Method for evaluating the accuracy of state‐of‐charge (SOC)/state‐of‐health (SOH) estimation of BMSs

The battery management system (BMS), which is widely adopted in various eco‐friendly automobiles that use batteries for propulsion, increases energy efficiency and extends the usable life by monitoring and optimally managing batteries. Eco‐friendly automobile original equipment manufacturers demand improvements in state‐of‐charge (SOC) and state‐of‐health (SOH) accuracies. However, because standards for evaluating accuracy do not exist, evaluations are performed using methods suggested by BMS developers or researchers, which differ from each other. Thus, the reliability of the SOC and SOH accuracy results is low. To accurately evaluate the SOC/SOH, it is necessary to reflect various conditions such as the power of the battery during actual vehicle operation and the temperature inside and outside the battery at the time. Therefore, a method is proposed herein that individually or simultaneously evaluates the SOC and SOH estimation accuracies of the BMS while considering an eco‐friendly vehicle's actual driving environment, instead of simple charge and discharge test conditions and room temperature. The reference values for evaluating the accuracy of SOC/SOH estimation in a battery can be defined as follows: Full charge and full discharge are conducted before applying the evaluation pattern to calculate the precise available capacity of the battery. Subsequently, it is fully charged and discharged to its initial SOC. A load pattern reflecting the load conditions of the subsequent application is repeatedly applied. After concluding the evaluation pattern, the remaining capacity is discharged, thereby determining the reference value. The proposed method is applied to two types of BMSs to evaluate its accuracy in SOC/SOH estimation and the validity of the proposed method was verified.

The battery management systems (BMSs) are currently used in various eco-friendly automobiles, such as in hybrid electric vehicles (EVs), plug-in hybrid EVs, EVs, and fuel cell EVs, that use batteries for propulsion.A BMS optimally manages a battery to increase its energy efficiency and lifespan by monitoring the battery's voltage, current, and temperature in real time and estimating its state of charge (SOC) and state of health (SOH). 1 A battery is a device that converts chemical energy into electrical energy via a redox reaction between the anode and cathode materials.Furthermore, it exhibits nonlinear voltage characteristics depending on the charging and discharging currents.][4][5] In addition to the Coulomb counting, several methods have been proposed to estimate the SOC.These include OCV, model-based, data-driven, hybrid, 6 and combined methods. 7In addition, various methods, such as the total leastsquares algorithm for capacity estimation, 8 hybrid neural networks, 9 and improved long short-term memory (LSTM) algorithms, 10 have been studied to estimate the SOH. 11,12ccordingly, original equipment manufacturers manufacture eco-friendly vehicles that demand certain accuracies in terms of the SOC and SOH, that is, typically within 5% and 10% for the SOC and SOH, respectively.Several studies and patents related to SOC and SOH estimation algorithms exist; however, accuracy standards have not been stipulated. 13The SOC/ SOH estimation accuracy evaluated by BMSs differs by company, thereby resulting in its low reliability.In some areas, SOC and SOH accuracy is determined by simple charging and discharging, which is extremely low or unmeasurable when actual charging and operating conditions such as low-speed city driving and highspeed freeway driving and environmental conditions such as summer and winter are considered.
Table 1 shows the studies verifying the SOC and SOH estimation performance by applying various battery state estimation algorithms.Park et al. proposed a method for simultaneously estimating SOC and SOH by applying the model to a dual extended Kalman filter (DEKF) after improving the accuracy of the model through the multiple adaptive forgetting factor-recursive least-squares algorithm for identifying battery model parameters. 14The SOC estimation performance was verified by applying the arbitrary constant-current (CC) profile, and the SOH estimation   16 Álvarez Antón et al. estimated the SOC by applying the support vector machine method and verified the SOC estimation performance through a CC-CV current load profile. 17All of the studies mentioned above were tested at room temperature (25°C).Sun et al. proposed the SOC estimation method by applying an extended state observer. 18This paper applied the CC pulse load profile for verifying SOC estimation performance.In addition, Liu et al. estimated the SOC using a state observer and verified the SOC estimation performance through hybrid pulse power characterization (HPPC) and the United States Advanced Battery Consortium (USABC) load profiles. 19The above study did not mark the temperature environment.Tian et al. estimated SOC and SOH using a convolution neural network and verified SOC estimation performance through a short charging load profile at 25°C and 40°C. 20Lin et al. estimated SOH through a probabilistic neural network and verified it through a CC-CV load profile in a 25°C environment condition. 21In previous studies, it can be confirmed that the state estimation algorithm is focused, and the detailed evaluation procedure for performance verification is insufficient.Furthermore, there are many international standards that can be referred to in BMS research and development of EVs.experiments are not defined.The USABC test procedure manual presents standards for verifying functions, such as the safety and performance of EV battery packs.Various international standards provide procedures such as test conditions and environments for BMS research and development, but for example, the description of the SOC evaluation process, which is the basis of the experimental procedure, is insufficient.These points confirm that there are no specifications and standards to verify accuracy after developing the state estimation algorithm, and the developed algorithms evaluate the state estimation accuracy by applying different load profiles, such as the dynamic CC profile, CC-CV profile, and USABC.Further, because there are no existing criteria for evaluating the accuracy of international certification standards, a method is required to evaluate the accuracy of the state estimation algorithm by considering the actual BMS operation state.Therefore, this paper suggests an evaluation method for the accuracy of SOC/SOH estimation at BMS by applying data considering the environment and driving conditions of real-world EVs.To verify the proposed battery SOC/SOH accuracy evaluation procedure, EV power load profile is applied to the EV battery pack after the battery is set up to a specific SOC.It is then compared and verified with the SOC calculated in BMS.Also, after the battery is set to a specific SOC, the driving load profile of the actual vehicle is fully reflected and charged/discharged, and the internal parameters are compared through a capacity extraction experiment after the load profile is applied to verify SOH estimation performance.The proposed method was applied to two types of BMS to evaluate the accuracy of SOC and SOH estimation to verify its effectiveness.It is possible to improve the performance of predicting the daily charging mileage and life by utilizing the SOC and SOH of EVs with accuracy through the method proposed in this paper.After developing the state estimation algorithm considering battery types, a load profile can be applied and evaluated according to the SOC/SOH estimation accuracy evaluation method proposed in this study; therefore, all battery types can be applied to the evaluation profile.When developing SOC/SOH estimation algorithms, it is necessary to check the specifications of the microcontrollers applied to the BMS and consider the calculation complexity and computational efficiency.When the SOC/SOH estimation accuracy is evaluated by modifying and applying an evaluation pattern, it can be evaluated by reflecting the operational characteristics of the target application.
This paper is structured as follows: Section 2 shows the evaluation procedure method.Section 3 describes the experiments and verification results for verification, and Section 4 describes the conclusions.

| Defining SOC and SOH estimation error
The estimation accuracies for the SOC and SOH should first be defined.Herein, "SOC/SOH estimation accuracy" is defined as the difference between the SOC/SOH value estimated by the target BMS being evaluated and the SOC/SOH of a reference.The definitions and formulas for the SOC and SOH are presented in several studies and are typically defined as follows: SOC is the remaining capacity of a battery and the ratio of the current remaining capacity to the nominal capacity, which indicates the currently available battery capacity.The SOC and SOH of a battery are defined using Equations ( 1) and ( 2), respectively,

SOH Maximum charging capacity of battery
Rated capacity of battery : .
Therefore, the SOC reference can be confirmed using a battery residual capacity test based on the SOC definition, because a SOC reference that operates and compares the SOC in real time does not exist.Even if a SOC reference exists, it will yield a higherror estimated value, which is meaningless.Hence, a SOC estimation evaluation method is proposed herein.
The procedures involved are as follows: (i) Set the battery to a SOC setting, (ii) charge and discharge the battery based on the pattern specified in the evaluation, (iii) complete the pattern, and (iv) compare the last SOC calculated by the BMS with the ratio of the capacity discharged to the cut-off voltage of the battery (reference SOC).The formula for calculating the SOC estimation error (%) is as follows:

SOC estimation error SOC estimation by BMS Discharged capacity after pattern
Rated capacity ( The SOH reflects the battery lifespan and is defined as the ratio of the usable capacity or internal resistance of the current battery to that of a new battery.The end of life (EOL) of a battery must be defined in advance.For example, a battery is regarded to have reached its end of life if its capacity is reduced by 20% or its internal resistance is increased by 150% compared with the conditions of a new battery.A degraded battery must be assessed to evaluate the SOH estimation accuracy.A battery selected for evaluation must have been degraded through realworld usage or artificially degraded through a repeated charge/discharge pattern in a laboratory environment.The reference SOH of the target battery can be determined by battery capacity or internal resistance tests.Typically, the SOH can be observed by comparing the capacity and internal resistance ratio of a degraded battery to those of a new battery.Hence, an SOH estimation evaluation method is proposed.
The procedures involved are as follows: (i) set the battery to a SOC setting, (ii) charge and discharge the battery in close resemblance to a real-world vehicle driving pattern, (iii) complete the pattern, (iv) discharge the battery to the cut-off voltage, (v) verify the charge and discharge capacity based on a full charge and full discharge, respectively, and (vi) compare the capacity and internal resistance ratio of the battery to those of a new battery on the same model.The SOH estimation error (%) is calculated as follows:

SOH estimation error SOH estimation by BMS Battery SOH that can confirm degradation
The SOH estimation error is contributed by the capacity and internal resistance of the SOH of the battery, whose degradation can be observed.The internal resistance method can be classified into alternating current-internal resistance (AC-IR) and direct current-internal resistance (DC-IR) methods.For instance, when the estimated SOH error is set to the change in capacity, charge and discharge operations are performed in an evaluation pattern at a particular SOC on a degraded battery whose new capacity is 100 Ah, and the battery is discharged to the cut-off voltage.If the capacity of the target battery is 90 Ah after a full charge and full discharge, then when the BMS-estimated SOH is 95% after the pattern is completed, the reference SOH is calculated to be 90% (90/100 × 100), and the estimated SOH accuracy is 5% (95 − 90).

| Configuration diagram for the evaluation of SOC and SOH estimation error
The SOC/SOH estimation accuracy (error) can only be evaluated using battery modules or battery packs with a BMS equipped with evaluation algorithms.Although testing/evaluation can be performed using a battery simulator, the results would not reflect the actual characteristics of a real battery.Therefore, in this study, the SOC/SOH estimation accuracy was evaluated using a real battery on which the BMS algorithm was applied.
Figure 1 shows the configuration for evaluating SOC/SOH accuracy.The evaluation equipment comprises a charger and discharger that charges and discharges the battery, respectively, a charge/discharge controller operated with an SOC/SOH evaluation pattern, an environmental chamber that simulates the external environment of the target battery, and BMS, a signal conversion device that converts the sensor value of the battery into a controller area network (CAN), a power supply that simulates the BMS power and control signals (e.g., ignition key, charger connection signal, etc.), and a data storage and analysis device that stores and analyzes data by receiving the SOC/SOH estimated by the BMS and data measured from the charger/discharger in the CAN format.A temperature simulator controls the internal temperature of the battery and is divided into two types depending on the thermal management of the battery.The air-cooled type is a temperature chamber that can control the temperature and air volume inside the battery, and the water-cooled type can control the temperature and flow rate of the coolant.

| Evaluation pattern and procedure (sequence) for evaluating SOC and SOH estimation errors
EV mode can be divided into driving mode, stop mode, and charging mode.Simple charge/discharge test generally corresponds to CC-CV mode conditions.In an EV, it is a mode in which charging (slow or fast charging) or the electric load is operated when the vehicle is not running (the electric motor is not driven).Meanwhile, in the driving mode, the torque of the motor is required according to the required speed of the vehicle, and the current of the battery is consumed to satisfy the torque of the motor.In addition, the battery is charged by regenerative braking during coasting or deceleration after acceleration.
A mode in which it is difficult to estimate the SOC accuracy in terms of SOC accuracy during the entire charging/discharging process of the vehicle corresponds to a driving mode, not a stop or charging mode.The reason is that the voltage and current of the battery change much more per unit of time.In addition, this is also because the amount of change in current and voltage changes more nonlinearly according to the operation of the motor.In addition, the mode that requires more accurate SOC estimation will be the driving mode, which requires accurate mileage estimation according to the remaining battery capacity.Therefore, simple charging and discharging test conditions are required for estimating the SOC accuracy, but it is more important to evaluate the SOC accuracy in the actual EV driving conditions and driving environment.Figure 2 shows the voltage and current of the battery during fast charging and city driving (urban dynamometer driving schedule [UDDS]) of the IONIQ5 EV.
In this study, the algorithm to be embedded in the vehicle BMS was evaluated.Therefore, instead of the simplified CC or CV pattern for SOC/SOH accuracy evaluation, the power pattern derived from the vehicle driving schedule was used.However, BMS is applied to various applications, and the load power pattern is not the same, so it is necessary to analyze the load consumption pattern of the target application and design the evaluation pattern.For instance, a chassis dynamometer pattern can be used to evaluate the range of an EV on a full charge.The UDDS and highway fuel-economy test cycle (HWFET) are typical modes used in the chassis dynamometer. 22,23For an identical or similar ecofriendly vehicle equipped with a high-voltage battery pack subject to BMS, the voltage and current data of the high-voltage battery gathered while driving in UDDS and HWFET modes with the chassis dynamometer can be used as power evaluation patterns.Voltage and current battery data can be simulated using vehicle modeling if actual driving data for an eco-friendly vehicle are unavailable.
Figure 3 illustrates the power data of the high-voltage battery used in the UDDS and HWFET modes (speed) of an electric passenger vehicle (IONIQ5 EV). 24The corresponding UDDS and HWFET cycles may be combined, or each cycle may be repeated to apply the actual vehicle driving time to the evaluation pattern time.Actual driving measurement data can be obtained by attaching voltage and current sensors to a vehicle's battery if the SOC/SOH evaluation pattern of an eco-friendly vehicle tested on a real road is required.

| Preparing batteries for evaluation and verifying SOC and SOH references
To simultaneously evaluate the SOC and SOH at a realworld vehicle level, a degraded battery is recommended instead of a new one.With a new battery, there is not enough change in capacity or internal resistance before and after use to evaluate the SOH estimation error.In terms of SOC evaluation, a degraded battery is set as the BMS target because the SOC can differ owing to the different rated capacities afforded based on the battery degradation amount.The capacity or the internal resistance should be analyzed before the evaluation pattern is completed to verify the SOC and SOH estimation accuracies of the reference.
The capacity is typically based on the discharged capacity, and the discharge capacity afforded by the current integration is analyzed by repeating the pattern, full charge → rest → full discharge → rest, as suggested by the battery manufacturer.A new battery should be inspected if the rated capacity suggested by the manufacturer is similar to the discharge capacity obtained from a test.
For a degraded battery, its SOH reference can be verified using the current discharge capacity based on the ratio of rated capacity to the current available capacity of a new battery.If only a new battery is available, then it should be degraded in advance via charge and discharge degradation cycles.A degradation cycle involves one full charge and one full discharge at a certain C-rate, which is defined as one cycle.The capacity and internal resistance changes due to repeated test cycles should be verified in advance.
Degradation cycle testing may require several months to complete.To reduce the test time, evaluation can be performed by increasing the C-rate or increasing the battery's ambient temperature from room temperature (25°C) to a high temperature (e.g., 50°C or higher).In this study, the internal resistance measurement was limited to that of the DC-IR.Moreover, the AC-IR can be evaluated using electrical impedance spectroscopy for impedance measurement by frequency. 25igure 4 schematically shows the HPPC profile and calculation of the DC-IR at each SOC.The DC-IR can be calculated (ΔV/ΔI) using the voltage change (ΔV) and current change (ΔI) during the charging and discharging of the current within a duration (Δt).A DC-IR test was conducted in this study using an HPPC test. 26

F I G U R E 2 Battery voltage and current according to electric vehicle mode (charging mode [left] vs. driving mode [right]
). SOC, state of charge; SOH, state of health.

| Initial battery SOC selection
The accuracy of the SOC estimation was evaluated after fully charging a degraded battery, setting it to a particular SOC setting (e.g., 80% or 60%, etc.), resting the battery (at least 30 min for lithium-ion batteries), and resetting (correcting) the SOC of BMS.Patterns in the order of full charge, rest, full discharge, and rest may be added to verify the capacity before a particular SOC setting.The SOC of a degraded battery is set as the SOC for a current battery that allows the SOC to be set based on the capacity of the degraded battery, and not a new battery.As an example, for the rated capacities of new and degraded batteries of 110 and 100 Ah, respectively, the 80% SOC setting with a 20% SOC of the discharge capacity is rated based on the degraded battery's 100 Ah rated capacity and not on the new battery's 110 Ah; this implies that 20 Ah must be discharged.
Although the initial SOC setting may represent the most commonly used SOC among the entire window of the SOC, it should be evaluated by varying the initial SOC setting of the entire section in which the vehicle operates.
SOC reset (correction) indicates a vehicle sequence that can correct itself when an internal BMS SOC accuracy error is set to the SOC (corrected with the SOC by the OCV after the battery rest phase) when the ignition is off at rest and turned back on after a minimum of 30 min (depending on battery type and configuration).Please refer to the manufacturer's suggestion for further details.

| Independent SOC estimation error evaluation procedure (sequence)
After the initial SOC setting and stabilization time, charge/discharge tests were performed based on the selected charge/discharge evaluation pattern.After a rest period, the battery was discharged to the cut-off voltage, and the test was completed.If required, the battery should be fully charged and discharged again by the SOC amount to perform the test under the following SOC settings.Subsequently, the accuracy of SOC estimation is evaluated as follows: • Verify the rated capacity via the capacity test before the evaluation pattern ---(1) • Verify the estimated SOC of BMS when the evaluation pattern is completed (current is 0).---(2) • After the evaluation pattern is completed, verify the capacity discharged to the cut-off voltage.---(3) • After the evaluation pattern is completed via steps (1)  and ( 3), calculate the reference SOC using ((3)/ (1) × 100).---(4) • Calculate the SOC estimation error by subtracting (2)  from (4). Figure 5 illustrates an example of the voltage and current graph of a battery when estimating the SOC considering vehicle state information, such as the ignition key and battery evaluation temperature.First, an evaluation pattern was designed to reflect the load characteristics of the target application.Thereafter, the SOC of the battery for evaluation was set as the target value, and the evaluation pattern was repeatedly applied several times during discharge.The battery was internally stabilized by applying a rest time of at least 1 h after discharge to the evaluation SOC.Finally, the remaining capacity was fully discharged, and the procedure for evaluating SOC estimation accuracy was completed.

| SOC/SOH simultaneous estimation error evaluation procedure
When evaluating the SOH and SOC simultaneously, whether the SOH is to be regarded as a change in capacity or internal resistance must be decided in advance.The power pattern for simulating the vehicle operation described above can be used as the evaluation pattern when evaluating the SOH based on the amount of change in capacity, whereas the HPPC test pattern can be used when evaluating the SOH based on the amount of change in internal resistance.A BMS installed in an actual vehicle continuously monitors the battery status.It verifies the amount of charges/discharges and the number of battery cycles, as well as stores the last battery status information in a nonvolatile memory even when the ignition is turned off.However, when the BMS is used for evaluation, a pattern must be added to verify the battery's health before and after the evaluation because the charging/discharging history is unknown.Therefore, full charge and discharge patterns are added before the evaluation pattern is completed, whereas the charge pattern is added after the evaluation pattern is completed.When slow charging (standard charge) occurs after the evaluation pattern is completed, the charge pattern may appear in the form of CV control after the CC charge.The evaluation pattern can be set to be charged with different CC values per the SOC to achieve quick charging.Figure 6 shows an evaluation procedure that can simultaneously evaluate the estimation error accuracy of the SOC/SOH.Five procedures were arranged in the ranges of full charge, rest time, full discharge, SOC 80% setting, and evaluation pattern application to determine the algorithm estimation accuracy.The reference capacity was determined using the initial full charge and full discharge.Subsequently, the battery was stabilized by applying a rest time of at least an hour, and full charging was performed.In the second full charge, the SOC was set to 80%; based on the full discharge capacity, 20% of the SOC was discharged.The battery was discharged by applying an evaluation pattern load profile suitable for the target application after a rest period of 1 h or more.The error in the SOC estimation accuracy can be calculated by comparing the remaining capacity ratio after the evaluation pattern is completed to the last SOC in the BMS.In terms of the SOH, the rated capacity of the new battery must be examined based on the same battery currently used.The error can be calculated by verifying the charge/discharge capacity of the current battery through full charge-discharge before the evaluation pattern is completed and calculating the ratio of the available capacity of the current battery to the standard rated capacity based on the SOC estimation accuracy calculation method.The SOH estimation accuracy is calculated as follows:

| SOC/SOH estimation accuracy evaluation experiment
The battery module capable of determining the degree of degradation was positioned in a thermostat set to a certain temperature to conduct an evaluation for at least 6 h to set the internal temperature of the battery to the experimental temperature.All experiments were marginally set for safety and terminated immediately if they exceeded the figures provided by the manufacturer.In the case of the battery applied in this paper, in consideration of safety conditions, when the temperature of the battery module is 65°C or higher, and when the voltage is 25.7 V or higher and 14.5 V or lower, the test termination conditions are immediately applied.Before applying the evaluation pattern, full charging and discharging were performed to measure the capacity of the battery module, and a rest time of more than 1 h was applied to stabilize the internal state of the battery.The CC-CV mode was applied for full charging, the upper-end voltage was set to 25.2 V while charging in the CC mode, and the cutoff current was set to 1/20 C while charging in the CV mode.The CC mode was applied for full discharge, and the lower limit end voltage was set to 15 V.After full discharge, to set the SOC to 80%, the full charge was performed in the CC-CV mode, and 20% of the SOC based on the discharge capacity was discharged through the CC mode.Before the application of the evaluation pattern, the battery was positioned at the evaluation temperature for at least 12 h, and the internal temperature of the battery reached the evaluation temperature.After the evaluation pattern was applied, the battery was sufficiently positioned for 12 h or more at room temperature, and the SOC/SOH estimation accuracy was evaluated.This study evaluated it at room temperature ranging from 23°C to 25°C.Therefore, after applying a rest time of 1 h or more, the evaluation pattern was repeated three times, during which the remaining capacity was calculated, the reference SOC was extracted, and a full discharge was performed.Thereafter, the experiment was terminated by full charging and discharging to update the current discharging capacity of the battery.

| SOC/SOH estimation accuracy evaluation
Since the specifications of the MCU applied to the application vary, and SOC/SOH estimation needs to be performed while carrying out various functions, an appropriate algorithm should be selected.Therefore, rather than having a fixed algorithm to apply the evaluation method, no state estimation technique can assess accuracy through the evaluation method presented in this paper.This paper evaluated the estimation accuracy of SOC and SOH for two types of BMSs that are equipped with DEKFs and LSTM algorithms.The target battery cell sample was a six-series module for Hyundai IONIQ5. 27The charger and discharger used to measure the data were calibrated and certified for accuracy, and the reference values of SOC and SOH were defined using the current integration method based on the capacity measured through the discharge capacity test of the battery.In this study, the reliability of the evaluation method was the main factor in evaluating the SOC/SOH results because their reference values were defined by Ah counting, which is sensitive to sensor noise.To analyze the current sensor uncertainty of Ah counting, a Type A evaluation of the standard uncertainty can be utilized to evaluate the reference data. 28The average current measurement in the discharging mode was −55.5999A, and the measured standard deviation of the mean was 0.0084.The current sensor noise was considered negligible in Ah counting, as shown in Figure 7.In Figure 7, M denotes the mean and σ denotes the standard deviation.A new battery module (110 Ah) was placed over 435 charge/ discharge cycles; the charge/discharge pattern is shown in Figure 8.In addition, as the life cycle progressed, the capacity value of the battery was checked regularly, and the results are shown in Table 3. Subsequently, a degraded battery with a capacity of 105.215Ah was used as the sample.Based on the capacity change, the reference SOH is 95.65%.The evaluation environment for estimating the SOC/SOH was configured as shown in Figure 9.
The battery test environment temperature was set to room temperature (25°C), and the initial SOC was set to 80%.The combination of one UDDS cycle (1369 s) and one HWFET cycle (765 s) was regarded as one cycle (2134 s) in the evaluation pattern, and a repeated pattern comprising three cycles (1.85 h) was used.Figure 10 shows the onecycle power evaluation pattern of the three-cycle patterns used to evaluate the battery SOC/SOH accuracy.The experimental data used to define the reference SOC and SOH were measured using a charger and discharger directly connected to the battery.The data confirms the exact remaining capacity and allows the user to set the SOC by adjusting the current value.Additionally, voltage and current were measured in each BMS to evaluate the accuracy of the SOC and SOH estimation algorithms, and the results were calculated using the algorithm.The SOC and SOH values calculated by the BMS were passed to the charger and discharger via the CAN communication line, and the data were eventually stored.
The remaining capacity was calculated after the charge/discharge pattern was discharged to the cut-off voltage.Subsequently, a full-charge pattern was used.After the test, the capacity discharged to the cut-off voltage after the completion of the charge/discharge pattern was 44.2 Ah.The ratio of the battery capacity to the current capacity before the test was calculated.The reference SOC was 44.69%.Immediately after completing the charge/ discharge pattern, the SOC values of BMS #1 and #2 were 42.69% and 44.14%, respectively.Moreover, based on the reference SOC, the SOC estimation accuracies of BMS #1 and #2 were 2.0% and 0.55%, respectively.Figure 11 shows the estimated SOC values of the two BMS and the reference SOC values in the evaluation pattern interval.
Based on the results of the SOC/SOH estimation accuracy evaluation test, the SOH values of BMS #1 and #2 were 92.43% and 93.00%, respectively.Therefore, based on the reference SOH, the estimation accuracies of BMS #1 and #2 were 3.22% and 2.65%, respectively.Figure 12 shows the estimated SOH values of the two BMS and reference SOH values over the entire evaluation interval.Table 4 shows the evaluation results of applying the SOC/SOH accuracy evaluation method proposed in this paper to two BMS.

| CONCLUSIONS
A method for evaluating the SOC and SOH accuracies estimated by a BMS was introduced in this study.The proposed evaluation pattern can be applied to evaluate an actual vehicle under various operating conditions and environments.It is also characterized by verifying the current rated capacity and setting the initial SOC before the evaluation pattern is completed to allow the SOH algorithm to be updated based on the evaluation pattern while the operational history of the evaluation target BMS is reset.Additionally, a method for simultaneously evaluating the SOC and SOH for two BMSs used in EV battery modules was proposed.
The procedure was evaluated based on the accuracy of the SOC/SOH, and the validity of the results was confirmed.The SOC/SOH estimation accuracy of the BMS installed in the vehicle can be accurately determined by applying the proposed evaluation method.By increasing the accuracy of SOC/SOH estimation afforded by the BMS, the estimation of an EV's mileage per charge or the battery's remaining lifespan is expected to improve.
The performance of batteries is significantly influenced by various environmental factors, including temperature and load conditions.Notably, lower temperatures can lead to distinctive behavior in lithium batteries, especially at high C-rates as opposed to low C-rates.Additionally, various battery types, such as cylindrical, pouch, and prismatic, exhibit differences in characteristics such as energy density and heat generation.Consequently, there is a pressing need for an SOC/SOH estimation accuracy evaluation method capable of accounting for these multifaceted conditions.While this paper does not delve into experimental verification of the aforementioned factors, it sets the stage for future research to propose criteria for assessing the accuracy of battery state estimation.Given the absence of a clearly defined SOC/SOH estimation accuracy evaluation method, the introduction of an evaluation method that accommodates diverse load conditions and algorithms could be implemented in practical BMS to yield more precise results.

T A B L E 1 25 2 25 3 25 4 2 5 6 7
Comparison of battery state estimation and accuracy verification type./capacity estimator based on the discrete variational derivative combined with the DEKF for electric power applications KF Arbitrary CC profile (SOC) CC-CV profile (SOH) On-line battery state-of-charge estimation based on an integrated estimator KF Dynamic CC (SOC) State of charge estimation for Li-ion battery based on model from extreme learning machine KF Discharging pulse profile (SOC) Battery state-of-charge estimator using the SVM technique SVM CC-CV profile (SOC) 23 ± Combined internal resistance and state-of-charge estimation of lithium-ion battery based on extended state observer Observer CC profile (SOC) -Integrated system identification and state-of-charge estimation of battery systems Observer HPPC, USABC profile (SOC) -Flexible battery state of health and state of charge estimation using partial charging data and deep learning Deep learning Short charging profile (SOC/SOH) charge/state-of-health (SOC/SOH) evaluation equipment configuration diagram.BMS, battery management system; CAN, controller area network.SONG and SHIN | 4277

F I G U R E 3
Power pattern per mode in electric passenger vehicle (top: urban dynamometer driving schedule; bottom: highway fueleconomy test cycle).

F I G U R E 4
Hybrid pulse power characterization (HPPC) profile (left) and direct current-internal resistance calculation (right).SOC, state of charge; SOH, state of health.

F I G U R E 5
Example of procedure for evaluation of state of charge (SOC) estimation accuracy.SONG and SHIN | 4281 Verify the capacity or internal resistance (DC-IR) of a new battery identical to the subject battery before the test.---(1) • Verify the charge and discharge capacities before the test or the full charge capacity after the test.---(2) • Verify the internal resistance (DC-IR) if the evaluation pattern is an HPPC pattern.---(3) • Verify the SOH in the BMS after completing the test.---(4) • Calculate the reference SOH based on the capacity change rate after the evaluation pattern is completed via Steps (1) and (2).((2)/(1) × 100) ---(5_1) • Calculate the reference SOH based on the resistance change rate after the evaluation pattern is completed via Steps (1) and (3).(100 − ((3) − (1))/(1) × 100) ---(5_2) • The SOH estimation error is calculated as (5_x) − (4).

F I G U R E 6
Example of procedure for evaluating state-of-charge and state-of-health estimation accuracy simultaneously.(A): Full charge; (B): rest; (C): full discharge; (D): SOC set to 80%; (E): evaluation pattern.

F I G U R E 7
Graphical illustration of evaluating the standard uncertainty of noise quantity from the current sensor in discharging mode: (A) Distribution of current measurement using probability density function and (B) experimental current measurement.F I G U R E 8 Degradation pattern.SOC, state of charge.

T A B L E 3 3 F
Capacity change per cycle of IONIQ5 module.I G U R E 9 Evaluation environment for estimating state-ofcharge/state-of-health (SOC/SOH) accuracy.BMS, battery management system.F G U R E 10 One-cycle power evaluation pattern (urban dynamometer driving schedule + highway fuel-economy test cycle).F I G U R E 11 Graph of state-of-charge (SOC) estimation results.F I G U R E 12 Graph of state-of-health (SOH) estimation results.

Table 2
Abbreviations: EV, electric vehicles; HEV, hybrid electric vehicle; SOC, state of charge; SOH, state of health; SVM, support vector machine; USABC, United States Advanced Battery Consortium.
T A B L E 4 Result of SOC/SOH estimation error.: SOC, state-of-charge; SOH, state-of-health. Abbreviations