Combined State of Charge and State of Energy Estimation for Echelon-Use Lithium-Ion Battery Based on Adaptive Extended Kalman Filter

Enguang Hou, Zhen Wang, Xiaopeng Zhang, Zhixue Wang, Xin Qiao, Yun ZhangBatteries 2023 · DOI

TL;DR

This paper introduces a novel method for simultaneously estimating the State of Charge (SOC) and State of Energy (SOE) of echelon-use lithium-ion batteries (EULIBs). By combining a third-order equivalent circuit model with Long Short-Term Memory (LSTM) for parameter optimization and an Adaptive Extended Kalman Filter (AEKF) for robust estimation, the approach achieves high accuracy (error < 1.19%) and strong adaptability, even with uncertain initial conditions.

What Problem Does This Paper Address?

Echelon-use lithium-ion batteries (EULIBs) are retired power batteries repurposed for secondary applications. While they still possess significant capacity, their performance degrades, and their characteristics become less consistent. Accurately assessing their remaining useful life and operational safety is crucial, and this relies heavily on knowing their State of Charge (SOC) and State of Energy (SOE). Traditional methods struggle with the degraded and inconsistent nature of EULIBs, often requiring precise initial values and lacking adaptability to changing conditions. This paper aims to bridge this gap by developing a robust and accurate estimation method for both SOC and SOE in EULIBs, even when their initial states are unknown and their performance has degraded.

Methods

The proposed method integrates several key components:

  • Third-Order Resistor-Capacitance Equivalent Model (TRCEM): This model is chosen to capture the complex electrochemical dynamics of EULIBs more accurately than simpler models. It includes multiple RC branches to represent different polarization effects.
  • Long Short-Term Memory (LSTM) for Parameter Optimization: LSTM, a type of recurrent neural network, is employed to dynamically optimize crucial battery parameters like Ohmic Internal Resistance (OIR), Actual Energy (AE), and Actual Capacity (AC) in real-time. This addresses the challenge of parameter drift due to battery degradation.
  • Adaptive Extended Kalman Filter (AEKF): The core estimation framework is an AEKF. This filter is designed to adaptively update its process noise and observation noise covariance matrices. This adaptive capability allows the filter to adjust to changing battery conditions and uncertainties, enhancing its robustness and accuracy, especially when initial SOC/SOE values are not precisely known.

The overall approach involves establishing the TRCEM, using LSTM to refine its parameters, and then feeding these optimized parameters into the AEKF for the combined SOC and SOE estimation.

Datasets

The paper utilizes experimental data from a lithium-ion battery undergoing charge and discharge cycles at room temperature. The experiments were conducted using a BTS20 testing system. The battery's degradation was simulated by selecting data at different capacity decay levels: 90%, 60%, and 30% of its rated capacity. This dataset serves as the basis for validating the proposed estimation method under various degradation scenarios.

Key Insights and Ideas

  • The Synergy of LSTM and AEKF: The paper highlights that combining LSTM's ability to learn complex temporal dependencies for parameter estimation with the AEKF's adaptive filtering capabilities creates a powerful tool for dynamic battery state estimation. LSTM handles the non-linear parameter variations, while AEKF manages the state estimation under uncertainty.
  • Adaptability is Key for EULIBs: The research strongly emphasizes that for echelon-use batteries, adaptability is as crucial as accuracy. The AEKF's ability to adjust noise covariances allows it to effectively track the true state even when initial estimates are significantly off, a common scenario with aged batteries.
  • TRCEM Provides Necessary Detail: The choice of a third-order model (TRCEM) is justified by its ability to better represent the electrochemical processes, leading to more accurate estimations compared to simpler models, especially under varying operational conditions.
  • Initial Value Uncertainty is Handled: A significant finding is that the proposed method demonstrates strong adaptive ability and can accurately track the actual SOC and SOE values, even when the initial estimates are set to 100%, 60%, or 20%, showcasing its resilience to initial condition errors.

Key Results

  • High Accuracy: The combined LSTM-AEKF method achieved an estimation error of less than 1.19% for both SOC and SOE across various degradation levels (90%, 60%, and 30% capacity).
  • Robust Adaptability: The method demonstrated strong adaptive capabilities, effectively tracking actual values despite significant differences between initial and true SOC/SOE values.
  • Performance Across Degradation Levels: The accuracy remained high even as the battery degraded significantly, indicating the method's suitability for the entire echelon-use lifecycle. For instance, when decayed to 30% capacity, the SOE error ranged from -1.09% to 1.16%, and the SOC error ranged from -1.06% to 1.19%.
  • Comparison to Existing Methods: The paper presents a comparison (Table 3) showing its accuracy (1.19%) surpasses several other methods, such as those achieving 2.34%, 2%, and 1.93% errors, while maintaining adaptability.

Limitations and Future Work

  • Computational Cost: While not explicitly detailed as a limitation, the combination of LSTM and AEKF might introduce higher computational demands compared to simpler methods. Future work could focus on optimizing this for real-time embedded systems.
  • Generalizability: The study was conducted on a specific type of lithium-ion battery. Further research could explore the method's performance on different battery chemistries and configurations.
  • Real-World Deployment: While simulations are promising, real-world testing under diverse operating conditions (temperature variations, different charging/discharging profiles) would be valuable for full validation.
  • Parameter Identification: Although the paper mentions parameter identification methods, further exploration into more efficient and robust online parameter identification techniques could enhance the overall system.

Summary updated

Abstract

To ensure the safety and reliability of an echelon-use lithium-ion battery (EULIB), the performance of a EULIB is accurately reflected. This paper presents a method of estimating the combined state of energy (SOE) and state of charge (SOC). First, aiming to improve the accuracy of the SOE and SOC estimation, a third-order resistor-capacitance equivalent model (TRCEM) of a EULIB is established. Second, long short-term memory (LSTM) is introduced to optimize the Ohmic internal resistance (OIR), actual energy (AE), and actual capacity (AC) parameters in real time to improve the accuracy of the model. Third, in the process of the SOE and SOC estimation, the observation noise equation and process noise equation are updated iteratively to make adaptive corrections and enhance the adaptive ability. Finally, an SOE and SOC estimation method based on LSTM optimization and an adaptive extended Kalman filter (AEKF) is established. In simulation experiments, when the capacity decays to 90%, 60% and 30% of the rated capacity, regardless of whether the initial value is consistent with the actual value, the values of the SOE and SOC estimation can track the actual value with strong adaptive ability, and the estimated error is less than 1.19%, indicating that the algorithm has a high level of accuracy. The method presented in this paper provides a new perspective for estimating the SOE and SOC of a EULIB.
BibTeX
@inproceedings{hou2023combined,
  author = {Enguang Hou and Zhen Wang and Xiaopeng Zhang and Zhixue Wang and Xin Qiao and Yun Zhang and Xin Qiao and Yun Zhang},
  title = {{Combined State of Charge and State of Energy Estimation for Echelon-Use Lithium-Ion Battery Based on Adaptive Extended Kalman Filter}},
  booktitle = {Batteries},
  year = {2023},
  doi = {10.3390/batteries9070362},
}
Figure