A Lithium-Ion Battery Fractional Order State Space Model and its Time Domain System Identification

Hongjie Wu,Shifei Yuan,Chengliang Yin

DOI: https://doi.org/10.1007/978-3-642-33741-3_8

2012-11-07

Abstract:This paper deals with a fractional order state space model for the lithium-ion battery and its time domain system identification method. Currently the equivalent circuit models are the most popular model which was frequently used to simulate the performance of the battery. But as we know, the equivalent circuit model is based on the integer differential equations, and the accuracy is limited. And the real processes are usually of fractional order as opposed to the ideal integral order models. So here we propose a lithium-ion battery fractional order state space model, and compare it with the equivalent circuit models, to see which model fit with the experiment results best. Then the hybrid pulse power characterization (HPPC) test has been implemented in the lithium-ion battery during varied state-of-charge (SOC). Based on the Levenberg–Marquardt algorithm, the parameters for each model have been obtained using the time-domain test data. Experimental results show that the proposed lithium-ion fractional order state space model has a better fitness than the classical equivalent circuit models. Meanwhile, five other cycles are adopt here to validate the prediction error of the two models, and final results indicate that the fractional model has better generalization ability.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the accuracy problem of lithium - ion battery modeling. Specifically, although the existing Equivalent Circuit Model (ECM) is widely used to simulate the performance of lithium - ion batteries, it is based on integer - order differential equations, resulting in limited model accuracy. However, the actual process usually exhibits fractional - order characteristics rather than the ideal integer - order model. Therefore, the paper proposes a Fractional Order State Space Model for lithium - ion batteries and verifies the validity of this model through the time - domain system identification method, aiming to improve the accuracy of the model in describing the dynamic performance of the battery.

The paper solves this problem through the following steps:

1. **Model construction**:
- Constructed integer - order models (such as the first - order equivalent circuit model ECM - 1) and fractional - order models (such as FM - 1).
- The transfer function form of the fractional - order model is:
\[
G(s,\alpha)=\frac{a_{2}s^{1}+a_{1}s^{0.5}+a_{0}}{1 + b_{2}s^{1}+b_{1}s^{0.5}}
\]
- The transfer function form of the integer - order model is:
\[
G(s)=\frac{R_{1}sR_{0}R_{1}C_{1}+R_{0}+R_{1}}{sC_{1}R_{1}+1}
\]

2. **Experimental design**:
- Conducted an Open Circuit Voltage Test to obtain open - circuit voltage data at different states of charge (SOC).
- Implemented a Hybrid Pulse Power Characterization (HPPC) test to obtain dynamic characteristic data of the battery at different SOCs.

3. **Parameter identification**:
- Used the Levenberg - Marquardt algorithm to identify the parameters of each model based on the time - domain test data.
- Optimized the parameter estimates by minimizing the sum of squared prediction errors (SSE).

4. **Model comparison and verification**:
- Compared the proposed fractional - order model with the classical integer - order equivalent circuit model. The results show that the fractional - order model has better fitting and generalization abilities.
- The verification results show that the fractional - order model has a lower average error in multiple working cycles (such as Dynamic Stress Test DST, Federal Urban Driving Schedule FUDS, etc.), showing higher accuracy.

Through these steps, the paper successfully demonstrates the superiority of the fractional - order model in lithium - ion battery modeling and provides more accurate model support for battery management systems.