Recently, Professor Xiaosong Hu of VPSL and his collaborators published an expert insights article “State estimation for advanced battery management: Key challenges and future trends” on the authoritative journal Renewable and Sustainable Energy Reviews (IF: 10.556). For the first time, the state of the art in State of Charge (SOC), State of Energy (SOE), State of Health (SOH), State of Power (SOP), State of Temperature (SOT), and State of Safety (SOS) estimation are all elucidated in a tutorial yet systematical way, along with existing issues exposed. In addition, from six different viewpoints, some future important research opportunities and evolving trends of this prosperous field are disclosed, in order to stimulate more technologically innovative breakthroughs in SOC/SOE/SOH/SOP/SOT/SOS estimation.

Significance and Innovation

Battery management is of paramount importance for operational efficiency, safety, reliability, and cost effectiveness of ubiquitous battery-powered energy systems, such as electrified vehicles and smart grids with renewables. Owing to complicated electrochemical dynamics and multi-physics coupling, a trivial, black-box emulation of batteries that senses only voltage, current, and surface temperature obviously cannot result in high-performance battery management systems. How to accurately and robustly estimate and monitor critical internal states constitutes a key enabling technology for advanced battery management. Credible knowledge of State of Charge (SOC), State of Energy (SOE), State of Health (SOH), State of Power (SOP), State of Temperature (SOT), and State of Safety (SOS) is a prerequisite for effective charging, thermal, and health management of batteries. Battery state estimation has already evolved as a vast field of research, with a large number of approaches reported in the literature. Instead of surveying all the existing methods by comprehensively listing relevant papers, this work, for the first time, intends to provide useful, inspirational insights into key issues, technical challenges, and future research directions of this area in a concise, tutorial way. Besides, this article also intended to inspire more researchers and practitioners to contribute innovative ideas and tools for SOC/SOE/SOH/SOP/SOT/SOS estimation of power batteries in this field, which is of great significance to the research, design and development of next-generation battery management systems.

Article overview

Battery State Overview

Key battery dynamics can be often described precisely by a coupled electrochemical-thermal-aging model, each sub-model of the coupling multi-physics model with its own timescale. Specifically, some battery states, such as SOC/SOP/SOE, generally change in real time, thanks to fast-changing microscopic electrochemical parameters. Macroscopic temperature distribution updates in an intermediate timescale, due to physical structures of batteries and heat transfer characteristics. Battery SOH of the slowest timescale, manifested by slow-changing parameters, such as internal impedance/resistance increase and capacity fade, varies slightly in a short period of time. Battery safety state can be overall determined by evaluating the aforementioned states. Taking the safety level requirements of battery systems into account, SOS-relevant system designs and management may alter. Fig. 1 presents a schematic diagram highlighting the multi-timescale nature of SOC/SOE/SOH/SOP/SOT/SOS.

Fig. 1. The multi-timescale nature of multiple battery critical states

Definition of battery critical states

State of Charge (SOC): As one of the critical factors in battery management systems (BMS), SOC can be expressed by various problem formulations. In general, SOC refers to the available capacity expressed as a percentage of the nominal capacity. Using the tank of a fuel vehicle as an analogy, SOC provides the same functionality as the fuel gauge. Besides, from the perspective of battery electrochemistry, SOC refers to the charge contained in both anode and cathode electrode particles. The SOC variation reflects the distribution of lithium concentration in the electrode particles. Because the amount of available charge is highly dependent on the amount of lithium stored in the electrodes, SOC can be directly calculated in terms of mean lithium concentration.

State of Energy (SOE): Characterizing the remaining energy of the battery has been widely used in recent years to predict the remaining driving mileage of electric vehicles. Compared with SOC, its significance is that, under different SOC states, the battery's potential level will show a significant downward trend during the discharge process, so that the same charge change will bring different energy changes. Therefore, the widely used SOC can only represent the remaining power of the battery, and there is no intuitive reflection of the remaining energy.

State of Health (SOH): Electrochemical batteries inevitably experience gradual performance degradation during their service life, owing to side reactions. This leads to the so-called aging phenomenon that causes losses of lithium inventory and active materials. SOH is often used to quantitatively assess the level of battery aging in terms of capacity fade and internal resistance. A capacity fade of 20% and/or an internal resistance increase of 100% are often considered as the End-of-Life (EOL) of a battery in automotive applications.

State of Power (SOP): It generally refers to the available of power that a battery can supply to or absorb from the vehicle powertrain over a time horizon. Battery SOP can be viewed as a product of the threshold current and the corresponding voltage, while various operational constraints should be explicitly considered and respected. These constraints generally include the battery voltage, current, SOC, and even temperature.

State of Temperature (SOT): Up to date, there are few studies rigorously defining State of Temperature (SOT). Either internal temperature or temperature distribution has been reported in the existing literature. In general, thermal dynamics of a battery are manifested macroscopically by temperature distribution. For control purposes and ease of implementation, core, average, and surface temperatures of a battery are often representative of SOT. From a perspective of microscopic mechanism, SOT is elaborated primarily from heat source and structural heat conduction.

State of Safety (SOS): In the field of battery safety, different countries and organizations have classified the battery disaster level, but there will be slight differences. The hazard levels are also measured in a numerical way. For instance, hazard risk was defined as the product of hazard severity and hazard likelihood. If the battery dynamics are further considered, the SOS can be formulated as the reciprocal of a probability function for possible abuses, including voltage, temperature, charging and discharging currents, internal impedance, battery expansion, and battery deformation. According to this definition, SOS decreases as the abuse function increases.

Existing methods and key issues

SOC estimation methods and key issues: SOC estimation methods can be divided into direct calculation, model-based, and data-driven three main categories, as shown in Figure 2.

Fig. 2 Classification of SOC estimation methods

SOE estimation methods and key issues: Compared with SOC estimation, the nonlinear change characteristics of the cell equilibrium potential must be considered when making SOE estimation, so it will become more complicated and more challenging. So far, a considerable number of scholars have devoted themselves to improving the reliability of SOE estimation, and their research summary is shown in Figure 3.

Fig. 3 Classification of SOE estimation methods

SOH estimation methods and key issues: Many existing methods can be used to estimate the battery's SOH, and can be roughly divided into four categories: physical model-based methods, empirical model-based methods, data-driven methods, and incremental capacity-based methods Analysis (ICA) method. A summary of existing SOH estimation methods is shown in Figure 4.

Fig. 4 Classification of SOH estimation methods

SOP estimation methods and key issues: Compared with the widely studied SOC or SOH estimation, there are relatively few studies on battery SOP estimation. SOP estimation methods can be divided into two major categories, including feature map-based methods and model-based methods, as shown in Figure 5.

Fig. 5 Classification of SOP estimation methods

SOT estimation methods and key issues: Based on measurement data such as temperature (including surface temperature and ambient temperature), impedance, and various observers combined with a simplified thermal model or empirical impedance model, an online estimation of battery temperature distribution can be achieved. The classification of SOT estimation is shown in Fig. 6, including a method based on impedance-temperature monitoring, a method based on thermal models, and a fusion method.

Fig. 6 Classification of SOT estimation methods

SOS method and key issues: Safety assessment of battery systems based on disaster levels is an important task. However, SOS estimation is a recent issue in the field of battery state estimation, and there are many key issues that need to be further addressed. A summary of current SOS estimation methods is shown in Figure 7, which mainly includes qualitative evaluation and quantitative calculation of battery safety.

Fig. 7 Classification of SOS estimation methods

Outlooks

Advanced sensing technologies: With the rapid development of sensing technologies, the physicochemical reactions inside a battery are likely to be preciously measured using advanced sensors. It is straightforward that the most accurate and direct way to obtain battery states is to directly measure them, provided that feasible sensors exist. For example, piezoelectric sensors, the high-precision contact-type displacement sensors and fiber optic sensors. Moreover, low-cost sensing technologies, attempting to address the entrenched complications arising from battery degradation, rate dependency, external interference, etc., will be conducive to a new, realistic implementation of efficient battery monitoring and health management.

Multi-state joint estimation: Battery states are coupled and interact with each other. To estimate one state independently while ignoring others can only obtain relatively satisfactory results under certain constraints. The multi-state joint estimation and prediction, considering a multi-field coupling of the internal electro-thermal-aging-mechanical conditions of a battery, is a promising but still challenging research direction. It is meaningful and valuable to elevate state-of-the-art joint estimation techniques one step beyond, achieving powerful co-estimation techniques of more than three states, by sufficiently examining the hierarchy and coupling of states in different spatial and temporal scales. Additionally, in comparison with applications of just one state estimation, more computational efforts would be required for multi-state estimation scenarios. In light of this, devising advanced approaches such as fractional-order calculus and multi-timescale estimator to effectively improving the accuracy of multi-state estimation, with an acceptable computational efficiency, becomes another essential future direction.

Scalability: With the continual advancement of state estimation algorithms themselves, more attention should be paid to scalability issues of state estimation, from cell to module, or even pack levels. In the case of a battery module, or pack, how should the estimation scheme adapt, in order to ensure a reasonable balance among accuracy, computational cost, and safety would be a ponderable question. Put another way, different estimation configurations should be sufficiently examined, including distributed and lumped configurations, to accommodate system-level estimation requirements. Currently, extensive research efforts are still needed to better address the scalability of state estimation to battery module and pack levels.

Holistic SOS monitoring and safety management: In the past, safety was mainly studied as an attribute of a system, rather than a system state. Now, the safety of a battery system has been researched as a state, which gains increasing attention from both academia and industry. Due to the requirements for the accurate numerical calculation of SOS, the SOS monitoring of a battery in various physical fields and conditions, such as mechanical, electrical, thermal, and electrochemical, will become a significant research direction for life-cycle battery prognosis and health management. Nowadays, enough accuracy and practicability of the SOS implementation is still tremendously challenging to obtain. By defining more sub-functions and recalibrating existing sub-functions based on investigations in various physical fields and conditions, the overall performance of SOS estimation can be further improved. Moreover, individual sub-function that could affect the SOS of a battery system is necessary to be further studied in depth.

Artificial intelligence: Artificial intelligence algorithms can complement traditional mathematical algorithms with strong capabilities of classification and linear/nonlinear regression. With the increasing integrity and accuracy of battery system big-data, integrated state estimation and prediction, combined with intelligent optimization algorithms, such as deep learning and reinforcement learning, will become growingly popular in the community of battery management and integration. For battery health state estimation, apart from SOH that evaluates the current health status, the remaining useful life (RUL) estimation is also essential for practical applications. It is worth noting that battery capacity/power capabilities gradually degrade over real-world operations until its end of life (EOL). How to use a small sample of early aging phase to construct a reliable artificial intelligence-based approach for precise battery RUL or aging-path estimation is of extreme importance for advanced battery management. Besides, exterior characteristic parameters such as the current, voltage, and temperature, are often selected as training input. Consequently, the trained artificial intelligence-based methods are unable to describe the internal electrochemical behaviors of a battery, which limits their applications under varying operating conditions. More importantly, artificial intelligence-based methods often need a large amount of data for training, and the training process is time-consuming. Therefore, studying the estimation of various state variables, incorporating internal electrochemical parameters into the algorithm training, and improving computational efficiency constitute major challenges in improving the performance of artificial intelligence-based methods.

Comprehensive evaluation with multiple metrics: The apparent variability and complexity in operating conditions, such as strong electromagnetic interference, wide temperature and current ranges, are representative of an electrified vehicle application scenario. Improvements of the robustness, precision, and fault tolerance of optimization algorithms for battery state estimation is an imperative research direction, which determines whether estimation algorithms can be really applied in practice well. Studies which investigated the theoretical analysis of a fundamental relationship between state estimation accuracy and the measurement data, with the consideration of sensor noise and measurement uncertainty, can guide the experimental design for system identification and data selection for online estimation. Besides, considering the bus communication delay in BMSs, sensors will inevitably do asynchronous sampling. In addition to adopting modified sampling strategies (e.g., the voltage is first sampled and the current later), improvements of state estimation algorithms are needed accordingly. Also, the parameters of battery models may vary frequently due to changing driving conditions, in terms of SOC, temperature, and current profiles, etc. Thus, the online learning abilities of state estimation methods, which can update model parameters dynamically, are of prominent importance to some battery operational scenarios, like electrified vehicle applications.

Fig. 8 Future trends of state estimation in advanced battery management system

Pulication information

Source：

Xiaosong Hu*, Fei Feng, Kailong Liu, Lei Zhang, Jiale Xie, Bo Liu, State estimation for advanced battery management: Key challenges and future trends, Renewable and Sustainable Energy Reviews, 114:1-13, 2019.

Extra introduction

Renewable and Sustainable Energy Reviews

Renewable and Sustainable Energy Reviews (RSER) is an international authoritative journal in the field of energy. Articles on this journal include review articles, expert insights, original research, case studies, and new technology analysis. Among them, the expert insights are commissioned mini-reviews from field leaders on topics of significant interest. This article is the first expert insight on this journal in the field of new energy vehicles.