As vehicles adopt higher-voltage electrical systems (e.g., 48V), the risk of hazardous electric arcs from component failure increases, threatening vehicle safety and reliability. Existing arc detection methods are often inaccurate, leading to false alarms or missed events. This internship will address this challenge by developing and validating next-generation, AI-based detection solutions that operate reliably under real-world conditions, ensuring the safety of future automotive systems.
- During your internship, you will establish a comprehensive arc detection database through systematic measurement and data collection across various load types: Resistive (R), Inductive (I) and Power electronic loads (P); arc types: series and parallel arcs; arc scenarios: electrode degradation, plug/connector faults, guillotine wire cuts, pendulum-type contact and intermittent contacts; environmental conditions: various noise levels, impedance variations and electromagnetic disturbances.
- You will benchmark existing arc detection methods (ZoneArc, Vbased, Ibased and other commercial solutions) through development and verification using collected data.
- In addition, you will validate them under real test conditions using test bench infrastructure and assess their performance via Monte Carlo simulations for robustness analysis.
- Furthermore, you will prototype novel AI/ML-based arc detection algorithms utilizing: Deep Learning (DL), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Support Vector Machines (SVM) and AutoEncoder architectures, AI-generative modelling to enhance and expand training datasets.
- Last but not least, you will validate all methods using professional test bench equipment. The Standalone ArcBench with electronic loads (resistive, power, current, PWM capabilities) and the Integrated ArcBench@LabCar with realistic load benches, including ZoneECU, EPS (steering system), brake systems and Cooling Fan components.