor. This strategic placement can capture vibrational signatures essential for the early detection of mechanical faults. The dataset collected from this sensor was used to process and benchmark the performance of different anomaly detection algorithms, validating the solution's effectiveness in identifying incipient bearing degradation in a real case scenario.