Experimental Data-Driven Parameter Identification and State of Charge Estimation for a Li-Ion Battery Equivalent Circuit Model

Paper at a glance

What the researchers studied

The authors fit a second-order RC battery model to experimental data using a genetic algorithm, compare simulated terminal voltage with measurements under the same current input, and use an adaptive extended Kalman filter for state-of-charge estimation. Section 2 defines the circuit; Section 3 describes its identification and experimental comparison. Read the publisher’s article.

Connection to Lemonergy

Use the two-branch battery pulse example to inspect how the instantaneous resistive voltage change differs from the two slower RC responses. The parameter recovery benchmark then demonstrates fitting selected parameters to a known waveform.

These are different evidence questions: a correct forward calculation, recovery from synthetic data, and agreement with a physical experiment. The paper is related reading for the third question; Lemonergy’s supplied examples do not reproduce its experiments, optimizer or state estimator.

Apply the reading to your model

1. Identify which parameters you will estimate and which remain fixed.
2. Save the current profile, voltage samples, units and sampling interval with the model.
3. Compare the complete voltage trace and residuals using the results guide.
4. Record the origin of the comparison data. A generated waveform must remain distinguishable from a measurement.

Continue learning

• Run, edit and save a validation model
• Battery pulse response with two RC branches
• Parameter recovery from an authored pulse waveform