Publications

Degradation capacity spatial-temporal embedding RUL prediction framework for lithium-ion batteries

Published in IEEE Journal of Emerging and Selected Topics in Industrial Electronics, April 16, 2025

Accurately predicting the remaining useful life (RUL) of lithium-ion batteries is crucial to ensure the safe and reliable operation of the energy storage and power supply systems. However, RUL prediction is significantly affected by the challenges posed by multi-dimensional nonlinearity. Embedding spatial-temporal variables helps reveal nonlinear relationships, making the degradation capacity spatial-temporal embedding an effective approach for extracting patterns and trends of battery degradation within multi-dimensional subspaces. Motivated by this, this work proposes a phase space reconstruction (PSR) approach that utilizes the C-C method combined with a convolutional neural network-bidirectional LSTM and hyperspace attention mechanism (CNN-BiLSTM-HAM) to address this challenge. First, the C-C method performs PSR according to time delay and embedding dimension values to transform degradation capacity data from one-dimensional time series into a multi-dimensional format. Next, the reconstructed capacity matrix is processed by the CNN to generate feature vectors that preserve the spatial structure and location information of the input data, while discarding irrelevant information. These feature vectors serve as input for training the BiLSTM. Finally, the HAM is used to allocate the weights of different feature subsets simultaneously. The proposed CNN-BiLSTM-HAM model and C-C method are validated using the NASA dataset. Experimental results demonstrate that the proposed method yields accurate RUL prediction, with the absolute error, mean absolute error, and root mean square error all being less than 2, 1.3%, and 2%, respectively.

Smart electric vehicle charging algorithm to reduce the impact on power grids: A reinforcement learning based methodology

Published in IEEE Open Journal of Vehicular Technology, April 9, 2025

The increasing penetration of electric vehicles (EVs) presents a significant challenge for power grid management, particularly in maintaining network stability and optimizing energy costs. Existing model predictive control (MPC)-based approaches for EV charging and discharging scheduling often struggle to balance computational efficiency with real-time operationability. This gap highlights the need for more advanced methods that can effectively mitigate the impact of EV activities on power grids without oversimplifying system dynamics. Here, we propose a novel scheduling methodology using a pre-trained Reinforcement Learning (RL) framework to address this challenge. The method integrates real grid simulations to monitor critical electrical points and variables while simplifying analysis by excluding the influence of real grid dynamics. The proposed approach formulates the scheduling problem to minimize costs, maximize rewards from ancillary service delivery, and mitigate network overloads at specified grid nodes. The methodology is validated on a benchmark electric grid, where realistic charging station utilization scenarios are simulated. The results demonstrate the method’s robustness and ability to efficiently cope with the EV smart scheduling problem.

A stable lithium-ion battery SOH estimation framework for suppressing measurement noise with unknown distribution

Published in IEEE Transactions on Transportation Electrification, March 26, 2025

Existing methods for estimating the state of health (SOH) of lithium-ion battery (LIB) typically rely on the assumption that the distribution of noise (or outliers) in the measurement data is known. However, this assumption rarely holds true for LIB operating under real-word conditions. This article proposes a stable framework for accurate SOH estimation that accommodates noises with unknown distribution in both measurement data and label values. The framework combines generalized correntropy loss (GCL) with Savitzky-Golay (SG) filter and extreme learning machine (ELM) to obtain measurement data filter named SG-GCL and SOH estimator named generalized ELM (GELM), respectively. The SG-GCL filtering of the measurement data keeps the root mean square error (RMSE) within 0.0365%, and Pearson correlation between extracted feature and SOH improves by 0.4963, which in turn leads to the reduction of the RMSE metrics of the ELM for the estimation of the SOH by 43.69%. From the filtering results, feature extraction and estimation results proved its necessity and effectiveness. GELM effectively suppresses the influence of label value noise on the model in the training process, which reduces the SOH estimation RMSE index by more than 0.66%. The results from experiments with different distributional noise conditions show that the proposed SOH estimation framework has excellent and stable performance.

A unified model for active battery equalization systems

Published in IEEE Transactions on Control Systems Technology, November 20, 2024

Lithium-ion battery packs demand effective active equalization systems to enhance their usable capacity and lifetime. Despite numerous topologies and control schemes proposed in the literature, conducting quantitative analyses, comprehensive comparisons, and systematic optimization of their performance remains challenging due to the absence of a unified mathematical model at the pack level. To address this gap, we introduce a novel, hypergraph-based approach to establish the first unified model for various active battery equalization systems. This model reveals the intrinsic relationship between battery cells and equalizers by representing them as the vertices and hyperedges of hypergraphs, respectively. With the developed model, we identify the necessary conditions for all equalization systems to achieve balance through controllability analysis, offering valuable insights for selecting the number of equalizers. Moreover, we prove that the battery equalization time is inversely correlated with the second smallest eigenvalue of the hypergraph’s Laplacian matrix of each equalization system. This significantly simplifies the selection and optimized design of equalization systems, obviating the need for extensive experiments or simulations to derive the equalization time. Illustrative results demonstrate the efficiency of the proposed model and validate our findings.

Modeling of PEMEL hydrogen production systems: Comprehensive multivariate sensitivity analysis considering mass-energy dynamic equilibrium

Published in Applied Energy, October 11, 2024

The proton exchange membrane electrolyzer (PEMEL) system offers significant advantages for utilizing renewable energy for hydrogen production, owing to its high efficiency and wide operating range. However, the current research mainly focuses on optimizing the stack and overlooking the modeling of nonlinear behavioral characteristics and sensitivity analysis related to the dynamics of the mass and energy transfer process at the system level. Thus, this paper proposes a method for dynamically modeling and conducting multivariate parameter sensitivity analysis of the PEMEL hydrogen production system, considering the mass-energy equilibrium. Temperature, pressure, water discharge, and other factors are integrated to model the PEMEL hydrogen production system based on the Balance of Plant (BoP). System efficiency, system power, and power consumption per unit of hydrogen production are utilized as performance indicators. To verify the model’s feasibility, the output characteristics were compared with the experimental results from both stack and system levels, yielding a Coefficient of Determination (R-squared) exceeding 97%, which indicates a strong fit between simulation and reality. Subsequently, three sensitivity analysis methods, Taguchi method, analysis of variance (ANOVA), and Pareto analysis, were employed to determine the contribution of different parameters to the key system performance indicators and identify the optimal operating combination. Orthogonal arrays and signal-to-noise ratio (SNR) analysis facilitated this assessment. Results reveal that the operating temperature had the highest contribution to system efficiency and power consumption per unit of hydrogen production, reaching 69.48% and 70.09%, with the standardized effect far exceeding the minimum threshold of 2.069. Pressure holds the highest influence on the system power, at 77.98%. Finally, within the operational feasibility domain of the system, the calculation ensured that power consumption per unit of hydrogen production remains below 4.7 kW·h/m3.

MINN: Learning the dynamics of differential-algebraic equations and application to battery modeling

Published in IEEE Transactions on Pattern Analysis and Machine Intelligence, September 9, 2024

The concept of integrating physics-based and data-driven approaches has become popular for modeling sustainable energy systems. However, the existing literature mainly focuses on the data-driven surrogates generated to replace physics-based models. These models often trade accuracy for speed but lack the generalizability, adaptability, and interpretability inherent in physics-based models, which are often indispensable in modeling real-world dynamic systems for optimization and control purposes. We propose a novel machine learning architecture, termed model-integrated neural networks (MINN), that can learn the physics-based dynamics of general autonomous or non-autonomous systems consisting of partial differential-algebraic equations (PDAEs). The obtained architecture systematically solves an unsettled research problem in control-oriented modeling, i.e., how to obtain optimally simplified models that are physically insightful, numerically accurate, and computationally tractable simultaneously. We apply the proposed neural network architecture to model the electrochemical dynamics of lithium-ion batteries and show that MINN is extremely data-efficient to train while being sufficiently generalizable to previously unseen input data, owing to its underlying physical invariants. The MINN battery model has an accuracy comparable to the first principle-based model in predicting both the system outputs and any locally distributed electrochemical behaviors but achieves two orders of magnitude reduction in the solution time.

Data-driven state of health estimation method of lithium-ion batteries for partial charging curves

Published in IEEE Transactions on Energy Conversion, May 30, 2024

State of health (SOH) is one of the most important performance indicators of lithium-ion batteries (LIBs). Accurate estimation of SOH is a prerequisite for the safe and reliable operation of LIBs. Traditional SOH estimation methods predominantly rely on complete charging cycle data acquired through laboratory testing. However, in practical application, the charging behaviors of electric vehicle users are random and unpredictable, making the partial charging curves difficult to utilize the traditional methods. This work introduces a novel data-driven approach to estimating a battery’s SOH for partial charging cases. Firstly, a curve fitting method is proposed to extract health indicators (HIs) from partial charging voltage data, where novel HIs based on the energy-voltage curve are extracted. A composite Gaussian process regression-based data-driven method is proposed to achieve highly accurate SOH estimation. The method’s adaptability to real-world partial charging habits is evaluated through three representative scenarios derived from extensive charging behavior reports of EV users. The impact of partial charging on HI extraction is analyzed based on the three identified scenarios. The proposed method is verified using a combination of our laboratory testing data and the Oxford open dataset. The results show that the proposed framework demonstrates the ability to estimate SOH accurately and strong robustness to various partial charging behaviors.

Real-time reconfiguration-based all-cell flexibility and capacity maximum utilization of second-life batteries

Published in IEEE Transactions on Transportation Electrification, May 10, 2024

The capacity underutilization caused by cell inconsistency hinders the efficient utilization of lithium-ion battery packs. This is particularly critical for the second-life battery utilization where high cell inconsistency exists. To address this issue, this article proposes a multiscale reconfiguration control method enabled by an efficient reconfigurable battery topology, aiming to maximize the pack’s capacity utilization. In this regard, a novel four-switch reconfigurable battery topology is proposed, offering the advantages of all-cell flexibility and reasonable complexity. Building upon this, an all-cell equalization method is proposed, combining intramodule current sharing and three forms of intermodule energy distribution to achieve maximum pack capacity utilization. Moreover, real-time reconfiguration ensures effective charge transmission when the pack voltage deviates from the expected threshold. A laboratory-scale prototype of the reconfigurable battery pack is tested, and the experimental results confirm that the proposed design and reconfiguration control can improve pack capacity utilization and efficiency by 10.96% and 14.34%, respectively, without any redundant design. This method provides a feasible solution for grouping and system management of second-life battery systems consisting of highly inconsistent cells.

A flow-rate-aware data-driven model of vanadium redox flow battery based on gated recurrent unit neural network

Published in Journal of Energy Storage, November 7, 2023

The vanadium redox flow battery (VRB) system involves complex multi-physical and multi-timescale interactions, where the electrolyte flow rate plays a pivotal role in both static and dynamic performance. Traditionally, fixed flow rates have been employed for operational convenience. However, in today’s highly dynamic energy market environment, adjusting flow rates based on operating conditions can provide significant advantages for improving VRB energy conversion efficiency and cost-effectiveness. Unfortunately, incorporating the electrolyte flow rate into conventional multi-physical models is overly complex for VRB management and control systems, as real-time operations demand low-computational and low-complexity models for onboard functionalities. This paper introduces a novel data-driven approach that integrates flow rates into VRB modeling, enhancing data processing capabilities and prediction accuracy of VRB behaviors. The proposed model adopts a gated recurrent unit (GRU) neural network as its fundamental framework, exhibiting exceptional proficiency in capturing VRB’s nonlinear voltage segments. The GRU network structure is carefully designed to optimize the predictive ability of the model, with flow rate considered as a crucial input parameter to account for its influence on VRB behavior. Model refinement involves analyzing well-designed simulation results obtained during VRB operations under various flow rates. Laboratory experiments were also designed and conducted, covering different conditions of currents and flow rates to validate the proposed data-driven modeling method. Comparative analyses were performed against several state-of-the-art algorithms, including equivalent circuit models and other data-driven models, demonstrating the superiority of the proposed GRU-based VRB model considering flow rates. Thanks to the GRU’s outstanding capability in processing time series data, the proposed model delivers impressively accurate terminal voltage predictions with a low error margin of no more than 0.023 V (1.3%) under wide operating ranges. These results indicate the efficacy and robustness of the proposed approach, highlighting the novelty and significance of accounting for flow rates in accurate VRB modeling for management and control system design.

Hybrid physics-based and data-driven prognostic for PEM fuel cells considering voltage recovery

Published in IEEE Transactions on Energy Conversion, September 4, 2023

Predicting the degradation behaviors is challenging and essential for prognostics and health management for proton exchange membrane fuel cells (PEMFCs). However, existing methods based on data-driven or model-based methods can face the problem of significant performance inconsistencies in different prediction stages. We investigate the cause and attribute it to the ignorance of the voltage recovery phenomena of PEMFCs observed during the frequent start-stop processes during practical applications. A novel prognostic method is proposed to provide a more comprehensive analysis of PEMFC aging that integrates data-driven and model-based methods. Specifically, a physics-based aging model considering voltage recovery (PA-VR) is first reported as a model-based method to enhance the prediction effect at voltage mutation points. Then, the moving window method with iterative function is used to combine the data-driven method with the PA-VR model, which realizes the online update of model parameters. Finally, the weightings on individual approaches are dynamically determined at different stages throughout the PEMFC lifecycle. The proposed hybrid method achieves an effective improvement in prediction performance by combining the overall degradation trend predicted by the PA-VR model and the local dynamic characteristics predicted by the data-driven method.

Nonlinear model inversion-based output tracking control for battery fast charging

Published in IEEE Transactions on Control Systems Technology, August 30, 2023

We propose a novel nonlinear control approach for fast charging of lithium-ion batteries, where health- and safety-related variables, or their time derivatives, are expressed in an input-polynomial form. By converting a constrained optimal control problem into an output tracking problem with multiple tracking references, the required control input, i.e., the charging current, is obtained by computing a series of candidate currents associated with different tracking references. Consequently, an optimization-free nonlinear model inversion-based control algorithm is derived for charging the batteries. We demonstrate the efficacy of our method using a spatially discretized high-fidelity pseudo-two-dimensional (P2D) model with thermal dynamics. Conventional methods require computationally demanding optimization to solve the corresponding fast charging problem for such a high-order system, leading to practical difficulties in achieving low-cost implementation. Results from comparative studies show that the proposed controller can achieve performance very close to nonlinear and linearized model predictive control but with much lower computational costs and minimal parameter tuning efforts.

Multilevel data-driven battery management: From internal sensing to big data utilization

Published in IEEE Transactions on Transportation Electrification, August 7, 2023

A battery management system (BMS) is essential for the safety and longevity of lithium-ion battery (LIB) utilization. With the rapid development of new sensing techniques, artificial intelligence, and the availability of huge amounts of battery operational data, data-driven battery management has attracted ever-widening attention as a promising solution. This review article overviews the recent progress and future trend of data-driven battery management from a multilevel perspective. The widely explored data-driven methods relying on routine measurements of current, voltage, and surface temperature are reviewed first. Within a deeper understanding and at the microscopic level, emerging management strategies with multidimensional battery data assisted by new sensing techniques have been reviewed. Enabled by the fast growth of big data technologies and platforms, the efficient use of battery big data for enhanced battery management is further overviewed. This belongs to the upper and macroscopic levels of the data-driven BMS framework. With this endeavor, we aim to motivate new insights into the future development of next-generation data-driven battery management.

Ensemble nonlinear model predictive control for residential solar battery energy management

Published in IEEE Transactions on Control Systems Technology, July 18, 2023

In a dynamic distribution market environment, residential prosumers with solar power generation and battery energy storage devices can flexibly interact with the power grid via power exchange. Providing a schedule for this bidirectional power dispatch can facilitate the operational planning for the grid operator and bring additional benefits to the prosumers with some economic incentives. However, the major obstacle to achieving this win–win situation is the difficulty in: 1) predicting the nonlinear behaviors of battery degradation under unknown operating conditions and 2) addressing the highly uncertain generation/load patterns, in a computationally viable way. This article thus establishes a robust short-term dispatch framework for residential prosumers equipped with rooftop solar photovoltaic (PV) panels and household batteries. The objective is to achieve the minimum-cost operation under the dynamic distribution energy market environment with stipulated dispatch rules. A general nonlinear optimization problem is formulated, taking into consideration the operating costs due to electricity trading, battery degradation, and various operating constraints. The optimization problem is solved in real-time using a proposed ensemble nonlinear model predictive control (EnNMPC)-based economic dispatch strategy, where the uncertainty in the forecast has been addressed adequately albeit with limited local data. The effectiveness of the proposed algorithm has been validated using real-world prosumer datasets.

Knee-point-conscious battery aging trajectory prediction of lithium-ion based on physics-guided machine learning

Published in IEEE Transactions on Transportation Electrification, April 11, 2023

Early prediction of aging trajectories of lithium-ion (Li-ion) batteries is critical for cycle life testing, quality control, and battery health management. Although data-driven machine learning (ML) approaches are well suited for this task, unfortunately, relying solely on data is exceedingly time-consuming and resource-intensive, even in accelerated aging with complex aging mechanisms. This challenge is rooted in the highly complex and time-varying degradation mechanisms of Li-ion battery cells. We propose a novel method based on physics-guided machine learning (PGML) to overcome this issue. First, electrode-level physical information is incorporated into the model training process to predict the aging trajectory’s knee point (KP). The relationship between the identified KP and the accelerated aging behavior is then explored, and an aging trajectory prediction algorithm is developed. The prior knowledge of aging mechanisms enables a transfer of valuable physical insights to yield accurate KP predictions with small data and weak correlation feature relationship. Based on a Li[NiCoMn]O 2 cell dataset, we demonstrate that only 14 cells are needed to train a PGML model for achieving a lifetime prediction error of 2.02% using the data of the first 50 cycles. In contrast, at least 100 cells are needed to reach this level of accuracy without the physical insights.

Physics-based model predictive control for power capability estimation of lithium-ion batteries

Published in IEEE Transactions on Industrial Informatics, February 2, 2023

The power capability of a lithium-ion battery signifies its capacity to continuously supply or absorb energy within a given time period. For an electrified vehicle, knowing this information is critical to determining control strategies such as acceleration, power split, and regenerative braking. Unfortunately, such an indicator cannot be directly measured and is usually challenging to be inferred for today’s high-energy type of batteries with thicker electrodes. In this work, we propose a novel physics-based battery power capability estimation method to prevent the battery from moving into harmful situations during its operation for its health and safety. The method incorporates a high-fidelity electrochemical-thermal battery model, with which not only the external limitations on current, voltage, and power but also the internal constraints on lithium plating and thermal runaway, can be readily taken into account. The online estimation of maximum power is accomplished by formulating and solving a constrained nonlinear optimization problem. Due to the relatively high system order, high model nonlinearity, and long prediction horizon, a scheme based on multistep nonlinear model predictive control is found to be computationally affordable and accurate.

Machine learning-based fast charging of lithium-ion battery by perceiving and regulating internal microscopic states

Published in Energy Storage Materials, January 7, 2023

Fast charging of the lithium-ion battery (LIB) is an enabling technology for the popularity of electric vehicles. However, high-rate charging regardless of the physical limits can induce irreversible degradation or even hazardous safety issues to the LIB system. Motivated by this, this paper proposes a machine learning-based fast charging strategy with multi-physical awareness within a battery-to-cloud framework. In particular, a reduced-order electrochemical-thermal model is built in the cloud to perceive the microscopic states of LIB, leveraging which the soft actor-critic (SAC) deep reinforcement learning (DRL) algorithm is exploited for the first time to train a fast charging strategy. Hardware-in-Loop tests and experiments with practical LIBs are carried out for validation. Results suggest that the battery-to-cloud architecture can mitigate the risk of a heavy computing burden in the real-time controller. The proposed strategy can effectively mitigate the unfavorable over-temperature and lithium deposition, which benefits the safety and longevity during fast charging. Given a similar charging speed, the proposed machine learning approach extends the LIB cycle life by about 75% compared to the commonly-used empirical protocol. Meanwhile, the proposed strategy is proven superior to the state-of-the-art rule-based and the model-based strategies in terms of charging rapidity, charging safety and computational complexity. Moreover, the trained low-complexity strategy is highly adaptive to the ambient temperature and initial charging state, which promises robust performance in practical applications.

Peak power estimation of vanadium redox flow batteries based on receding horizon control

Published in IEEE Journal of Emerging and Selected Topics in Power Electronics, February 17, 2022

The peak power of a vanadium redox flow battery (VRB) reflects its capability to continuously absorb or release energy. Accurate estimation of peak power is essential for the safe, reliable, and efficient operation of VRB systems, but also challenging as it is limited by various factors, such as currents, flow rates, temperature, and state of charge. This article proposes a new online model-based peak power estimation scheme for VRBs. First, the model parameters and system states are accurately estimated using the recursive least squares with forgetting and the unscented Kalman filter, respectively. Next, based on a linear time-varying VRB model and the estimated states, the peak power estimation is formulated into an optimal control problem, and the problem is solved using the receding horizon control (RHC). The influence of the predictive horizon on the estimated peak power is discussed. Finally, the effectiveness of the proposed RHC-based peak power estimation scheme is experimentally verified on a 5-kW/3-kWh VRB platform.

Multistage state of health estimation of lithium-ion battery with high tolerance to heavily-partial charging

Published in IEEE Transactions on Power Electronics, January 21, 2022

State of health (SOH) is critical to the management of lithium-ion batteries (LIBs) due to its deep insight into health diagnostic and protection. However, the lack of complete charging data is common in practice, which poses a challenge for the charging-based SOH estimators. This article proposes a multistage SOH estimation method with a broad scope of applications, including the unfavorable but practical scenarios of heavily partial charging. In particular, different sets of health indicators (HIs), covering both the morphological incremental capacity features and the voltage entropy information, are extracted from the partial constant-current charging data with different initial charging voltages to characterize the aging status. Following this endeavor, artificial neural network based HI fusion is proposed to estimate the SOH of LIB precisely in real time. The proposed method is evaluated with long-term aging experiments performed on different types of LIBs. Results validate several superior merits of the proposed method, including high estimation accuracy, high tolerance to partial charging, strong robustness to cell inconsistency, and wide generality to different battery types.

Control-oriented modeling of all-solid-state batteries using physics-based equivalent circuits

Published in IEEE Transactions on Transportation Electrification, November 25, 2021

Considered as one of the ultimate energy storage technologies for electrified transportation, the emerging all-solid-state batteries (ASSBs) have attracted immense attention due to their superior thermal stability, increased power and energy densities, and prolonged cycle life. To achieve the expected high performance, practical applications of ASSBs require accurate and computationally efficient models for the design and implementation of many onboard management algorithms so that the ASSB safety, health, and cycling performance can be optimized under a wide range of operating conditions. A control-oriented modeling framework is thus established in this work by systematically simplifying a rigorous partial differential equation (PDE)-based model of the ASSBs developed from underlying electrochemical principles. Specifically, partial fraction expansion and moment matching (PFE-MM) are used to obtain ordinary differential equation-based reduced-order models (ROMs). By expressing the models in a canonical circuit form, excellent properties for control design, such as structural simplicity and full observability, are revealed. Compared to the original PDE model, the developed ROMs have demonstrated high fidelity at significantly improved computational efficiency. Extensive comparisons have also been conducted to verify its superiority to the prevailing models due to the consideration of concentration-dependent diffusion and migration. Such ROMs can thus be used for advanced control design in future intelligent management systems of ASSBs.

Offline and online blended machine learning for lithium-ion battery health state estimation

Published in IEEE Transactions on Transportation Electrification, November 19, 2021

This article proposes an adaptive state-of-health (SOH) estimation method for lithium-ion (Li-ion) batteries using machine learning. Practical problems with feature extraction, cell inconsistency, and online implementability are specifically solved using a proposed individualized estimation scheme blending offline model migration with online ensemble learning. First, based on the data of pseudo-open-circuit voltage measured over the battery lifespan, a systematic comparison of different incremental capacity features is conducted to identify a suitable SOH indicator. Next, a pool of candidate models, composed of slope-bias correction (SBC) and radial basis function neural networks (RBFNNs), are trained offline. For online operation, the prediction errors due to cell inconsistency in the target new cell are then mitigated by a proposed modified random forest regression (mRFR)-based ensemble learning process with high adaptability. The results show that compared to prevailing methods, the proposed SBC-RBFNN-mRFR-based scheme can achieve considerably improved SOH estimation accuracy (15%) while only a small amount of early-age data and online measurements are needed for practical operation. Furthermore, the applicability of the proposed SBC-RBFNN-mRFR algorithms to real-world operation is validated using measured data from electric vehicles, and it is shown that a 38% improvement in estimation accuracy can be achieved.

Model order reduction techniques for physics-based lithium-ion battery management: A survey

Published in IEEE Industrial Electronics Magazine, August 17, 2021

To unlock the promise of electrified transportation and smart grid, emerging advanced battery management systems (BMSs) shall play an important role in health-aware monitoring, diagnosis, and control of widely used lithium-ion (Li-ion) batteries. Sophisticated physics-based battery models incorporated in the advanced BMS can offer valuable battery internal information to achieve improved operational safety, reliability and efficiency, and to extend the lifetime of the batteries. However, developed from the fundamental electrochemical and thermodynamic principles, the rigorous physics-based models are saddled with exceedingly high cognitive and computational complexity for practical applications. This article reviews prevailing order reduction techniques of physics-based Li-ion battery models to facilitate the development of next-generation BMSs. We analyze and comparatively characterize these techniques, mainly from perspectives of model fidelity, computational efficiency, and the scope of applications. By representing many effective and flexible reduced-order models as equivalent circuits, designers and practitioners, who do not have electrochemical expertise but with knowledge of circuit theory, can readily gain insights into multi-physical dynamics as well as their coupling effects inside the batteries. In addition, recommendations are made on how to select appropriate physics-based models for various model-based applications in battery management. Finally, the prospect of physical model-enabled BMSs is discussed, including the potential challenges and future research directions.

Adaptive ensemble-based electrochemical-thermal degradation state estimation of lithium-ion batteries

Published in IEEE Transactions on Industrial Electronics, July 14, 2021

In this article, a computationally efficient state estimation method for lithium-ion (Li-ion) batteries is proposed based on a degradation-conscious high-fidelity electrochemical–thermal model for advanced battery management systems. The computational burden caused by the high-dimensional nonlinear nature of the battery model is effectively eased by adopting an ensemble-based state estimator using the singular evolutive interpolated Kalman filter (SEIKF). Unlike the existing schemes, it shows that the proposed algorithm intrinsically ensures mass conservation without imposing additional constraints, leading to a battery state estimator simple to tune and fast to converge. The model uncertainty caused by battery degradation and the measurement errors are properly addressed by the proposed scheme as it adaptively adjusts the error covariance matrices of the SEIKF. The performance of the proposed adaptive ensemble-based Li-ion battery state estimator is examined by comparing it with some well-established nonlinear estimation techniques that have been used previously for battery electrochemical state estimation, and the results show that excellent performance can be provided in terms of accuracy, computational speed, and robustness.

Deep deterministic policy gradient-DRL enabled multiphysics-constrained fast charging of lithium-ion battery

Published in IEEE Transactions on Industrial Electronics, April 7, 2021

Fast charging is an enabling technique for the large-scale penetration of electric vehicles. This article proposes a knowledge-based, multiphysics-constrained fast charging strategy for lithium-ion battery (LIB), with a consciousness of the thermal safety and degradation. A universal algorithmic framework combining model-based state observer and a deep reinforcement learning (DRL)-based optimizer is proposed, for the first time, to provide a LIB fast charging solution. Within the DRL framework, a multiobjective optimization problem is formulated by penalizing the over-temperature and degradation. An improved environmental perceptive deep deterministic policy gradient (DDPG) algorithm with priority experience replay is exploited to tradeoff smartly the charging rapidity and the compliance of physical constraints. The proposed DDPG-DRL strategy is compared experimentally with the rule-based strategies and the state-of-the-art model predictive controller to validate its superiority in terms of charging rapidity, enforcement of LIB thermal safety and life extension, as well as the computational tractability.

Electrochemical model-based fast charging: Physical constraint-triggered PI control

Published in IEEE Transactions on Energy Conversion, March 17, 2021

This paper proposes a new fast charging strategy for lithium-ion (Li-ion) batteries. The approach relies on an experimentally validated high-fidelity model describing battery electrochemical and thermal dynamics that determine the fast charging capability. Such a high-dimensional nonlinear dynamic model can be intractable to compute in real-time if it is fused with the extended Kalman filter or the unscented Kalman filter that is commonly used in the community of battery management. To significantly save computational efforts and achieve rapid convergence, the ensemble transform Kalman filter (ETKF) is selected and tailored to estimate the nonuniform Li-ion battery states. Then, a health- and safety-aware charging protocol is proposed based on successively applied proportional-integral (PI) control actions. The controller regulates charging rates using online battery state information and the imposed constraints, in which each PI control action automatically comes into play when its corresponding constraint is triggered. The proposed physical constraint-triggered PI charging control strategy with the ETKF is evaluated and compared with several prevalent alternatives. It shows that the derived controller can achieve close to the optimal solution in terms of charging time and trajectory, as determined by a nonlinear model predictive controller, but at a drastically reduced computational cost.

Dispatch planning of wide-area wind power-energy storage scheme based on ensemble empirical mode decomposition technique

Published in IEEE Transactions on Sustainable Energy, December 3, 2020

This article addresses the challenging task of developing a procedure for the day-ahead dispatch planning of wind power which emanates from a wide geographical area. Using the complete ensemble empirical mode decomposition technique, it is shown that the low-frequency components of the area aggregated wind power account for the largest proportions of the perturbing energy harnessed from the wind. By taking advantage of the slow-varying characteristics of the low-frequency components, accurate forecast of these components is readily obtained and incorporated into the developed dispatch planning procedure. The dispatchability of the wide-area wind generation is facilitated by the buffering actions offered by a centralized power dispatch energy storage system, operating under a proposed power flows control strategy. The efficacy of the developed procedure is illustrated using a pumped hydroelectric system as the dispatch energy storage medium.

Constrained ensemble Kalman filter for distributed electrochemical state estimation of lithium-ion batteries

Published in IEEE Transactions on Industrial Informatics, February 18, 2020

This article proposes a novel model-based estimator for distributed electrochemical states of lithium-ion (Li-ion) batteries. Through systematic simplifications of a high-order electrochemical–thermal coupled model consisting of partial differential-algebraic equations, a reduced-order battery model is obtained, which features an equivalent circuit form and captures local state dynamics of interest inside the battery. Based on the physics-based equivalent circuit model, a constrained ensemble Kalman filter (EnKF) is pertinently designed to detect internal variables, such as the local concentrations, overpotential, and molar flux. To address slow convergence issues due to weak observability of the battery model, the Li-ion’s mass conservation is judiciously considered as a constraint in the estimation algorithm. The estimation performance is comprehensively examined under a wide operating range. It demonstrates that the proposed EnKF-based nonlinear estimator is able to accurately reproduce the physically meaningful state variables at a low computational cost and is significantly superior to its prevalent benchmarks for online applications.