EV Health Monitoring

Non-invasive high-precision current sensing and its application to health monitoring of electric vehicles (EV)

Primary Investigators: Prof. Kaushik Roy & Prof. Byunghoo Jung

Research Assistant: Chamika M. Liyanagedera

[Motivation]

In the era of the proliferation of electric vehicles, there is one importance issue that is getting a spotlight less than deserved -- long-term reliability issue. Once issues important for rapid market penetration and deployment of EVs such as mileage per charging, charging speed and production cost are largely addressed, the long-term reliability will emerge as a crucial issue for the growth of EV market. Every EV automakers would love to tie their brands to reliability, since reliability is a key for superior customer experience and sales. This work is about the long-term reliability of EVs. The battery to EV is like the heart to the human body. As the heart pumps blood everywhere for the proper function of the human body, the battery pumps currents everywhere for the proper function of EV. One can easily imagine that, as there are several noticeable symptoms before the catastrophic failure in the heart and the other parts of the human body, there must be noticeable symptoms before the failure in the battery and the other parts of EV. Consequently, as monitoring the blood flow pumped by the heart is crucial for human health condition monitoring, monitoring the currents pumped by battery is crucial for EV health condition monitoring. There are two different approaches for EV current sensing: (1) series resistive sensing that is accurate but invasive, and (2) sensing utilizing magnetic field sensors such as Hall-effect sensor that is potentially non-invasive but inaccurate. Because of the limitations, none of the existing sensing systems can enable an intelligent health condition monitoring of EVs.

[Project Goals]

Develop:

  • A non-invasive current sensing module with an accuracy better than that of invasive resistive sensors

  • Utilize the new sensing technique to enable a deeper level of EV health condition monitoring that is not feasible currently

  • And, eventually, combined with data fusion and deep learning, provide health condition diagnosis that can significantly improve the long-term reliability and safety, and reduce the maintenance costs of EV

[Uniqueness -- Sensing Capability]

  • Non-invasive -- just wrap around the wire without removing insulator

  • High-precision -- better than that of invasive resistive sensors

  • Wide dynamic range

  • High update rate

  • High interference rejection

  • Flexible digital connectivity: CAN, I2C, SPI, UART, USB and Bluetooth (BLE)

[Uniqueness -- Health Monitoring & Diagnosis]

  • Flexible digital connectivity: CAN, I2C, SPI, UART, USB and Bluetooth (BLE)

  • Platform independent SDK: Windows, iOS, Android, AndroidAuto, CarPlay

  • Sensor fusion engine

  • Advanced monitoring engine

  • Machine learning engine

[Invited Talk]

2018 International Conference on Electric Vehicle, Smart Grid and Information Technology

"Health condition monitoring and diagnosis of electric vehicles"