Intelligent Battery Systems: System‐Level Integration of Data‐Driven Learning, Optimisation and Predictive Management

<div class="section abstract"><div class="htmlview paragraph">The fuel economy of a vehicle can be improved by recuperating the kinetic energy when the vehicle is decelerated. However, if there is no electrical traction component, the recuperated energy can be used only by the other electrical systems of the vehicle. Thus, the fuel economy improvement can be maximized by balancing the recuperated energy and the consumed energy. Also, suitable alternator and battery management is required to maximize the fuel economy. This paper describes a design optimization process of the alternator and battery system equipped with recuperation control algorithms for a mid-sized sedan based on the fuel economy and system cost. A vehicle model using AVL Cruise is developed for cycle simulations and validated with experimental data. The validated model is used for the parametric study and design optimization of the alternator and battery systems with single and dual energy storage. In this paper, recuperation systems with Flooded, AGM, and Li-ion batteries are compared and different design optimization processes are presented depending on the battery types and system architectures.</div></div>

• Examined various battery types, BMS architectures, and modelling techniques. • Explored deterministic, non-deterministic, and hybrid SOC estimation methods. • Discussed SOH estimation and RUL prediction using various methods. • Exploration of digital twin technologies and battery digital twins (BDTs) for enhancing BMS functionality. • Discussion of future research directions in battery state estimation. Accurate state estimation is fundamental for the safety, reliability, and performance of electric vehicle (EV) batteries. This review highlights the critical need for precise evaluation of key battery states such as state of charge (SOC), state of health (SOH), and remaining useful life (RUL). It surveys the existing modelling and estimation techniques used in lithium-ion battery (LIB) systems. It explores parameter identification methods and the challenges associated with real-time monitoring and predictive diagnostics in dynamic EV environments. The paper further investigates the transformative role of digital twin (DT) technologies in addressing these challenges. By integrating battery digital twins technologies(BDTs): Internet of Things (IoT), cloud computing, artificial intelligence (AI), and extended reality (XR), it offers a virtual replica of the battery system. That enables continuous monitoring, predictive maintenance, and performance optimization. The review delves into the architecture, functions, challenges and future perspectives of battery digital twins (BDTs). And it emphasizes their potential to enhance traditional battery management systems (BMS) through intelligent, adaptive control. Through comparative analysis and identification of research gaps, this review provides a roadmap for future advancements in state estimation and digital twin integration.

Battery ensures power solutions for many necessary portable devices such as electric vehicles, mobiles, and laptops. Owing to the rapid growth of Li-ion battery users, unwanted incidents involving Li-ion batteries have also increased to some extent. In particular, the sudden breakdown of industrial and lightweight machinery due to battery failure causes a substantial economic loss for the industry. Consequently, battery state estimation, management system, and estimation of the remaining useful life (RUL) have become a topic of interest for researchers. Considering this, appropriate battery data acquisition and proper information on available battery data sets may require. This review paper is mainly focused on three parts. The first one is battery data acquisitions with commercially and freely available Li-ion battery data set information. The second is the estimation of the states of battery with the battery management system. And third is battery RUL estimation. Various RUL prognostic methods applied for Li-ion batteries are classified, discussed, and reviewed based on their essential performance parameters. Information on commercially and publicly available data sets of many battery models under various conditions is also reviewed. Various battery states are reviewed considering advanced battery management systems. To that end, a comparative study of Li-ion battery RUL prediction is provided together with the investigation of various RUL prediction algorithms and mathematical modelling.

In recent years, approaches to control performance and resource optimization for embedded control systems have been receiving increased attention. Most of them focus on theory, whereas practical aspects are omitted. Theoretical advances demand flexible real-time kernel support for multitasking and preemption, thus requiring more sophisticated and expensive software/hardware solutions. On the other hand, embedded control systems often have cost constraints related with mass production and strong industrial competition, thus demanding low-cost solutions. In this paper, it is shown that these conflicting demands can be softened and that a compromise solution can be reached. We advocate that recent research results on optimal resource management for control tasks can be implemented on simple multitasking preemptive real-time kernels targeting low-cost microprocessors, which can be easily built in-house and tailored to actual application needs. The experimental evaluation shows that significant control performance improvement can be achieved without increasing hardware costs.

Systematic experimental pulse test investigation for parameter identification of an equivalent based lithium-ion battery model

On-line estimation of state-of-charge of Li-ion batteries in electric vehicle using the resampling particle filter

This paper examines the health-conscious optimal control of lithium-ion batteries. The paper focuses specifically on the optimizing battery control policies to minimize solid electrolyte interphase (SEI) layer growth. However, its approach is applicable to other degradation mechanisms. The paper exploits the fact, established by the mathematical control literature1–4, that Fick’s second law of diffusion is differential flat for both linear and nonlinear diffusion problems. Differential flatness only applies to each diffusion medium in the battery separately. In the single particle model (SPM), for instance, each electrode’s diffusion sub-model alone is differential flat. Ensuring differential flatness for the entire battery is a two-step process. First, one must truncate redundant integrators from the two electrodes’ diffusion sub-models. Second, one must then relate the truncated integral state variable to the remaining state variables through an affine transformation. The differential flatness of the resulting battery model makes it possible to solve the health-conscious battery management problem efficiently using pseudospectral methods. This exploitation of the structure (i.e., differential flatness) of lithium-ion battery dynamics for more computationally efficient trajectory optimization is novel and significant. The existing literature already examines the problem of health-conscious optimal battery charging and discharging5,6. However, the optimal control tools used in this literature are not tailored to exploit the structure (e.g., differential flatness) of lithium-ion battery models. This can lead to higher computational burdens compared to the novel and more efficient approach we present here. The approach has the added advantage that it can be used for both offline trajectory optimization and online model-predictive control.The paper demonstrates the above ideas using the SPM. The optimization problem is formulated by equation [1-6] and the SPM with film growth model. The cost function in Equation [1] aims to track the reference state of charge (SOC) and at the same time minimize the film growth rate, which is defined by equation [7]-[14] in Ramadass’s paper7. The weight β represents a tradeoff/balance between aggressive charging and battery degradation. Constraints [2-6] place limits on battery charge/discharge current, reflecting battery management hardware capabilities. Furthermore, these constraints limit battery SOC to prevent over-charging and over-discharging. Finally, these constraints also place bounds on battery state variables contributing to damage phenomena such as lithium plating and mechanical degradation. For instance, we bound the concentration gradients in the battery cell and the overpotentials driving the lithium plating side reaction8–10. The optimization approach presented in this paper transforms these equations and constraints from a dynamic programming problem to a nonlinear programming (NLP) problem. Solving this NLP problem using traditional optimization methods leads to a health-conscious battery management policy. The value of this paper lies not in this policy, but rather in the computational efficiency with which it is obtained. Specifically, to the best of the authors’ knowledge, this paper represents the first attempt to exploit lithium-ion battery model structure for more efficient solution of the health-conscious optimal management problem.ACKNOWLEDGMENTSThe research was funded by ARPA-E AMPED program grant # 0675-1565. The authors gratefully acknowledge this support.REFERENCES1. M. FLIESS, J. LÉVINE, P. MARTIN, and P. ROUCHON, Int. J. Control, 61, 1327–1361 (1995).2. B. Laroche, P. Martin, and P. Rouchon, Int. J. Robust Nonlinear Control, 10, 629–643 (2000).3. M. Fliess and R. Marquez, Int. J. Control, 73, 606–623 (2000).4. T. Meurer, Automatica, 47, 935–949 (2011).5. R. Klein and N. Chaturvedi, Am. Control … (2011).6. V. Boovaragavan and V. R. Subramanian, J. Power Sources, 173, 1006–1011 (2007).7. P. Ramadass, B. Haran, P. M. Gomadam, R. White, and B. N. Popov, J. Electrochem. Soc., 151, A196 (2004).8. J. Christensen and J. Newman, J. Solid State Electrochem., 10, 293–319 (2006).9. X. Zhang, A. M. Sastry, and W. Shyy, J. Electrochem. Soc., 155, A542 (2008).10. N. Chaturvedi, R. Klein, J. Christensen, J. Ahmed, and A. Kojic, IEEE Control Syst. Mag., 30, 49–68 (2010).

Distributed Energy Resources (DER) have increased penetration into the electric grid in recent years due to the growth of many renewable energy technologies as well as more efficient small conventional generation sources. Microgrids have grown in popularity along with the increased usage of DERs as a way to manage the multiple DERs on a local level.1Microgrids contain energy storage devices to store and distribute the intermittent energy provided by DERs (see figure 1).Coupling a solar energy resource with a Li-ion battery energy storage, we study the autonomy (the percentage of time the system meets demand) and utilization (amount of capacity used) during yearly operation.2By including a demand schedule with the coupled solar and battery system, we study both how different demands, system sizes and solar insolation patterns will affect the autonomy and utilization of the system. Both islanded and grid-tied demand structures will be discussed.The battery model involved in this study is an electrochemical and transport based Li-ion battery model.3For the solar portion of the model we will assess both ideal insolation curves as well as site based insolation data. For the real-site data, several solar monitoring sites within the US will be employed. The real data will offer a contrasting look at how much the coupled system can be affected on a daily and seasonal basis. In addition, using real data will allow for a greater assessment about the stability of a coupled system in areas with strong insolation, but high variability against areas of weaker insolation, but low variability. A balance between a high level of both autonomy and utilization will be studied in conjunction with preferable demand structures. This balance will affect the structure and operation of a microgrid with both solar and battery systems. Acknowledgements The authors acknowledge financial support from the U.S. Department of Energy’s Advanced Research Projects Agency- Energy (ARPA-E), and the Solar Energy Research Institute in India and the United States (SERIIUS), as well as, Washington University in St. Louis’ McDonnell Academy Global Energy and Environmental Partnership (MAGEEP).

Analytical models have been developed to diminish test procedures for product realization, but they have only been partially successful in predicting the performance of battery systems consistently. The complex set of interacting physical and chemical processes within battery systems has made the development of analytical models of significant challenge. Advanced simulation tools are needed to be more accurately model battery systems which will reduce the time and cost required for product realization. As an alternative approach begun, the development of cell performance modeling using non-phenomenological models for battery systems were based on artificial neural networks (ANN) using Matlab 7.6.0(R2008b). ANN has been shown to provide a very robust and computationally efficient simulation tool for predicting state of charge for Lead Acid cells under a variety of operating conditions. In this study, the analytical model and the neural network model of lead acid battery for electric vehicle were used to determinate the battery state of charge. A precision comparison between the analytical model and the neural network model has been evaluated. The precise of the neural network model has error less than 0.00045 percent.

In electric and hybrid-electric aircraft, the battery systems are usually composed of up to thousands of battery cells connected in series or parallel to provide the voltage and power/energy requirements. The inconsistent cells could affect the battery pack and its performance or even endanger electric and hybrid-electric aircraft security; thus, the early fault diagnosis of the battery system is essential. A well-designed battery management system along with a set of reliable voltage and current sensors is required to properly measure and control the cells operational variables in a large battery pack. In this study, based on the battery working mechanism, a new, fast, and robust fault diagnostic scheme is proposed for a lithium-ion battery (LIB) pack that can be employed for applications such as electric and hybrid-electric aircraft. In this method, some faults such as the overcharge, overdischarge occurring in LIB packs can be detected and isolated, based on some predefined factors gained from the battery models in healthy, overcharge, and overdischarge conditions. Finally, the effectiveness of the proposed fast fault diagnosis scheme is experimentally validated with LIBs under a typical flight cycle.

Fundamental physical and (electro) chemical principles of rechargeable battery operation form the basis of the electronic network models developed for Nickel-based aqueous battery systems, including Nickel Metal Hydride (NiMH), and non-aqueous battery systems, such as the well-known Li-ion. Refined equivalent network circuits for both systems represent the main contribution of this paper. These electronic network models describe the behavior of batteries during normal operation and during over (dis) charging in the case of the aqueous battery systems. This makes it possible to visualize the various reaction pathways, including convention and pulse (dis) charge behavior and for example, the self-discharge performance.

State of charge (SOC) is a critical factor to guarantee that a battery system is operating in a safe and reliable manner. Many uncertainties and noises, such as fluctuating current, sensor measurement accuracy and bias, temperature effects, calibration errors or even sensor failure, etc. pose a challenge to the accurate estimation of SOC in real applications. This paper adds two contributions to the existing literature. First, the auto regressive exogenous (ARX) model is proposed here to simulate the battery nonlinear dynamics. Due to its discrete form and ease of implemention, this straightforward approach could be more suitable for real applications. Second, its order selection principle and parameter identification method is illustrated in detail in this paper. The hybrid pulse power characterization (HPPC) cycles are implemented on the 60AH LiFePO4 battery module for the model identification and validation. Based on the proposed ARX model, SOC estimation is pursued using the extended Kalman filter. Evaluation of the adaptability of the battery models and robustness of the SOC estimation algorithm are also verified. The results indicate that the SOC estimation method using the Kalman filter based on the ARX model shows great performance. It increases the model output voltage accuracy, thereby having the potential to be used in real applications, such as EVs and HEVs.

The lithium-air (Li-air) battery, with its usable energy density close to 1,700Wh/kg1, has captured worldwide attention as a promising battery solution for electric vehicles. However, a major hurdle facing the development of Li-air battery systems is their poor round-trip efficiency owing to the formation of electrically insulating lithium peroxide (Li2O2) at the cathode surface. Computational modeling has proven to be an indispensable tool in the analysis of battery systems, however, all existing simulations thus far treat the electrode/Li2O2 matrix as a homogenous continuum and utilize simply-shaped electrode morphologies, such as spheres and rods, to construct volume-averaged expressions (see Fig. 1a) for porosity and surface area and are insufficient to probe the effect of precise electrode microstructures and Li2O2 growth2,3. Utilizing a pore-scale transport-resolved model of the Li-air battery, the complex electrode and Li2O2morphologies can be directly incorporated and their effects on the system-level performance can be evaluated. In this work, we present a pore-scale transport resolved model (see Fig. 1b) of the Li-air battery that fully accounts for the electrode microstructure and peroxide growth4. This approach requires no empirical correlations regarding the electrode morphology. 3D reconstructed images of real carbon fiber cathode (Fig. 2) are used as geometric inputs to the model. Results obtained from our pore-scale model agree well with experiments and the validated model is then used to predict the galvanostatic discharge behavior of a Li-air cell for a variety of electrode morphologies and design parameters. Fig. 3 shows the cell voltage versus specific capacity curves for nanofiber electrodes of varying mean pore spacing from our 3D pore-scale model. The cell discharge capacity is limited by the spacing between nanostructures, which may lead to pore blocking and hence the reduction of active surface area. To capture the different growth morphologies two potential peroxide forming reactions, based on an intermediate oxygen reduction reaction where O2 reduces to becomes superoxide (O2 -), are incorporated as shown in Fig. 1c. Superoxide may form at the surface of the electrode and then follow one of two possible paths; reaction at the electrode surface to form lithium peroxide or diffusion into the bulk liquid to nucleate lithium peroxide in the electrolyte5. Lithium peroxide created at the electrode/electrolyte surface will produce a thin film growth morphology while peroxide particles generated further away in the electrolyte contributes to the formation of larger particle shaped or toroidal shaped growth. The competition between the different reactions can be explicitly accounted for in the pore-scale modeling framework. A discrete element method is used to simulate the nucleation and movement of lithium peroxide. Through extensive 3D simulations, we systematically examine the effect of drawing current, ORR rate coefficients, oxygen solubility, mean pore size and distribution on Li2O2growth and cell performance as to better understand the underlying physics of capacity fading during cycling. The methodology presented here can be applied to other electrochemical systems that include an insoluble product formation as a result of the reaction process and will be a valuable tool for rational design of electrode microstructures for improved cell performance.

Due to concerning risk of environment pollution, there is an urgent need to opt some means to reduce the risk. Electric Vehicle is the most evolving means in this field. But at the same time it faces certain challenges in the field of its battery management. The main challenges are its low travel range and large charging time which it needs to solve in order to control the dominance of Internal Combustion Engine based systems. To lower down the charging time, the charging current need to be increased, this causes an adverse effect in the battery by increasing its temperature and deteriorates its life span. It is prime concern to enhance the battery performance in order to make it safe. The paper proposes, power control tactics of associated photo voltaic and wind energy system. It allows the utmost usage of abundantly accessible renewable energy. In this we included battery management with a fuzzy controller, which controls the phenomenon of charging and discharging with load change and output power, and makes the arrangement reliable and efficient. The design is validated using MATLAB Simulink.

Battery systems are inherently networked systems, in which battery cells or packs are grouped by serial and parallel connections to meet power and capacity specifications. In realtime operation, battery packs must be controlled collaboratively so that their capacities can be fully utilized. By employing both analog charge/discharge currents and discrete-event bypass switches in networked battery systems, this paper develops a new hybrid control methodology to control the states of charge (SOCs) and optimize performance. The paper first establishes necessary and sufficient conditions for controllability under hybrid control mechanism. These conditions are then used to derive control strategies to achieve coordinated SOC control, state consensus, and performance optimization. Control algorithms are developed, their convergence properties proved, and optimality established. Simulation examples are used to demonstrate the methods and their performances using a production battery system.