I have a question about the paper"SAE 2013: Simplified Extended Kalman Filter Observer for Battery SOC Estimation
In this paper, you model the hysteresis effect, and use k as the hysteresis factor.
how can I understand the following sentence in Page5"The value of k was obtained by inspection by minimizing the difference between the experimental and simulated VOC vs. SOC curves. VOC(VSOC) is the instantaneous value of the OCV on the minor loop."
what's the experimental VOC(VSOC),is caculted by(VOCch(VSOC)+VOCdisch(VSOC) )/2?
and what's simulated VOC(VSOC), is the difference between measured terminal voltage (Vt) and VRC?
Thank you very much for these nice examples!
We try to characterize a electronic car battery for appropriate SoC-estimation in the car.
Therefore we measured pulse data and are doing parameter estimation.
Now the question: How do you use the gained data later on in the devices (eg our car)? You normally generate look up tables and implement them in the ECU? Or you bring the complete model to the ECUs? What would be the typical way of proceeding?
Thank you in advance!
Kirankumar - You can use drive cycles (validation) data with Simulink Design Optimization to tune parameters of a model. I would encourage doing so. However, the circuit parameter values may need to have good initial guesses for them to converge.
Pulse data is nice because it allows you to isolate the contributions of the different equivalent circuit elements.
We are trying to estimate parameters for Lithium-Ion battery with Rin+R1||C1 structure. For moment we shall not consider the thermal effect. All three parameters shall be 2-D look-up tables with breakups along SOC and Ibat(temperature in future).
We are looking for high accurate estimation(<10mV). Till now, we used System identification tool box to estimate the parameter for all the scenarios as per the breakups mentioned above. But, while validation with drive pattern data(validation data) we get huge error. We are thinking to try Parameter estimation toolbox to get better results.
My query is:
1) is it possible to directly feed the validation data to parameter identification toolbox for estimation? or its always necessary to provide the pulse data for estimation?
2) Intent is we do not have all the pulse data with different SOC and Ibat.
Great Job (Lithium Battery Model, Simscape Language and Simulink Design Optimization)
Now we have an easy way to parameterize battery with accuracy on all conditions.
One question please:
1) During of Model Advisor usage we got
the following error
On Modeling Standards for DO-178C/DO-331 - Check safety-related optimization settings and
On Modeling Standards for DO-178C/DO-331 - Check safety-related code generation settings
Abnormal exit: BLAS loading error: mkl.dll: Not enough storage is available to process this command.
What could I do in order to over come this problem?