Author:
Gavin McVeigh – Senior Director System Architecture at Dukosi
Date
06/30/2025
As SoC and SoH cannot be directly measured, various methods of computational complexity are used to determine their values. The accuracy of quantifiable metrics—temperature, voltage, and current—and the synchronization of data captured for all cells are critical inputs for the algorithms used to determine overall performance.
Confidence in the accuracy of the SoC estimate allows manufacturers to safely extract more usable energy from the cells, reducing the size, weight and cost of a battery pack for a given capacity, while also helping to accurately predict vehicle range to improve the ownership experience by giving the driver confidence in the dashboard value they see.
The aggregation of monitoring data, including temperature, voltage, and charge/discharge cycles over time helps to assess the battery's SoH during its first life, as well as building trust for safe and cost-effective reuse in next-life applications, promoting sustainability by maximizing the useful lifetime of every cell in the battery.
SoC and SoH estimation methods
Dukosi’s recent white paper, linked at the end of this article, discusses methods for estimating the state of charge (SoC) and state of health (SoH), focusing on their sensitivity to measurement and synchronization errors. A simulation-based approach applied to an equivalent circuit model (ECM) in various uses cases created a baseline dataset with assumed ideal accuracy. Measurement and synchronization errors were then introduced into the accurate dataset, and then fed into SoC estimation algorithms. The results were subsequently used as inputs for SoH estimation methods, which aim to determine the cell's total capacity and equivalent series resistance (ESR).
The accuracy of these methods was evaluated by comparing the SoC and SoH estimates to the actual values obtained from the initial simulation. Since the study is based on synthetic data, the model of the lithium-ion battery cell used to generate this data must be defined first. This “enhanced self-correcting” model is shown in Figure 2. Here OCV represents the open-circuit voltage of the cell at a given SoC, and R0 is the cell’s equivalent series resistance (ESR).
Click image to enlarge
Cell types and algorithms
SoC estimation is heavily influenced by the chemistry and design of the battery cell. Lithium iron phosphate (LFP) cells, commonly used in stationary battery energy storage systems, are increasingly being adopted in automotive applications. Algorithmically, these cells display a linear open-circuit voltage response with significant hysteresis, complicating accurate SoC estimation. Similarly, nickel manganese cobalt oxide (NMC) cells are widely used in automotive applications and other fields requiring high energy density.
In automotive contexts, both LFP and NMC cells are typically tested using an urban dynamometer driving schedule (UDDS) profile, while energy storage applications employ a fast frequency response (FFR) profile for LFP cells. The accuracy of SoC estimates depends heavily on the algorithm chosen. In this study, Coulomb counting and Sigma Point Kalman filters (SPKF) were selected as the estimation algorithms. Table 1 outlines the parameter settings for the simulations, while Table 2 details the eight simulation scenarios explored for each case. For each scenario and case combination presented in Tables 1 and 2, an ESC model was simulated using noise-free inputs, discretizing its equations at a sampling frequency of 100Hz and implementing them in Simulink.
Click image to enlarge
Click image to enlarge
SoC estimation
Coulomb counting is a common method used to estimate the state of charge of a battery. It's based on the principle that the total charge passing through the battery over time is proportional to the change in its SoC.
The SPKF is a nonlinear filtering technique that has gained popularity in recent years for battery state estimation, including SoC. It offers several advantages over traditional Kalman Filters, especially when dealing with highly nonlinear systems. The SPKF repeatedly executes two sub-steps every measurement interval. First, it uses the ESC model equations to predict the model state and cell voltage. Then, it compares the predicted to the measured voltage and adjusts its state estimate based on this feedback.
Pseudo-random Gaussian noise was introduced to the current, voltage, and temperature data obtained from the sensors to produce the use cases of Table 2. The SoC estimation results are listed in Table 3.
SoH estimation
The SoH is typically evaluated by tracking changes in total cell capacity and ESR as the cell ages. Total capacity is estimated in this study by applying methods of linear regression to the cell model equation. The estimation is formed using total-least-squares where the performance of weighted ordinary least squares (WLS) and approximate weighted total least squares (AWTLS) are available for comparison.
Since ESR can be easily observed from voltage measurements, it can be estimated accurately and quickly with a high-quality sensor system. Although various methods are discussed in the literature, this study concentrates on two approaches using total least squares.
The results are assessed by comparing the estimated SoH (total capacity or ESR) to the true SoH from the original simulation. The total capacity and ESR results are summarized in Tables 4 and 5.
Click image to enlarge
Click image to enlarge
Dukosi Cell Monitoring System (DKCMS)
Headquartered in Edinburgh, U.K., and with a presence in global markets, Dukosi has developed an innovative approach to battery cell monitoring with potential applications in electric vehicles, industrial transportation, and also in battery energy storage systems (BESS). Its “smart cell” solution offers high levels of accuracy, safety, reliability, and efficiency in battery performance.
Unlike conventional battery management systems, where sensors monitor packs or modules of cells, the Dukosi Cell Monitoring System (DKCMS) gathers data from each cell independently using a custom designed Cell Monitor IC that can accurately capture both voltage and temperature of each cell, along with necessary cell balancing functionality. By measuring parameters at the source, Dukosi's system is optimized for high accuracy and eliminates the potential for signal degradation, interference or latency variance that can occur in traditional monitoring methods.
The synchronously timed data collection of all cells is delivered with deterministic latency to the Dukosi System Hub chip, which is ideally integrated into the conventional BMS host.
Dukosi C-SynQ communications and near field connectivity
To facilitate seamless communication between the individual Cell Monitor chips and the BMS processor, Dukosi has developed a proprietary communication protocol, called C-SynQ, which ensures synchronous, reliable and efficient contactless communication, even in challenging environments.
The C-SynQ protocol works with Dukosi’s near field communications system to enable contactless data exchange between the Cell Monitor chips and the BMS via an efficient single bus antenna. This solution eliminates the need for wiring connecting the BMS to every module, which then runs more wires to every cell in each module. In comparison, Dukosi’s contactless architecture significantly reduces the complexity of traditional systems, leading to several advantages, such as reduced weight and volume, further automation of manufacturing, and flexible scalability at the cell level, enabling large energy storage systems that can easily be designed to fit any size, capacity, or space.
The Dukosi Cell Monitoring System (DKCMS) supports regulations such as the EU Battery Passport and completed comprehensive qualification testing following AEC-Q100 standards, meeting the stringent requirements and expected longevity of electric vehicles and stationary battery energy storage systems.
Conclusion
The available energy of a battery is contingent upon its state of charge (SoC) and overall capacity. To achieve high confidence in the accuracy of calculated SoC and SoH, the source data from each cell must be captured synchronously with the highest possible accuracy.
The Dukosi Cell Monitoring System is a unique battery architecture that synchronously captures all cell data, and delivers it to the BMS processor via near field network and patented C-SynQ protocol with deterministic latency. With its Cell Monitors attached to individual cells, it is capable of capturing incredibly accurate temperature and voltage data, which in turn empowers more accurate SoC and SoH estimations.