Data Driven Modelling of Li-ion Cell Batteries

Repository for project titled Data-driven modelling of Li-ion Batteries.

Requirements

• Python>=3.8
• Numpy
• Scipy
• Matplotlib
• Pandas

Usage

• main.py for execution in IDE
• main.ipynb for execution as a notebook

Overview

This project is an implementation of RC link modelling of Li-ion Batteries using convex optimization to fit desired parameters.

Open Source Datasets Used

LG 18650HG2

• Kollmeyer, Phillip; Vidal, Carlos; Naguib Mina; Skells, Michael (2020), “LG 18650HG2 Li-ion Battery Data and Example Deep Neural Network xEV SOC Estimator Script”, Mendeley Data, V3, doi: 10.17632/cp3473x7xv.3

Battery Models Used

RC2 model:

• 

RC2 model with hysteresis:

• 

OCV-SOC Curve Extraction

• The OCV-SOC curve is computed as the averge of the charge and discharge OCV-SOC curves extracted from OCV test data.
• 
• 
• The computed OCV-SOC curve is the stored for simulation of cell using known cell parameters.
• 

Model Parameter Extraction

• The cell parameters are extracted from the dynamic test data using the minimize function from scipy.optimize.
• The Cumulative Root Mean Squared Error (CRMSE) between the simulated terminal voltage and actual terminal voltage for a given dynamic test is considered as the loss function to be minimized.
• The bounds of the cell parameters are defined with plausible values for the resistances and time constants of each RC branch.
• Once the parameters are trained from a training data, it is then validated on different dynamic test data by comparing the computed CRMSEs.

Results

RC2 model:

OCV-SOC curve extracted from 25degC/549_C20DisCh.csv, training done on 25degC/551_Mixed1.csv and validation done on 25degC/551_LA92.csv:

Training:

• 

Validation:

• 

Parameter	R0(ohm)	R1(ohm)	R2(ohm)	C1(farad)	C2(farad)	Training CRMSE(mV)	Validation CRMSE(mV)
Value	0.01951358	0.01541913	0.5	1395.97247796	84959.64540431	20.01416826514894	20.095525362647045

RC2 model with hysteresis: TBD

License

MIT License