Article
Energy & Fuels
Jinpeng Tian, Rui Xiong, Weixiang Shen, Jiahuan Lu
Summary: A method based on deep neural network is proposed for fast and accurate estimation of SOC for LiFePO4 batteries, with an error of less than 2.03% over the entire battery SOC range. By integrating the DNN with a Kalman filter, the robustness of SOC estimation against random noises and error spikes can be improved.
Article
Thermodynamics
Xiong Feng, Junxiong Chen, Zhongwei Zhang, Shuwen Miao, Qiao Zhu
Summary: This paper presents a novel neural network structure called CWRNN, which effectively addresses long-term dependencies, reduces training and computation costs, and is validated under different temperature conditions.
Article
Energy & Fuels
Zhimin Xi, Rui Wang, Yuhong Fu, Chris Mi
Summary: This paper proposes a time-delayed recurrent neural network for lithium ion battery modeling and SOC estimation, and analyzes its performance. It is found that overexcited neurons may cause poor performance of the neural network.
Article
Energy & Fuels
Elias Dias Rossi Lopes, Marlon Marques Soudre, Carlos Humberto Llanos, Helon Vicente Hultmann
Summary: An important component of electric vehicles is the Battery Management System (BMS), which is responsible for monitoring the state of charge (SoC) of the battery, a crucial factor in the vehicle's autonomy. This paper proposes the use of receding-horizon strategies, specifically Moving-Horizon State Estimation (MHSE) and Neural Network Moving-Horizon Estimation (NNMHE), for accurate SoC estimation. MHSE utilizes a constrained optimization problem with a larger observation window, while NNMHE employs a neural network to emulate the optimization solver, resulting in faster and approximate results. The effectiveness of this approach is validated with experimental data, achieving a coefficient of determination of almost 99% and a processing time reduction of about 20 times, making it suitable for embedded systems with limited computational resources.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Energy & Fuels
Qiao Wang, Min Ye, Meng Wei, Gaoqi Lian
Summary: In this paper, an optimized deep neural network-based low-cost SOC estimation method is proposed. By optimizing the moving window and adjusting the hyperparameters, the performance of the deep neural network is improved. The proposed method is validated using experimental data and compared to other estimators to prove its superiority.
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
(2022)
Article
Energy & Fuels
Ivo Horstkoetter, Philipp Gesner, Kerstin Hadler, Bernard Baeker
Summary: Understanding the degradation processes of lithium-ion cells is a current and pressing challenge, influenced by various operating conditions. Experimentation has shown that the discharge dynamics of a load profile significantly impact battery degradation, with higher current gradients resulting in larger degradation rates. This linear relationship between current gradient and degradation rate highlights the importance of considering dynamic influences in battery aging studies.
JOURNAL OF ENERGY STORAGE
(2021)
Article
Electrochemistry
Jinpeng Tian, Rui Xiong, Cheng Chen, Chenxu Wang, Weixiang Shen, Fengchun Sun
Summary: In this study, a data-driven solution is proposed to predict battery impedance spectra at different states. An encoder-decoder deep neural network is developed to simultaneously predict impedance spectra and state of charge using short-term pulse data sampled at 1 Hz, eliminating the need for specific hardware and reducing test requirements. Validation results show that the proposed method enables accurate predictions at different temperatures and ageing levels, with errors of impedance spectra and SOC restricted within 1.5 mΩ and 1.26%, respectively. Additionally, the predicted impedance spectra provide detailed physical insight into battery kinetics and allow for accurate extraction of critical parameters of an impedance model. This method makes EIS measurement more accessible for evaluating battery characteristics and demonstrates the potential of deep learning in battery research.
ELECTROCHIMICA ACTA
(2023)
Article
Thermodynamics
Shanshan Guo, Liang Ma
Summary: This study investigates the performance of four state-of-the-art deep learning algorithms in state-of-charge estimation, evaluating their accuracy, robustness, and efficiency using experimental data.
Article
Chemistry, Analytical
Yu-Chun Wang, Nei-Chun Shao, Guan-Wen Chen, Wei-Shen Hsu, Shun-Chi Wu
Summary: This study applies deep-learning techniques, specifically convolutional residual networks, to estimate the state-of-charge (SOC) of lithium-ion batteries. By stacking values of multiple measurable variables as inputs, process information for voltage or current generation can be effectively extracted, allowing accurate SOC regression.
Article
Green & Sustainable Science & Technology
Wei Xiong, Fang Xie, Gang Xu, Yumei Li, Ben Li, Yimin Mo, Fei Ma, Keke Wei
Summary: As electric vehicles become more common, there is increasing concern about the disposal of electric vehicle batteries. This study focuses on the reuse of retired batteries and the importance of accurate state of charge estimation. The study develops a battery model dependent on temperature and aging state and proposes a co-estimation method for parameter identification and state estimation. The proposed method is validated and found to be highly robust, with an error in state of charge estimation restricted to 1% over a wide temperature range and battery degradation range.
Article
Thermodynamics
Yong Zhou, Guangzhong Dong, Qianqian Tan, Xueyuan Han, Chunlin Chen, Jingwen Wei
Summary: Due to the complex behaviors of lithium-ion batteries, accurately estimating their state-of-health remains a critical challenge. Most existing battery health prognosis methods focus on low-frequency sampled time-domain response, which may not fully reflect the battery health status in automotive applications. This paper proposes a data-driven method using high and medium frequency impedance spectroscopy data to estimate battery state-of-health. Experimental results demonstrate the high accuracy and robustness of the proposed method, with an estimation error of 1.12%.
Article
Energy & Fuels
Jiangnan Hong, Yucheng Chen, Qinqin Chai, Qiongbin Lin, Wu Wang
Summary: This paper proposes a SOH estimation method for lithium-ion batteries based on a recurrent neural network and feature extraction, which is validated on the NASA dataset and shows high accuracy.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Thermodynamics
Junxiong Chen, Xiong Feng, Lin Jiang, Qiao Zhu
Summary: A novel method combining a denoising autoencoder neural network with a gated recurrent unit recurrent neural network is proposed to reduce the impact of measurement data noise on state of charge estimation. Experimental results demonstrate that the proposed method has better accuracy and robustness in SOC estimation.
Article
Thermodynamics
Zichuan Ni, Xianchao Xiu, Ying Yang
Summary: This paper presents a novel method based on canonical correlation analysis (CCA) for battery state of charge (SOC) estimation. The proposed method achieves high accuracy under input noise and is computationally efficient.
Article
Chemistry, Physical
Jorn M. Reniers, Grietus Mulder, David A. Howey
Summary: Lithium-ion batteries are increasingly used in liberalized electricity systems driven by economic optimization. A physics-based degradation model can decrease battery degradation and increase revenue. The approach increases battery lifetime in terms of years and cycles, while also improving revenue potential.
JOURNAL OF POWER SOURCES
(2021)