Method for determining a state of charge (SOC) of a battery

Abstract

A method for determining a state of charge (SOC) of a battery, the method comprising the steps of:-receiving (1) at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions, -determining (2 and 3), for each model, at least one estimate of the state of charge and at least one output voltage value on the basis of the at least one received parameter and at least two different models,-providing (4) an estimate of the state of charge of the battery on the basis of the determined state of charge with the lowest error rate.

Description

Technical Field

[0001] The present invention relates to a method for determining the state of charge (SOC) of a battery. The present invention also relates to a system for determining the state of charge (SOC) of a battery. Background Art

[0002] The lifetime and safety of sodium-ion batteries are very important for practical applications. However, optimal energy utilization and minimization of degradation effects are among the typical challenges faced. With challenges such as safety management, charge and discharge control, performance degradation of sodium-ion batteries, and capacity decay, the estimation of state of charge and state of health has become a hot and challenging issue. In fact, the design and implementation of diagnostic models are considered to be key to solving battery durability issues. The deployment of diagnostic schemes allows predicting and avoiding faults, assessing the state of health, estimating the state of charge, and based on this information, control and / or maintenance measures can be envisioned to ensure the continuity of battery operation. However, different battery states such as state of charge and state of health are not directly observable, which requires estimation and prediction algorithms such as diagnosis and prediction.

[0003] State of health is an important aspect of a battery management system (BMS) as it is considered a measure of lifetime. Therefore, a poor state of health estimation ultimately leads to damage to the battery and shortens its lifetime. As with other chemical-based energy storage systems, the use of the battery produces irreversible physical and chemical changes, so its performance tends to gradually deteriorate over its lifetime. Several aging protocols for sodium-ion battery aging have been tested, including calendar aging and cycle aging, and have shown an increase in internal resistance and a decrease in capacity. Therefore, the definition of the end of battery life depends on these aging indicators: capacity and resistance, the same as the degradation of lithium-ion batteries. However, these aging indicators are not measurable, so the main way to track battery aging online without interrupting the system is to estimate these indicators using diagnostic models.

[0004] The object of the present invention is to address at least one of these drawbacks. Summary of the invention

[0005] The object is achieved by a method for determining the state of charge (SOC) of a battery, the method comprising the following steps:

[0006] - receiving at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions,

[0007] - determining, based on at least one received parameter and at least two different models, at least one estimate of the state of charge and at least one output voltage value for each model,

[0008] - providing an estimate of the state of charge of the battery based on the determined state of charge having the lowest error rate.

[0009] The method of the present invention allows the determination of the state of charge (SOC) of a battery. The state of charge is defined as a percentage of the total capacity and is used to reflect the performance of the battery. The OCV-SOC function is defined by the volt value of the open circuit voltage (OCV) and the percentage value of the state of charge (SOC).

[0010] The battery referred to in the method is a battery whose OCV-SOC function is nonlinear, that is, the OCV-SOC function includes at least two "plateaus". The term "plateau" actually means that the function includes two regions that follow different curves of two different affine functions. For example, for:

[0011] Na 3 V 2 (PO 4 ) 2 F 3

[0012] Na 2 CoFe(CN) 6

[0013] Na 0.6 Ni 0.22 Al 0.11 Mn 0.66 O 2

[0014] Na 0.6 Ni 0.45 Zn 0.05 Mn 0.4 Ti 0.1 O 2

[0015] R-Na 1.92 Fe[Fe(CN) 6

[0016] Na 2 VTi(PO 4 ) 3

[0017] P 2 -Na 2 / 3 Ni 1 / 3 Mn 2 / 3 O 2

[0018] Furthermore, the at least two models are used to estimate the state of charge in parallel, which provides good accuracy for the state of charge estimation. Each model is able to estimate the state of charge of a specific area well. The at least two models are compiled in parallel, and the model with the lowest voltage estimate gives a lower accuracy of the state of charge. Comparison of at least two error rates is allowed to select the most accurate estimate.

[0019] The step of receiving at least one parameter may include the following steps:

[0020] - receiving at least one no-load voltage value,

[0021] - dividing the OCV-SOC function into at least two regions, each region having a different function,

[0022] - determining to which region the at least one no-load voltage value belongs,

[0023] - providing a percentage of the initial state of charge of the battery as a function of the area.

[0024] This step makes it possible to make a first estimate of the state of charge based on an area of ​​the OCV-SOC function curve. This first estimate given as a function of said area is more accurate than the estimate given as a function of the entire curve. In practice, this makes it possible to further limit the error rate when calculating the state of charge.

[0025] For at least two regions of the OCV-SOC function, the first region may be set to 0 to 3.5 volts, and the second region may be set to 3.5 volts to 4.5 volts. These values ​​correspond to the OCV values ​​in the OCV-SOC function.

[0026] At least two regions are defined based on the tension value or the percentage of the state of charge. Here, the threshold for dividing the curve of the OCV-SOC function into at least two regions is 3.5 volts (OCV value) or 40% (state of charge value). This separation occurs before the second plateau defined above.

[0027] The first of at least two different models can use an observer.

[0028] The observer used may be a Kalman filter.

[0029] The Kalman filter is used to estimate the state variables of a continuous nonlinear system which is linearized around its equilibrium point and represented using a state function. The Kalman filter gives better estimates in the linear part of the open circuit voltage (from 0 to 35% and from 45% to 100%).

[0030] A first of the at least two different models may use a sliding mode observer.

[0031] A first of the at least two different models uses an adaptive observer.

[0032] The second of the at least two different models uses a counting function. The estimation of the state of charge from 35% to 45% is difficult. Therefore, in addition to the estimation using the state function and the observer, an additional verification of the estimation of the state of charge using coulomb counting is added.

[0033] The state of charge may be updated at each iteration of at least two models.

[0034] Therefore, the estimate of the state of charge is updated at each iteration.

[0035] According to another aspect of the present invention, a system for determining the state of charge (SOC) of a battery is provided, the system comprising:

[0036] - Computation module;

[0037] - one or more processors;

[0038] - one or more computer-readable media storing instructions that, when executed by the one or more processors, cause the system to:

[0039] - receiving at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions,

[0040] - determining, based on at least one received parameter and at least two different models, at least one estimate of the state of charge and at least one output voltage value for each model,

[0041] - providing an estimate of the state of charge of the battery based on the determined state of charge having the lowest error rate.

[0042] The system is named "state of charge module" and is configured to estimate and / or calculate the state of charge of the battery.

[0043] The one or more computer readable media may be configured to:

[0044] - receiving at least one no-load voltage value,

[0045] - dividing the OCV-SOC function into at least two regions, each region having a different function,

[0046] - determining to which region the at least one no-load voltage value belongs,

[0047] - providing a percentage of the initial state of charge of the battery as a function of said area.

[0048] For the at least two regions of the OCV-SOC function, the first region is set to 0 to 3.5 volts, and the second region may be set to 3.5 volts to 4.5 volts.

[0049] A first of the at least two different models may use an observer.

[0050] The observer used may be a Kalman filter.

[0051] A first of the at least two different models may use a sliding mode observer.

[0052] A first of the at least two different models may use an adaptive observer.

[0053] A second of the at least two different models may use a counting function.

[0054] The counting function used may be coulomb counting.

[0055] The one or more computer-readable media may be configured to update the state of charge at each iteration of at least two models.

[0056] According to another aspect of the present invention, one or more non-transitory computer-readable media are provided, which store instructions, which, when executed by one or more processors, cause the system to perform a method for determining the state of charge (SOC) of a battery, the method comprising the following steps:

[0057] - receiving at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions,

[0058] - determining, based on at least one received parameter and at least two different models, at least one estimate of the state of charge and at least one output voltage value for each model,

[0059] - providing an estimate of the state of charge of the battery based on the determined state of charge having the lowest error rate.

[0060] BRIEF DESCRIPTION OF THE DRAWINGS AND DETAILED DESCRIPTION

[0061] Other advantages and characteristics of the invention will emerge from a reading of the detailed description of non-limiting embodiments and examples and from the following drawings:

[0062] Figure 1A battery management system (BMS) according to the present invention is shown.

[0063] Figure 2 A calculation model according to the invention is described.

[0064] Figure 3 A model of a SOC module according to the present invention is shown.

[0065] Figure 4a Curves depicting the OVC-SOC function for Li-ion NMC.

[0066] Figure 4b The graph depicts the OVC-SOC function of the lithium-ion LFP.

[0067] Figure 5 Describes the sodium ion NVPF / HC ((Na 3 V 2 (PO 4 ) 2 F 3 )-(Hard Carbon))'s OVC-SOC function curve.

[0068] Figure 6 A model using a Kalman filter according to the present invention is described.

[0069] Figure 7 A model using coulomb counting according to the present invention is described.

[0070] Figure 8a An NVPF-HC electrical model according to the present invention is described.

[0071] Figure 8b The electrochemical impedance spectroscopy (EIS) test results of the circuit are described.

[0072] These embodiments are by no means limiting and variants of the invention may be considered which consist only of a selection of the features described or illustrated, separated from other features described or illustrated subsequently (even if this selection is isolated in a sentence containing these other features), if the selection of the features is sufficient to confer a technical advantage or to distinguish the invention from the state of the art. Such a selection includes at least one functionally preferred feature without structural details and / or only part of the structural details, if this part alone is sufficient to confer a technical advantage or to distinguish the invention from the prior art.

[0073] First reference Figure 1A battery management system (BMS) is described. In a typical sodium-ion battery implemented in an actual application, coupled to external communication 101 is a communication link 102. The communication link 102 can be used to obtain configuration updates from an external data source and convey different information about the sodium-ion battery to a user, such as: health status, state of charge, and functional status.

[0074] The communication link 103 can be used to provide communication between each cell module and the battery management system. The data extraction 104 can read the different data measured. Using this data, the calculation model 105 calculates the state of charge. The calculation model 105 includes a battery model and an estimation and calculation module. The output state of charge is transmitted to the balancing algorithm 109 using the communication link 106. The balancing algorithm 109 allows balancing different sodium-ion cell modules. Cell module balancing is a method of compensating for these weaker cells by balancing the charge on all cell modules in the chain, thereby extending the battery life. Using a good balancing algorithm 109 can extend the life of the sodium-ion battery.

[0075] Sodium-ion batteries have important advantages over other technologies. Sodium-ion batteries can be fully discharged until 0V. In case of an abnormal event, the safety protocol 110 is activated using the communication link 108. The safety protocol 110 also controls the semiconductor device, which in this case fully discharges all cells or all batteries to achieve 0V. Therefore, the cell polarity does not present any potential. In this way, the battery can be safely disassembled and transported. We note that even in the case of a fault alarm, the sodium-ion battery can be charged normally without problems.

[0076] Using the proposed technique to monitor different cells of Na-ion batteries, aged cells or thermal runaway events can be easily detected, which helps in preventive maintenance.

[0077] according to Figure 2 , the calculation model 105 includes:

[0078] -NVPF / HC((Na 3 V 2 (PO 4 ) 2 F 3 )-(Hard Carbon))Model Module 10: The cell model estimates the cell voltage based on current and temperature measurements.

[0079] A state of health (SOH) module 11 for estimating different parameters of the state of health of the battery based on the measurements and the voltage estimation errors.

[0080] A state of charge (SOC) module 12 for estimating the state of charge based on the measurements and the state of health parameters.

[0081] The input parameters of the NVPF / HC model module 10 are: Vcell corresponding to the cell voltage, Tcell corresponding to the cell surface temperature, and Tamb corresponding to the operating temperature. In addition, V_pack corresponds to the voltage of all battery packs, SOH_R corresponds to the health state based on resistance, and SOH_Q corresponds to the health state based on capacity.

[0082] Based on measurements of current, operating temperature, and open circuit voltage (i.e., the voltage of the cell before it is connected to a load or charger), the NVPF-HC model module 10 estimates the cell voltage. Figure 8a As shown, the NVPF-HC model module 10 is based on an RC circuit. Uoc is the open circuit voltage. i is the current (positive electrode for charging, negative electrode for discharging). V- is the negative terminal of the battery. V+ is the positive terminal of the battery. Rs is the equivalent series resistance, which represents all ohmic contributions of the battery cell (contact resistance, electron migration in the collector and electrodes, ion migration in the electrolyte). Rsurf is the surface resistance, corresponding to the voltage drop at the interface between the active material particles and the electrolyte. This parameter is associated with the charge transfer of the two electrodes and any passivation layer present on their surfaces. Csurf corresponds to the "surface time constant" τsurf=Rsurf×Csurf, which is used to approximate the fast dynamics (usually less than 1 second) associated with the interface phenomena of the two electrodes (charge transfer, double layer capacitance, dynamics that may be associated with the passivation layer with capacitance and / or diffusion effects). Vsurf is the voltage on the RsurfCsurf circuit. Zd is the diffusion impedance, which combines the overvoltage associated with the atomic diffusion phenomena in the active material particles of each electrode and the ion diffusion phenomena in the electrolyte. Vd is the voltage across the impedance Zd.

[0083] V cell =V + -V -

[0084]

[0085] V cell =U OC +R s· i+V surf +V d

[0086] The open circuit voltage depends on the state of charge, temperature and operating stage (charging or discharging). Based on the values ​​of the open circuit voltage and the operating temperature, a parameter is specified. The parameter corresponding to the initial state of charge (SOC0) is specified. This initial value is important for the estimation of the state of charge and is considered as an input to the state of charge estimation module. The result of the state of charge estimation is the input to the model. At each iteration, the estimated state of charge is combined with the operating temperature to calculate the open circuit voltage.

[0087] refer to Figure 3 , describes a model of the SOC module 12 according to the present invention. In this embodiment, the battery is a sodium-ion NVPF-HC battery. The method is applied by the state of charge module 12. The method comprises the following steps:

[0088] - receiving (1) at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation zone (pace zone), the OCV-SOC function variation being divided into at least two zones,

[0089] - determining (2 and 3) for each model at least one estimate of the state of charge and at least one output voltage value based on at least one received parameter and at least two different models,

[0090] - Based on the determined state of charge having the lowest error rate, providing (4) an estimate of the state of charge of the battery.

[0091] State of charge (SOC) is a battery's charge level compared to its capacity. The formula is as follows:

[0092]

[0093] η is the Coulomb coefficient, i(t) is the current (positive for charging and negative for discharging), SOC 0 is a percentage of the battery's initial state of charge. Q is considered to be the actual battery's available capacity under given aging conditions. Therefore, Q is updated after each diagnostic process and is considered equal to the actual capacity provided by the battery at each charge / discharge measurement. The capacity also decreases as the battery ages.

[0094] according to Figure 8a The NVPF-HC electrical model, we get:

[0095]

[0096] And, the battery voltage is

[0097] V cell =U OC +R s·i + V surf +V d

[0098] In the battery voltage equation, we can only measure the battery voltage Vcell and the current i. The parameters (R s 、R surf 、C surf and Z d ) of the resistor-capacitor (RC) circuit are determined using electrochemical impedance spectroscopy (EIS) (as Figure 8b shown) and galvanostatic intermittent titration technique (GITT) tests at different states of charge and temperatures. This stage corresponds to the calibration stage of the method at the beginning of the battery life. These data are initial data, and even if they are not precise, they will be updated at each iteration of the method. Therefore, the cell model will converge to the measured voltage value, and the error will decrease to reach the possible minimum. Therefore, in the voltage equation V cell , the last parameter to be defined is the open-circuit voltage U oc . U oc depends on the state of charge, temperature, and operating stage (charging or discharging).

[0099] The NVPF-HC model module 10 provides (1) at least one parameter corresponding to the initial state of charge percentage of the battery as an input parameter to the SOC module 12.

[0100] As Figure 4a and 4b shown, the open-circuit voltage of lithium ions is a linear function, presenting only one stable plateau P0. This plateau P0 is linear in the case of NMC ( Figure 4a ), and stable in the case of LFP ( Figure 4b ). The open-circuit voltage configuration of lithium ions helps to represent the evolution of the open-circuit voltage using some assumptions. The following assumptions are usually used for lithium-ion batteries:

[0101]

[0102] U OC (SOC) = a.SOC + b

[0103] This assumption cannot be applied to the embodiment of sodium-ion NVPF-HC. For sodium-ion NVPF-HC, the evolution of the open-circuit voltage is not linear, as we can see in Figure 5 . The evolution of the open-circuit voltage contains two stable plateaus P1 and P2: the first plateau P1 has a state of charge from 0 to 40%, and the second plateau P2 has a state of charge from 40% to 100%. Therefore, the first region is set to 0 to 3.5 volts (0 to 40%), and the second region is set to 3.5 V to 4.5 V (40% to 100%).

[0104] The OCV-SOC function is defined based on the region where the characteristic is close to linear (approximate). For NVPF-HC cells, the open circuit voltage depends on the state of charge and temperature. The two OCV-SOC functions (corresponding to the P1 and P2 regions) correspond to two regions, which are defined as follows:

[0105] {U OC (SOC) = a 1· SOC+b 1 If SOC≤40%

[0106] U OC (SOC) = a 2· SOC+b 2 if SOC>40%

[0107] This subdivision of the OCV-SOC function, depending on the state-of-charge interval, can be applied to all active materials with different plateaus.

[0108] The measurable states of the system are the cell voltage Vcell, the current i, the cell surface temperature Tcell and the operating temperature Tamb.

[0109] According to the method, at least one estimate of the state of charge and at least one output voltage value are calculated based on the percentage of the initial state of charge provided by the NVPF-HC model module 10 and based on at least two different models. The two calculations are obtained based on at least two different models. The calculations of the at least two models are performed simultaneously in parallel.

[0110] In the proposed embodiment, the first model consists of the state of charge estimation using an observer, more specifically an extended Kalman observer. For the state of charge estimation of the NVPF-HC sodium-ion battery, the state function is:

[0111]

[0112] V surf,k =α.V surf,k-1 -β.i k

[0113]

[0114] in and

[0115] Based on the state function, the state vector of the Kalman observer is Where V 2 =V surf +V d Y k =V k For output,

[0116] Assume that the noise is Gaussian white noise. In fact, the model must combine all deterministic system information and the system variables are continuous. In the model of the present invention, the main goal of the extended Kalman is to estimate the state of charge representation. The parameters to be estimated by the observer and the input and output Y are as follows:

[0117] Given that

[0118] Therefore, X k+1 =AX k +Bi k , where A = [1 α] and

[0119] The input is current i k . And the output Y k =V k =a.SOC k +b+V surf，k +Ri k

[0120] Y k =[a 1].X k +Ri k +b

[0121] Where D = [a 1]

[0122] If SOC≤40%, then a=a 1 And b=b 1

[0123] If SOC>40%, then a=a 2 And b=b 2

[0124] For the estimation model using the Kalman filter, the model estimates 2 or predicts 21 the state of charge. All details of the Kalman filter model are as follows Figure 6 As shown. The Kalman filter in a discrete context is a recursive estimator. This means that to predict 21 the current state, only the previous state 20 and the current measurement need to be estimated. Therefore, no historical observations and predictions are required. The initial health state is the input data for the model. The model outputs an estimate of the state of charge. Based on the estimate of the state of charge, the output voltage is calculated. The model repeats the steps at each time step or iteration. Therefore, the estimate of the state of charge and the value of the output voltage are updated 22 at each iteration.

[0125] The state of the observer is represented by two variables:

[0126] The state estimate at time k

[0127] P k-1 / k-1 : Error covariance matrix (a measure of the accuracy of the estimated state).

[0128] The Kalman filter has two distinct phases: prediction 20 and update 21. The prediction step 21 uses the estimated state at the previous time to produce an estimate of the current state. In the update step 22, the predicted state is corrected using the observation at the current time to obtain a more accurate estimate. In other embodiments, the first model consists of a state of charge estimate using a sliding mode observer, an adaptive observer, or a fuzzy observer.

[0129] In addition to the first model, a second estimate and a second output voltage of the first model are calculated in parallel by a second model. In the proposed embodiment, the second model uses a coulomb counting model. The coulomb counting model is as follows Figure 7 shown.

[0130] Regarding the coulomb counting model, this model calculates the number of coulombs that charge / discharge a battery. The amount of charge transferred by the current is measured in coulombs and is given by:

[0131] C c =i×dt

[0132] Among them, C c is the amount of charge transferred, and dt is the time the current flows, in seconds.

[0133] To obtain the state of charge transferred during dt, the calculated charge is compared with the total capacity of the cell taking into account the Coulomb coefficient η. The sum of the SOC transferred at each iteration gives the total SOC:

[0134]

[0135] The same inputs to the Kalman filter, such as current and voltage measurements, are considered in the Coulomb counting model.

[0136] Therefore, the output of the second model is also an estimate of the state of charge and output voltage 3. Based on the output voltage of each model, an error rate is calculated. The error rate is calculated by comparing the output voltage of the model with the measured output voltage. When two error rates are calculated (one for each output voltage), they are compared to each other 4. The estimate of the state of charge with the lowest error rate corresponds to the state of charge of the battery.

[0137] Typically, at least one of the means of the device according to the above invention is a technical means, and preferably, each of the means of the device according to the above invention is a technical means.

[0138] Typically, each device of the apparatus according to the above invention may comprise at least one computer, central unit or calculation unit, analog electronic circuit (preferably dedicated electronic circuit), digital electronic circuit (preferably dedicated digital electronic circuit) and / or microprocessor.

[0139] Of course, the invention is not limited to the examples that have just been described and many adjustments may be made to these examples without exceeding the scope of the invention.

[0140] Of course, the different features, forms, variants and embodiments of the present invention can be associated with each other in various combinations to the extent that they are mutually compatible or not mutually exclusive. In particular, all variants and embodiments described above can be combined with each other.

Claims

1. A method for determining the state of charge (SOC) of a battery, the method The following steps are involved: - receiving (1) at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions, - determining (2 and 3) for each model at least one estimate of the state of charge and at least one output voltage value based on at least one received parameter and at least two different models, - Based on the determined state of charge having the lowest error rate, providing (4) an estimate of the state of charge of the battery.

2. The method according to claim 1, wherein the step of receiving at least one parameter The following steps are involved: - receiving at least one no-load voltage value, - dividing the OCV-SOC function into at least two regions, each region having a different function, - determining to which region the at least one no-load voltage value belongs, - providing a percentage of the initial state of charge of the battery as a function of the area.

3. The method according to any one of claims 1 to 2, wherein for the at least two regions of the OCV-SOC function, a first region is set to 0 to 3.5 volts and a second region is set to 3.5 volts to 4.5 volts.

4. The method of claim 1, wherein a first of the at least two different models uses an observer. The method according to claim 4 , wherein the observer used is a Kalman filter.

6. The method of claim 1, wherein a first of the at least two different models uses a sliding mode observer.

7. The method of claim 1, wherein a first of the at least two different models uses an adaptive observer.

8. The method according to any one of claims 1 to 7, wherein a second one of the at least two different models uses a counting function.

9. The method of claim 8, wherein the counting function used is coulomb counting.

10. The method according to any one of claims 1 to 8, wherein the state of charge is updated at each iteration of at least two models.

11. A system for determining the state of charge (SOC) of a battery, the system include: - Computation module, - one or more processors; - one or more computer-readable media storing instructions that, when executed by the one or more processors, cause the system to: - receiving (1) at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions, - determining (2 and 3) for each model at least one estimate of the state of charge and at least one output voltage value based on at least one received parameter and at least two different models, - Based on the determined state of charge having the lowest error rate, providing (4) an estimate of the state of charge of the battery.

12. The system of claim 11, wherein the one or more computer-readable media are configured to: - receiving at least one no-load voltage value, - split the OCV-SOC function into at least two regions, each with a different function, - determining to which region the at least one no-load voltage value belongs, - providing a percentage of the initial state of charge of the battery as a function of the area.

13. The system of any one of claims 11 to 12, wherein for the at least two regions of the OCV-SOC function, a first region is set to 0 to 3.5 volts and a second region is set to 3.5 volts to 4.5 volts.

14. The system of claim 11, wherein a first of the at least two different models uses an observer.

15. The system of claim 14, wherein the observer used is a Kalman filter.

16. The system of claim 11, wherein a first of the at least two different models uses a sliding mode observer.

17. The system of claim 11, wherein a first of the at least two different models uses an adaptive observer.

18. The system of any one of claims 11 to 17, wherein a second of the at least two different models uses a counting function.

19. The system of claim 18, wherein the counting function used is coulomb counting.

20. The system of any one of claims 11 to 19, wherein the one or more computer-readable media are configured to update the state of charge at each iteration of at least two models.

21. One or more non-transitory computer readable media storing instructions that, when executed by one or more processors, cause a system to perform a method for determining a state of charge (SOC) of a battery, the method The following steps are involved: - receiving (1) at least one parameter corresponding to a percentage of an initial state of charge of the battery based on at least one no-load voltage value and according to an open circuit voltage (OCV)-state of charge (SOC) function variation region, the OCV-SOC function variation being divided into at least two regions, - determining (2 and 3) for each model at least one estimate of the state of charge and at least one output voltage value based on at least one received parameter and at least two different models, - Based on the determined state of charge having the lowest error rate, providing (4) an estimate of the state of charge of the battery.

Images