Xin Sui
@energy.aau.dk
Department of Energy, The Faculty of Engineering and Science
Aalborg University
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Yusheng Zheng, Yunhong Che, Xiaosong Hu, Xin Sui, and Remus Teodorescu
Institute of Electrical and Electronics Engineers (IEEE)
Xin Sui, Shan He, and Remus Teodorescu
Institute of Electrical and Electronics Engineers (IEEE)
Yusheng Zheng, Yunhong Che, Xin Sui, and Remus Teodorescu
IEEE
Temperature significantly impacts the safety, performance, and degradation of lithium-ion batteries (LIBs), and therefore should be monitored properly by the battery management system (BMS). Hybrid estimation methods by combining physics-based thermal models and machine learning (ML) algorithms, become very promising for sensorless temperature estimation given the limited number of onboard temperature sensors. In this hybrid estimation framework, the physics-based thermal model provides prior knowledge for the ML algorithm to help achieve an accurate final estimation. Therefore, the impact of model accuracy on the overall estimation performance needs to be investigated comprehensively. To this end, this paper investigated the performance of the hybrid estimation framework under different model accuracies, which stem from parameter uncertainties and unmodeled dynamics. Results suggest that the hybrid estimation model can still achieve high accuracy even though trained with inaccurate prior knowledge, demonstrating its robustness to different uncertainties.
X. Sui, Y. Che, Y. Zheng, N. André Weinreich, S. He, and R. Teodorescu
IEEE
In battery management systems (BMSs), state estimation stands as a pivotal element yet encounters significant challenges. These include the poor observability inherent in fixed configuration battery packs, limited generalizability of pre-trained machine learning models, and the deficiency of higher-level management strategies. To address these obstacles, we propose a forward-looking perspective on the future BMS state estimation, introducing the concept of a "Smart Battery". Battery digital twin enables synthetic data generation and physics-informed AI development. This approach integrates battery digital twin to generate synthetic data, which is then used for data augmentation and physics-informed AI development. Additionally, it incorporates advanced data cleaning and selection techniques to preserve essential information and augment data management efficiency. Leveraging cutting-edge AI algorithms, such as transfer learning and meta-learning, aims to mitigate issues of model generalization and feature invalidation under various operating conditions. Furthermore, this paper emphasizes the importance of multi-task learning for batteries, enabling comprehensive health assessments. By fully utilizing both short-term estimations and long-term predictions, the proposed framework contributes to the advancement of higher-level health and thermal management designs. We aim to furnish pioneering insights for state estimation in future intelligent BMSs.
Yusheng Zheng, Yunhong Che, Jia Guo, Nicolai André Weinreich, Abhijit Kulkarni, Ahsan Nadeem, Xin Sui, and Remus Teodorescu
Institute of Electrical and Electronics Engineers (IEEE)
Daniel-Ioan Stroe and Xin Sui
Elsevier
Yusheng Zheng, Yunhong Che, Xiaosong Hu, Xin Sui, and Remus Teodorescu
Institute of Electrical and Electronics Engineers (IEEE)
Yusheng Zheng, Yunhong Che, Xiaosong Hu, Xin Sui, Daniel-Ioan Stroe, and Remus Teodorescu
Elsevier BV
Yunhong Che, Yusheng Zheng, Florent Evariste Forest, Xin Sui, Xiaosong Hu, and Remus Teodorescu
Elsevier BV
Yunhong Che, Yusheng Zheng, Xin Sui, and Remus Teodorescu
Elsevier BV
Yunhong Che, Xin Sui, and Remus Teodorescu
Elsevier BV
Yunhong Che, Søren Byg Vilsen, Jinhao Meng, Xin Sui, and Remus Teodorescu
Elsevier BV
Y. Zheng, N. A. Weinreich, A. Kulkarni, X. Sui, and R. Teodorescu
Institution of Engineering and Technology
A. Kulkarni, R. Teodorescu, X. Sui, and A. Oshnoei
Institution of Engineering and Technology
Xin Sui, Shan He, Yusheng Zheng, Yunhong Che, and Remus Teodorescu
IEEE
Artificial intelligence (AI) has been widely studied for batteries remaining useful lifetime prediction. However, the requirement of big datasets to train a robust AI model limits its practical application, particularly when batteries exhibit diverse degradation behaviors under different working conditions. Collecting sufficient data through laboratory testing can take several years. To tackle these challenges, a few-shot learning-based method for battery early lifetime prediction is proposed where only 6 cycles of charging data are required. The proposed method models batteries with different lengths of cycle life separately, considering that aging features recognized from early cycles might be different for long-life and short-life batteries. First, an auto encoder is trained to group batteries into long-life and short-life classes. The prototypical networks algorithm is employed to learn a metric space where samples from the same class are brought closer together than samples from different classes. Then based on the classification result, different lifetime models are selected, resulting in the final prediction. Few-shot learning technique is utilized to enable accurate and early health assessment of lithium-ion batteries. Compared to building a single model for all batteries throughout their lifetimes, the proposed method reduces the required data size, simplifies AI modeling, and improves prediction accuracy. Finally, the effectiveness of the proposed framework is verified using the accelerated aging dataset from 124 batteries.
Siyu Jin, Xinming Yu, Xin Sui, Wendi Guo, Maitane Berecibar, and Daniel-Ioan Stroe
IEEE
Pulse charging is recognized as a charging technique for maximizing the life of lithium-ion batteries. In this paper, 10 features are extracted from the battery PC operations for battery state of health prediction. By permuting, combining and comparing features, the prediction performance is improved when using two features as input.
Xin Sui, Shan He, and Remus Teodorescu
IEEE
Machine learning (ML) becomes an important technology in battery health assessment. The mapping from feature usually extracted from charging voltage or temperature to unmeasurable state of health (SOH) can be found by training a ML-based SOH estimator. However, the feature may become invalid when operation conditions change or be inaccessible from incomplete charging. For tackling these challenges, various entropies are investigated thoughtfully. Afterwards, spectral entropy and its variants, i.e., composite multi-scale entropy and hierarchical entropy are screened out. Ultrafast SOH feature extraction is therefore achieved where only 2 seconds of voltage data is needed. Finally, the effectiveness of the proposed method is verified by using the accelerated aging dataset from NMC batteries.
Yusheng Zheng, Nicolai André Weinreich, Abhijit Kulkarni, Yunhong Che, Hoda Sorouri, Xin Sui, and Remus Teodorescu
IEEE
Temperature plays a significant role in the safety, performance, and lifetime of lithium-ion batteries (LIBs). Therefore, monitoring battery temperature becomes one of the fundamental tasks for the safe and efficient operation of LIBs. Given the limited onboard temperature sensors, this paper proposes a sensorless temperature estimation method suitable for the smart battery system by obtaining the electrochemical impedance of batteries online via bypass actions. A suitable frequency is selected from the battery electrochemical impedance spectroscopy (EIS) to achieve an accurate and robust estimation of the battery temperature through online impedance measurement. Using the battery impedance with this selected frequency, the battery temperature can be estimated under different scenarios, with RMSE less than 1.5 ℃.
Abhijit Kulkarni, Hoda Sorouri, Yusheng Zheng, Xin Sui, Arman Oshnoei, Nicolai André Weinreich, and Remus Teodorescu
IEEE
The battery digital twin (BDT) is a modern tool that will be used in future intelligent battery management systems (BMS) for Li-Ion batteries (LIB) due to the transition of current technology toward Smart Battery (SB) with information and power processing capability at cell level. The BDT can predict the voltage output based on an impedance model at a given temperature and aging condition and this information can be used for advanced state estimation including sensorless state of temperature (SoT), state of health (SoH) and health management. This paper proposes an online impedance estimation method suitable for the smart battery system which includes a bypass device that can be switched to excite the battery impedance with different frequencies and minimum impact on the load. The performance of the proposed impedance model used in the BDT is compared experimentally in terms of accuracy of the voltage response to dynamic current profiles.
Xin Sui, Shan He, and Remus Teodorescu
IEEE
Using ensemble learning (EL) for battery state of health estimation has become a research hotspot. Because the performance of a single estimator can get boosted, which is applicable in the field of the battery especially when the amount of aging data is insufficient. Traditional EL is to aggregate base models through averaging, which will introduce errors from poor base models. To fully use the estimation results from base models, a statical post-processing method is proposed in this paper. The EL algorithm is initially constructed by combining random sampling and training multiple extreme learning machines. Then the post-processing is performed by fitting the kernel probability distribution of all sub-outputs and determining the most likely estimate, i.e., the statistical mode. As for comparison, the performance of other aggregations using average, weighted average, and mode from a normal distribution are investigated. Finally, the effectiveness of the proposed method is verified by conducting aging experiments on an NMC battery. The root-mean-squared error is as low as 0.2%, which is an approximate 80% improvement in accuracy over the traditional average-based method. The proposed method tackles the unstable estimation in learning with a small dataset, which is suitable for practical applications.
Yunhong Che, Daniel-Ioan Stroe, Xin Sui, Sϕren Byg Vilsen, Xiaosong Hu, and Remus Teodorescu
IEEE
Studying and analyzing battery aging behavior is crucial in battery health prognostic and management. This paper conducts novel and comprehensive experiments to evaluate battery aging under variable external stresses, including different dynamic load profiles and variable environmental temperatures. Respond analysis in the time and frequency domain is performed to account for the different aging rates under different current loadings, where the statistic calculation and fast Fourier transform are used for the analysis. The empirical model is used to fit the fade curve for the comparisons between constant and variable temperatures. The capacity decrease and internal resistance increase are extracted to evaluate capacity and power fade, respectively. The experimental results show that the urban dynamic operating conditions help to prolong the service life compared to the constant current aging case. In contrast, the aging under the highway profile accelerates the aging process. Although the average temperature is the same as under constant temperature conditions, variable temperature conditions accelerate battery aging.
Xin Sui, Shan He, SØren Byg Vilsen, Remus Teodorescu, and Daniel-Ioan Stroe
IEEE
Artificial neural networks are widely studied for the state of health (SOH) estimation of Lithium-ion batteries because they can recognize global features from the raw data and are able to cope with multi-dimensional data. But the performance of the model depends to some extent on the selection of the hyperparameters, which remain constant during model training. To improve the generalization performance as well as accuracy, an ensemble learning framework is proposed for battery SOH estimation, where multiple extreme learning machines are trained combined with bagging technology. The numbers of bags and neurons of the base model are then tuned by five commonly used hyperparameter optimization methods. Moreover, the SOH value with maximum probability density is selected as the output estimate to further improve the estimation accuracy. Finally, experimental results on both NMC and LPF batteries demonstrate that the proposed method with hyperparameter optimization can achieve stable and accurate battery SOH estimation. Regardless of which optimization method is used, the average percentage error for SOH estimation of NMC and LFP batteries can keep below 1% and 1.2%, respectively.
Jia Guo, Siyu Jin, Xin Sui, Xinrong Huang, Yaolin Xu, Yaqi Li, Peter Kjær Kristensen, Deyong Wang, Kjeld Pedersen, Leonid Gurevich,et al.
Royal Society of Chemistry (RSC)
Constant current charging and discharging is widely used nowadays for commercial lithium (Li) ion batteries (LIBs) in applications of portable electronic devices and electric vehicles.
Jichang Peng, Jinhao Meng, Dan Chen, Haitao Liu, Sipeng Hao, Xin Sui, and Xinghao Du
MDPI AG
With the widespread use of Lithium-ion (Li-ion) batteries in Electric Vehicles (EVs), Hybrid EVs and Renewable Energy Systems (RESs), much attention has been given to Battery Management System (BMSs). By monitoring the terminal voltage, current and temperature, BMS can evaluate the status of the Li-ion batteries and manage the operation of cells in a battery pack, which is fundamental for the high efficiency operation of EVs and smart grids. Battery capacity estimation is one of the key functions in the BMS, and battery capacity indicates the maximum storage capability of a battery which is essential for the battery State-of-Charge (SOC) estimation and lifespan management. This paper mainly focusses on a review of capacity estimation methods for BMS in EVs and RES and provides practical and feasible advice for capacity estimation with onboard BMSs. In this work, the mechanisms of Li-ion batteries capacity degradation are analyzed first, and then the recent processes for capacity estimation in BMSs are reviewed, including the direct measurement method, analysis-based method, SOC-based method and data-driven method. After a comprehensive review and comparison, the future prospective of onboard capacity estimation is also discussed. This paper aims to help design and choose a suitable capacity estimation method for BMS application, which can benefit the lifespan management of Li-ion batteries in EVs and RESs.
Remus Teodorescu, Xin Sui, Søren B. Vilsen, Pallavi Bharadwaj, Abhijit Kulkarni, and Daniel-Ioan Stroe
MDPI AG
Applications of lithium-ion batteries are widespread, ranging from electric vehicles to energy storage systems. In spite of nearly meeting the target in terms of energy density and cost, enhanced safety, lifetime, and second-life applications, there remain challenges. As a result of the difference between the electric characteristics of the cells, the degradation process is accelerated for battery packs containing many cells. The development of new generation battery solutions for transportation and grid storage with improved performance is the goal of this paper, which introduces the novel concept of Smart Battery that brings together batteries with advanced power electronics and artificial intelligence (AI). The key feature is a bypass device attached to each cell that can insert relaxation time to individual cell operation with minimal effect on the load. An advanced AI-based performance optimizer is trained to recognize early signs of accelerated degradation modes and to decide upon the optimal insertion of relaxation time. The resulting pulsed current operation has been proven to extend lifetime by up to 80% in laboratory aging conditions. The Smart Battery unique architecture uses a digital twin to accelerate the training of performance optimizers and predict failures. The Smart Battery technology is a new technology currently at the proof-of-concept stage.