SOC estimation industrial state of the art

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
  • #2527

    Hello everybody,

    I am Emil, in the past 7 years I have been a researchers around battery models and BMSs. I recently joined LEM, a company that focuses on electrical measurement devices for, among others, batteries.

    During my years as a researcher I have read tons of publications about all kinds of methods for state of charge (SOC) estimation. Filters including various kind of models (equivalent circuits, electrochemical, etc.), various flavours of machine learning, methods based on “exotic” measurements such as ultrasound or pressure, etcetera.

    From my – admittedly limited – interactions with industry, I got the impression that a “fine tuned coulomb counting” is actually still the most widespread method out there. Most of my interactions have been with people in small to medium companies, which probably simply didn’t have the manpower to optimize their algorithms. Still, often they seemed “content enough” with their coulomb counting.

    In short, I perceived a gap between research and industry, and I am curious to understand if this is due to my point of observation or other people share this impression.

    In your experience, what is the industry state of the art around SOC estimation? And why is that?
    1) Coulomb counting with OCV correction at rest;
    2) Filter (Kalman or other) based on equivalent circuit model;
    3) Filter based on an electrochemical model;
    4) Other methods, based on machine learning / data-driven approaches;
    5) Other methods, based on further measurements (e.g. impedance, ultrasound, etc.);
    6) Other methods, not in this list;
    And how do you think this will evolve in the near future?

    I am really curious to get other points of view on this topic! 🙂




    Hi Emil,
    Very interesting thoughts. I’m not an expert as you about all the found methods about the SOC estimation, but I’m used of the coulomb counter. I think it’s difficult to find a good, cheap and reliable method.

    If I want to make a new system for a serie production in the next 3 years, I don’t want to take any risk on the SOC estimation. A coulomb counter + OCV corrections is easy to implement. Of course I need to be sure that my system will have some opportunities to relax to get some efficient OCV corrections. Also I’m sure that my system will have one (or more) current sensor and enough computing power for this method.

    I can also look for a parallel path during the development to try one other method but most of the time, it’s difficult to find some time to invest into a parallel development path. I’m not an expert here neither, I don’t know if there is many company selling some ready to use system. If you know some, I will be happy to have a look. LEM may be ?


Viewing 2 posts - 1 through 2 (of 2 total)
  • You must be logged in to reply to this topic.