A Robust Battery Management System (BMS) for real-time and accurate Battery State and Health Monitoring is the most critical subsystem of any application which uses Battery Energy Storage Systems (BESS). The accurate monitoring of battery parameters is important to prevent any unforeseen catastrophic situation. We have proposed state-of-the-art Battery monitoring algorithms for estimating the State-of-Charge, State-of-Health, State-of-Power, and State-of-Function of Li-Ion Batteries.

The BMS algorithms will be developed on Intel FPGA and include functionalities such as overcharge, undercharge, and overcurrent protection circuits as well as Automatic cell balancing and uploading the data to the cloud. Such a novel design provides a power-efficient as well as a high-speed solution for the highly complex BMS.

Demo Video

Project Proposal

1. High-level project introduction and performance expectation

With the increasing focus on environmental protection, energy conservation, and climate change, battery storage technologies have been emerging as an important source of clean and green energy. Various battery-held applications such as hybrid/electric vehicles (HEVs/EVs), mobile robotics, renewable energy power grids, satellites & aerospace, etc. require efficient usage of Battery Energy Storage Systems (BESS) as well as their reliable monitoring and management. BESS can be utilized as the main/unique or the auxiliary power source in such applications. Lithium-ion batteries (LiBs) have emerged as one of the most promising battery technologies for renewable energy owing to their excellent characteristics such as best energy-to-weight ratios, low self-discharge rate, no memory effect, improved lifespan, and high energy efficiency. Although battery-based electronic systems are the need of the hour and are inevitably permeating the market, challenges still exist in their development. One such challenge is the “charge anxiety” problem, which refers to the fear of running out of battery power in the midst of a critical application leading to catastrophic situations. Efficient monitoring, as well as safety and protection of the BESS, are some other challenges that need to be addressed in order to enhance user confidence in the adoption of battery-based emerging technologies. Another critical challenge is the development and deployment of cost-effective, low complexity & power-efficient Battery Management System (BMS) on embedded platforms for low-power IoT applications since the BMS is also dependent on the BESS for its power consumption. 

The BMS is an electronic system that monitors and maintains the battery's critical states, health, and other aspects of the operation. It is usually regarded as the “Brain of the Battery'', forming a critical component of a BESS and is required for constant monitoring of the BESS to keep it working safely and efficiently. The BMS performs a wide range of operations such as charge control, over-current and over-voltage protection, cell equalization, on-board diagnosis, prognosis, etc. However, the most important function in a BMS is the monitoring of critical battery states which includes State-of-Charge (SOC), State-of-Health (SOH), State-of-Power (SOP), and State-of-Function (SOF). 

SOC estimation is regarded as one of the most important functions in a BMS. Accurate estimation of the remaining charge indicated by the SOC levels is essential for alleviating the challenges associated with the performance of the application. Various methods for SOC estimation exist in the literature, however, the accuracy and reliability of the open-loop methods are contestable. Kalman filter-based SOC estimation techniques have gained a lot of attention due to the high accuracy of SOC estimation compared to the existing techniques. The BMS also monitors how the performance of the battery degrades with aging and passage of time when being put into usage by constantly estimating the battery's internal parameters. The SOH is an indicator of the health of the battery and is used to determine the longevity of the BESS and its cost-benefit analysis for the intended applications.

SOF is another important function that describes how the battery meets the power demands and is often interpreted by the maximum available output power. The prediction of Remaining Useful Life (RUL) is a crucial parameter for intelligent health monitoring of the battery pack. Accurate estimation of SOF and RUL will provide an insight into the choice of particular battery technology for a specific application whose power demands are considerably different for different applications. Another functionality in a BMS is cell balancing/ equalization where cells having different charge levels are equalized to improve the performance of the battery pack and ensure safe and optimal utilization. Other functions include charging control, overcurrent and overvoltage detection and protection, cell and string monitoring, fault detection as well as power and heat management.

In this proposal, we aim to develop an IoT-oriented BMS for real-time and accurate battery state monitoring such as SOC, SOH, SOP, SOF, and RUL parameters. The real-time state monitoring by using the state-of-the-art technology of adaptive state estimation will help to alleviate the Charge-anxiety problem. The proposed system-on-chip design using a low complexity architecture on FPGA will lead to the development of a power-efficient and low-cost battery-operated system, thus reducing the overall cost and complexity of the project and improving the sustainability of the battery-based system. The complete BMS unit will be implemented on a low-cost Embedded System (such as Intel DE-10 Nano) by integrating with active cell balancing and protection units,  as well as the hardware-software solution for battery state monitoring to be developed as a viable product for the envisaged BESS application. Finally, the data will be available over a cloud-based application for real-time decision making, making it an intelligent Cyber-Physical system-based BMS.

2. Block Diagram

3. Expected sustainability results, projected resource savings

The aim of the proposal is to develop an IoT-oriented Battery Management System (BMS) for renewable energy applications. The BMS will provide real-time and highly accurate battery monitoring and protection leading to resilient and reliable battery-based applications. It will also help to achieve the objectives of several Sustainable Development Goals such as SDG 7 (Affordable, reliable, sustainable, and modern energy for all), SDG 11 (Sustainable cities and communities, sustainable transportation system), and SDG 13 (steps towards climate action).

The technology demonstration for the project is designed with the following objectives:

1. To propose a novel methodology for online and accurate State-of-Charge (SOC) and parameters estimation of the Battery Energy Storage System (BESS) to estimate the charge remaining in the BESS for the renewable energy application.

2. To augment the battery state monitoring with State-of-Health (SOH), State-of-Power (SOP), State-of-Function (SOF), and Remaining useful life (RUL) of the BESS in order to enhance the longevity of the battery units.

3. To develop a cost-effective and power-efficient Battery Management System (BMS) by integrating the proposed IPRs with cell balancing, battery and charge protection units, onboard fault diagnosis and prognosis, and thermal management unit for efficient management & optimal utilization of the BESS resources.

4. To design a Cyber-Physical System (CPS) by integrating the BMS Controller (embedded system) with a cloud-based application/ Mobile App that can perceive and understand the real-time requirement (from the physical environment) based on the battery status, health, and take intelligent decisions regarding real-time requirements of the system.

4. Design Introduction

The aim of the project is to develop a cost-effective IoT-oriented Battery Management System (BMS) Controller Unit for real-time and accurate health monitoring of Battery Energy Storage System (BESS). The BMS will consist of hardware and software solutions for battery states monitoring, charge equalization, and data communication. The real-time and accurate BMS will be developed by designing and implementing a state-of-the-art Dual Kalman-filter based accurate State-of-Charge (SOC) estimation methodology, which will act as an indicator to estimate the charge remaining in the battery. The Kalman Filter is an important set of robust equations that is utilized by designers in modern Battery management systems, to estimate the battery states and parameters in real-time. The dual KF (here proposed D-SRUKF) algorithm, another variant of Kalman Filter, has become an interesting area of research due to its capability to estimate the States and the internal parameters of the battery with high accuracy and stability required for highly non-linear applications such as battery state monitoring.

The BMS controller will also monitor how the performance of the battery degrades with the passage of time due to aging by constantly estimating the battery internal parameters and capacity of the battery along with the SOC. To augment the battery monitoring operation, the State-of-function (SOF) and Remaining useful life

(RUL) in terms of capacity utilization of the BESS system will also be estimated to enhance its longevity and ensure optimal utilization of the resources.

The technology will be developed as part of a Cyber-Physical System (CPS) by integrating the BMS Controller (embedded system) with a cloud-based application/ Mobile App that can perceive and understand the real-time requirement. The information about the battery status, its health, and SOF will be disseminated to make intelligent decisions to respond to the requirements in real-time. Finally, the entire BMS Controller unit will be developed on a low-cost Embedded System comprising charge protection and balancing units as well as a hardware-software solution for battery status monitoring which will be developed as a viable product for a BESS application such as Electric Vehicle (EV).

Why Intel FPGA? 

The proposed IoT-oriented Battery Management System will be used for highly critical applications such as Electric vehicles, Power Grid (integrating renewables to the grid), Satellite navigation systems, etc., where the emphasis is on high accuracy, stability, and reliability of the computational system.

There is a need for a dual-core ARM Cortex-A9 MPCore processor as we aim to perform real-time monitoring of the battery states using computationally demanding model-based techniques. The Neon media-processing engine which has a double-precision FPU (floating point unit) can provide high accuracy for floating-point arithmetic operations such as multiplication and division operation. The 64 KB on-chip SRAM and 1 GB DDR3 SDRAM (32-bit data) will help to store high precision floating point values and also aid in memory addressing which are an integral part of the proposed solution. It is ideal for multitasking which is also important for IoT-based applications.

Another reason for selecting Intel FPGA is that it offers a large number of Logic Elements(LE), multipliers, and variable precision DSP blocks. These are needed for the implementation of the co-estimation methodology using the Dual-Square Root Unscented Kalman Filter algorithm. 

Furthermore, a number of sensors will be integrated as part of the solution along with data acquisition system modules, and therefore, we require efficient clock management in the design. Therefore, we propose to use the DE10-Nano Cyclone V SoC FPGA Board as part of our IoT-oriented BMS project as it provides extensive I/O interfaces.

5. Functional description and implementation

The following work plan is being proposed to develop the solution and achieve the stated objectives. 

Work-Plan 1: Accurate Estimation of State-of-Charge (SOC) of Li-Ion Batteries using a novel methodology based on Dual Kalman Filters.

The increasing demand for real-time and accurate state estimation in embedded systems has been increasing, which places harsh restrictions on the physical space and electrical power available to the computing system, particularly systems such as CPS. Some of these problems can be alleviated by implementing a few functionalities as hardware and others as software. This allows the use of complex algorithms to run at higher speeds and also be implemented as a system-on-chip solution. Since batteries are complex electrochemical devices with a distinct nonlinear behavior depending on various internal and external conditions, their monitoring is a challenging task. State-of-Charge (SOC) estimation is important as it helps to make sure the availability of a reliable system. SOC is defined as the remaining quantity of releasable charge in a battery with respect to the maximum available capacity. SOC is influenced by C-rate, temperature, and the age of the battery. Accurate estimation of SOC technique is an important area of research and various techniques exist in the literature for its estimation depending upon the type of battery chemistry. A new and emerging class of SOC estimation techniques make use of Kalman Filters and they have shown considerable improvement in terms of convergence, accuracy, and stability compared to other conventional techniques. We propose to estimate the SOC by applying the Dual – Square Root Unscented Kalman Filter (D-SRUKF) algorithm based on an equivalent circuit model as shown in the figure below

The proposed real-time and accurate SOC estimation methodology has been implemented and compared with the existing technologies as well as the other model-based methodologies. The results have shown that the proposed dual SRUKF is more accurate than the state-of-the-art and remains stable once it converges to true SOC value.

Work-Plan 2: Scheme of the Co-estimation of SOC, SOH, and SOF 

The co-estimation algorithm utilizes the relationships among various battery states and is, therefore, more practical and accurate compared to discriminate state estimation algorithms. In this proposal, we aim to co-estimate the battery states using the following scheme. 

1. To take into consideration the impact of the battery aging on the SOC estimation. The capacity of the battery is updated online by the SOH estimation, therefore the accuracy of the SOC estimation after battery aging is improved.

2. The impact of the SOC on the battery available power - Battery Open circuit voltage (OCV) varies with SOC. Therefore with an update of the SOC and the correlated OCV, the accuracy of the State of Power estimation will be enhanced.

3. As the battery ages, its internal resistance increases. The impact of the battery aging on the battery's available power is taken into account. The internal resistance is estimated during the SOH estimation process and can be utilized in the SOP estimation with no extra computation.

Work-Plan 3: Model-based SOF and Remaining Useful Life (RUL) prediction. 

An important and often overlooked functionality is to ensure proper utilization of the remaining charge available in the battery for a given application. A particular scenario can be envisaged where an automatic Electric Vehicle has to make intelligent decisions about accelerating or de-accelerating the vehicle based on the information of the remaining charge available to reach the destination or a nearby charging station. This functionality is known as the State-of-function (SOF) of the battery which describes how it meets the power demands and is interpreted by the maximum available output power. The SOF is an important indicator of the battery status which can also help in determining the choice of a particular battery technology feasible for meeting different application demands. The Remaining Useful Life (RUL) indicator will demonstrate the estimated usage period for the device to function before it warrants a replacement. SOF and RUL indicators will help to reduce the depth of discharge of the battery and help in increase the lifespan and reduce the cost of repeated replacements. 

Work-Plan 4: IoT Oriented Battery Management System 

Lastly, all the above indicators (SOC, SOH, SOF, RUL) can be co-estimated and used to design a novel and state-of-the-art BMS which will also be connected to the Internet through appropriate communication protocols. This will lead to an IoT-oriented (cloud) Cyber-Physical System (CPS) solution for the consumer where a web portal (mobile friendly) or a Mobile Application (Android and iOS) can be developed through the Microsoft Azure IoT. This IoT solution will offer a wide range of services to the consumer where the information of the battery status will be transmitted to the cloud through the Internet and the battery held unit will be presented with information regarding its requirement through the server. The BMS unit will then take intelligent decisions based on information to actuate a particular functionality of the application unit. 

Work-Plan 5: Charge balancing, charge protection, and BMS Development. 

Cell balancing or charge equalization is an important operational unit in any modern BMS. Over the lifetime of a battery pack, the voltages and capacities of the series-connected cells may vary at different rates, which can limit the overall performance of the battery pack or even lead to catastrophic scenarios. Therefore, charge-balancing is important to improve the energy and power delivery to and from the battery pack. 

Cell balancing methods are classified into passive and active cell balancing. Passive methods use electrical components such as resistors which dissipate the excess energy in higher capacity cells as heat to limit the voltages of such cells. On the other hand, active methods achieve better energy efficiencies by transferring the energy among cells, and thereby no energy is lost in the form of heat and the battery pack remains within the operating temperature. As a result, active methods require complex control algorithms with additional electronic interfaces. The following state-flow model is proposed to be utilized for developing the charge balancing and charge protection units.

Work-Plan 6: Cross-Validation & Standardization 

The above-mentioned proposed units consisting of real-time and accurate SOC, SOH, SOF, and RUL estimations will be integrated with other BMS functionalities to develop a holistic intelligent BMS unit. At present, the proposed technology has been validated using standard tools and battery test data obtained from the open-source Calce Battery Research Group, University of Maryland. The state-estimation methodology as well as the system-on-chip BMS needs to be tested with real-time data for different drive cycles, along with cross-validation and standardization of the technology.

6. Performance metrics, performance to expectation

Battery devices are becoming the favorite choice for electrical energy storage and supply in a wide range of motive and stationary power applications. Without significant improvements in battery technologies and Battery Management Systems (BMS), the future uptake of these electrochemical energy storage devices will remain a challenge. The continuous need for higher energy and power density of batteries as well as environmental concerns regarding hazardous materials in batteries has led to a boost in the development and manufacturing of lithium-ion batteries. Over the past decade, lithium-ion batteries have achieved significant penetration into various markets, where high specific energy and power densities are desirable. Almost as long as rechargeable batteries have existed, systems able to give an indication about the state-of-charge (SOC) of a battery have been around. Several methods, including those of direct measurements, book-keeping, and adaptive systems are known in the art for determining the SOC of a cell or battery of cells. 

Most of the existing solutions make use of book-keeping methods where the accuracy of the algorithm is contestable. They use the OCV-SOC curve which is predetermined by the manufacturer and therefore, does not take into account the aging of the batteries. With the upcoming new battery technologies, the OCV-SOC curve is found to be flat, and therefore, such methods are not at all applicable. The dual UKF (DUKF) algorithm of SOC estimation has become an interesting area of research due to its capability to estimate the States (here SOC) and the internal parameters of the battery which changes or degrades over time due to aging. The Square-root Unscented Kalman Filter (SR-UKF) is an improved square-root version of the UKF with the added benefit of numerical stability, robustness, and high accuracy for systems that optimizes its non-linearity. The application of SR-UKF is no longer restricted to the control domain, and its high-performance capabilities make it a suitable candidate for myriad applications. An accurate SoC determination method and an understandable and reliable SOC display to the user will improve the performance and reliability, and will ultimately lengthen the lifetime of the battery. This will increase the confidence of the consumers in the Battery based applications that are most susceptible to technology failure due to cost economy as well import dependency. A number of techniques are available in the literature to uniquely estimate the SOC and SOH and to some extent the SOF. However, the stringent resource constraint placed on the system-on-chip demands the BMS controller to incorporate low complexity algorithms to implement these functionalities for cost-effectiveness and reduced form factor. A method to co-estimate all the parameters (SOC, SOH, and SOF) offers a unique solution to the problem.

The demand for fast and accurate state estimation in embedded systems has been increasing which places harsh restrictions on the physical space and electrical power available to the computing system, particularly distributed systems such as CPS. These restrictions mean that there may be insufficient computing power available to run complex algorithms for high-end CPS applications such as remote monitoring, simultaneous localization, and mapping (SLAM), micro and nano-satellites, system optimization in UBSS, etc. Some of these problems can be alleviated by implementing certain functionality as hardware. This allows the use of complex algorithms to run at a reasonable speed or by allowing multiple functions to be implemented on a single chip, in a system-on-chip.

7. Sustainability results, resource savings achieved

The IoT oriented Battery Management System has the following functionalities:

1. Battery state estimation (SOC, SOH, SOP) using the proposed novel and highly accurate D-SRUKF Algorithm.

2. Battery Voltage, Temperature and Current Monitoring (up to 12 cells)

3. Cell Balancing (Passive Balancing)

4. Cell Protection (Over-voltage, under-voltage, and Over-current Protection)

5. Data upload to the cloud and touch screen display to the user.

For achieving all the above functionality, we are using the Intel DE-10 Nano Board which will be the Master controller. The BMS state estimation algorithm runs on the HPS (ARM Cortex A9 processor). The C code for the novel and highly accurate DSRUKF algorithm is implemented in a modular fashion and each function is optimized to reduce the time as well as hardware complexity. Coordinate Rotation Digital Computer (CORDIC) is used as the main computational unit for triangularization and Cholesky update operations and Matrix computation is used for weight, sigma point, Mean, correlation, and covariance calculations. 

The block diagram of the proposed D-SRUKF Battery state monitoring algorithm is presented below:

For state estimation, the real-time Voltage, Current, and Temperature (VIT) of the cells are provided by the DC2260A Analog Device Evaluation board. It performs monitoring and Passive cell balancing of the battery pack. The evaluation board is mounted with LTC6811 IC which is a multi-cell battery stack monitor that can measure up to 12 series-connected cells. It is ideal for getting accurate real-time measurements as it has a refresh rate of 290us and a measurement error of less than 1.2mV. The DC2260A is connected with the DE-10 nano Board via the LTC 2x7 header where they communicate via SPI protocol. 

After the real-time voltage and current data acquisition, the battery states (which is running on HPS) are estimated for all the cells to get the SoC, SoH, SoP, and SoF values as output of the algorithm. This data is passed to the FPGA Fabric which is interfaced with the Terasic daughter cards. The RFS daughter card is connected to the 2x20 GPIO-0 header and is used for cloud upload and ambient temperature measurement. The LT24 card is used to display the instantaneous SoC, SoH, SoP, and SoF values to the user.

For Charging the battery pack and driving the load, DC2044A Analog Device Board is used. It is a 55V buck-boost multi-chemistry battery charger mounted with LTC4020 IC. The circuitry provides Constant current/Constant voltage(CC/CV) charging.

To validate the working of the designed BMS and to verify the value of SoC and SoH, DC2973A evaluation board is used. It is mounted with the new LTC3337 IC which is a Primary Battery SoH monitor with a precision coulomb counter for a single cell. The results of the proposed D-SRUKF BMS algorithm can be compared with the output of the evaluation board for functional verification, sustainability results and expected performance gains.

8. Conclusion

The increasing environmental concern about climate change and burning fossil fuels has shifted the focus towards more sustainable forms of energy including renewable energy resources to ensure continuity of energy supply in the economy. On the other hand, Indian cities are grappling with the growing menace of air pollution in the last couple of decades. As the cities are growing in size, the transport sector has emerged as a major contributor to air pollution. Hence, minimizing pollution by appropriate technology and policy interventions are the need of the hour. BESS has emerged as an important solution to deal with such issues of green energy technology in various important sectors.

We have developed a cost-effective Battery Management System (BMS) for real-time and accurate health monitoring of BESS. The BMS consists of hardware and software for battery management including, among others, algorithms determining battery states. The real-time and accurate BMS is developed by designing and implementing a state-of-the-art Dual Kalman-filter based accurate State-of-Charge (SOC) estimation methodology, which will act as an indicator to estimate the charge remaining in the battery and hence determine its capacity utilization. The BMS also incorporates functionalities to co-estimate the State-of-Health (SOH), State-of-Function (SOF), State-of-Power (SOP), and Remaining useful life (RUL) of the batteries following SOC estimation. The co-estimation methodology will provide information about the health of the batteries which will enhance their longevity and ensure optimal utilization of other resources deriving energy from the BESS. The BMS also includes charge protection and charge equalization functionalities to protect overcharging/ asymmetric charging of the batteries and prevent the failure of the BESS in critical applications. Finally, the BMS Controller unit (embedded system) is integrated with a cloud-based application/ Mobile App that can perceive and understand the real-time physical environment based on the battery status and health using an appropriate communication protocol. This will ensure that the system has ubiquitous access to the Internet and can make intelligent decisions based on sufficient power capability to carry out specific functions and make decisions to respond to the requirements in real-time.