Controlled estimation of the state of charge of a multi-cell battery of a system

Abstract

The invention relates to a method which is implemented in a system including a multi-cell battery. The method comprises a step (10-120) of determining a predicted state of charge using an estimation algorithm employing a battery model based on estimated impedance parameters of the cells and the voltages at the terminals of each of the cells, and based on the cumulative current consumption of the battery, then determining a corrected state of charge by means of this estimation algorithm, and verifying whether the use of the battery, the estimated impedance parameters and the corrected state of charge meet requirements; if they do, a new estimated state of charge is provided that is equal to the corrected state of charge, and, if they do not, a new estimated state of charge is provided that is equal to the predicted state of charge.

Description

DESCRIPTION TITLE: CONTROLLED ESTIMATION OF THE STATE OF CHARGE OF A CELLULAR BATTERY IN A SYSTEM The present invention claims priority from French application No. 2409104 filed on 26.08.2024, the content of which (text, drawings and claims) is incorporated herein by reference. Technical field of the invention

[0001] The invention relates to systems comprising at least one cellular battery, and more specifically to the real-time estimation of the current state of charge (or SoC) of such a cellular battery. State of the art

[0002] In many systems, such as in vehicles (possibly automobiles), rechargeable batteries are used which have cells designed to store electrical energy to power at least one electrical machine (possibly a motor) and / or an electrical power supply circuit.

[0003] For example, the cells can be electrochemical (notably lithium-ion (or Li-ion) or Ni-MH or Ni-Cd). Furthermore, in the case of a vehicle, the rechargeable battery can be either a so-called "main" battery (or traction or power battery) when it is responsible for supplying electrical current to an on-board network, via a converter, and at least one electric drive unit of the powertrain (or powertrain), or a so-called "service" battery when it is of the very low voltage type (typically between 12 V and 48 V) and responsible for supplying electrical current to the on-board network in the absence of a main battery (and therefore an electric drive unit) or in place of or in addition to the main battery.

[0004] In what follows and precedes, the term "onboard network" refers to an electrical power supply network to which equipment is connected. (or electrical (or electronic) components that consume electrical energy.

[0005] Typically, a (rechargeable) cellular battery is associated with a battery case that includes means for measuring voltage / current / internal temperature and a battery calculator. The latter centralizes current measurements, voltage measurements, and internal temperature measurements (inside the cellular battery), and determines or estimates parameters of this cellular battery based on these measurements, including its internal resistance, minimum voltage, and current state of charge (SoC).

[0006] This state of charge (or SoC) is an important parameter in a vehicle with at least partially electric powertrains because it allows, in particular, for the real-time estimation of at least part of the vehicle's mileage range, but also for optimizing the use of the cellular battery in order to slow down its degradation (and therefore increase its lifespan).

[0007] At least three techniques can be implemented to estimate the SoC of a cellular battery in real time.

[0008] One technique involves estimating the SoC of a cellular battery using only the current consumption readings (in Ah) from the battery. However, relying solely on current data is not optimal due to the accumulation of errors in current sensors over time and the inability to correct capacity errors as the cellular battery ages.

[0009] A second technique involves estimating the SoC of the battery cell using an extended Kalman filter (EKF) estimation algorithm applied only to the minimum and maximum cell voltages. In this case, the battery cell voltages are composed of the open circuit voltage (OCV) and the contribution of the direct current resistance (DCR). Given aging and dispersion within the battery cell early in its life, it is rare that the minimum (or maximum) voltage leads to a minimum SoC (respectively a maximum SoC). Therefore, this second solution is not accurate and can lead to poor anticipation of the functional limitations for the voltages of the cellular battery cells.

[0010] A third technique involves estimating the state of charge (SoC) of the battery cell using an extended Kalman filter (EKF) estimation algorithm and recursive least squares (RLS) applied to an equivalent electric circuit (EEC) battery model. This model may be combined with a hysteresis model for batches of battery cells, using a sequential approach to accommodate the battery calculator's computational constraints. More precisely, this technique allows for the determination of a predicted state of charge using the estimation algorithm. This algorithm considers the battery model as a function of estimated cell impedance parameters and the voltage across each cell, as well as the cumulative current consumption of the battery cell. A corrected state of charge is then obtained using this estimation algorithm.The results obtained at each sampling period with this third technique are not always accurate because the estimation algorithm requires some time (several iterations) before converging to the correct value. Therefore, the intermediate estimates that have not yet converged must be filtered out before use.

[0011] In general, SoC estimation errors when using cellular batteries (and therefore in real time) are very difficult to avoid because they can have several origins, such as an accumulated error of the sensor measuring the current, an inaccuracy in the measurement of the cellular battery capacity (since the SoC is the ratio between the remaining charge and the total capacity of the cellular battery), an error in modeling the cellular battery, real-time estimates of certain parameters of the battery model (such as impedances), or the chemical characteristics of the cells which influence their open circuit voltage (or OCV).

[0012] Furthermore, under certain critical operating conditions, such as driving with rapid acceleration that causes very sudden changes in battery current or power over very short periods, the battery model used for SoC estimation cannot react as quickly as the actual behavior of the battery. Consequently, the EKF algorithm can become unstable and provide estimated SoCs that are not representative of reality.

[0013] Furthermore, most existing versions of EKF estimation algorithms use open-circuit voltage (OCV) to dynamically correct the SoC during the wake-up phase after a sufficiently long battery relaxation time. These SoC corrections are entirely dependent on the OCV voltages, which are themselves a function of the SoC. Therefore, when the evolution of the OCV = f(SoC) curve is significant almost everywhere, the SoC correction is qualitative. However, when the OCV = f(SoC) curve has a significant, relatively flat portion, the SoC correction is poor (this is particularly true for LFP (Lithium Iron Phosphate or LiFePO4) or LS (Lithium Sulfur) cell types). It is therefore very difficult to use voltage information to correct the SoC when an error occurs.

[0014] The invention is therefore intended, in particular, to improve the situation. Presentation of the invention

[0015] In particular, it proposes for this purpose an estimation method, on the one hand, intended to estimate a state of charge of a rechargeable battery comprising cells and equipping a system, and, on the other hand, comprising a step in which a predicted state of charge is determined by means of a chosen estimation algorithm and using a battery model as a function of estimated impedance parameters of the cells and voltages across each of the cells, and as a function of a cumulative current consumption by the rechargeable battery, then a corrected state of charge is determined by means of this estimation algorithm.

[0016] This estimation method is characterized by the fact that in its step it is checked whether a current use of the rechargeable battery satisfies a normality criterion, whether the estimated impedance parameters satisfy a chosen reliability criterion, and whether the corrected state of charge satisfies a chosen robustness criterion, then, if all (three) criteria are satisfied, a new estimated state of charge equal to this corrected state of charge is provided, or, in the event of non-satisfaction of at least one of the criteria, a new estimated state of charge equal to the predicted state of charge is provided.

[0017] Thanks to this increased robustness of the state of charge estimates, we are guaranteed to obtain more accurate estimates of cellular battery parameters and to avoid state of charge divergence (and therefore reduce battery failures).

[0018] The estimation method according to the invention may include other features which may be taken separately or in combination, and in particular:

[0019] - in its step, the estimation algorithm can use an extended Kalman filter (or EKF);

[0020] - in its stage, the battery model can be an electrical circuit model of the so-called n-RC type, with n > 1;

[0021] - in its stage, the normality criterion can be a variation over a chosen period of a current flowing in the rechargeable battery or of an electrical power delivered or received by the rechargeable battery that is less than a chosen threshold;

[0022] - in its stage, the reliability criterion can be belonging to a first range of chosen values ​​of each of the estimated impedance parameters;

[0023] - in its stage, the robustness criterion can be belonging to a second interval of chosen values ​​of the corrected load state;

[0024] - in its step, after a complete recharge of the rechargeable battery, the new estimated state of charge can be corrected based on a target state of charge that is a function of the complete recharge and a state of estimated maximum charge of one of the cells, when current values ​​of at least two selected parameters of the rechargeable battery and a charging potential current used during full charging respectively meet normality criteria.

[0025] The invention also proposes a computer program product comprising a set of instructions which, when executed by processing means, is suitable for implementing an estimation method of the type presented above, in a system comprising a rechargeable battery including cells, to estimate a state of charge of the rechargeable battery.

[0026] The invention also proposes an estimation device, on the one hand, suitable for being part of a system comprising a rechargeable battery including cells, and, on the other hand, comprising at least one processor and at least one memory arranged to perform the operations of determining a predicted state of charge by means of a chosen estimation algorithm and using a battery model as a function of estimated impedance parameters of the cells and voltages across each of the cells, and as a function of a cumulative current consumption by the rechargeable battery, and then determining a corrected state of charge by means of this estimation algorithm.

[0027] This estimation device is characterized by the fact that its processor and memory are also arranged to perform the operations of checking whether a current use of the rechargeable battery meets a normality criterion, whether the estimated impedance parameters meet a chosen reliability criterion, and whether the corrected state of charge meets a chosen robustness criterion, then, if all these (three) criteria are met, to provide a new estimated state of charge equal to the corrected state of charge, or, in the event of non-satisfaction of at least one of the criteria, to provide a new estimated state of charge equal to the predicted state of charge.

[0028] The invention also proposes a system comprising, on the one hand, a rechargeable battery comprising cells, and, on the other hand, an estimation device of the type of that presented above.

[0029] For example, this system could be a vehicle, possibly of the automobile type. Brief description of the figures

[0030] Other features and advantages of the invention will become apparent upon examination of the detailed description below, and the accompanying drawings, in which:

[0031] [Fig. 1] schematically and functionally illustrates an example of an embodiment of a vehicle comprising a powertrain with an electric drive unit powered by a cellular battery associated with a battery computer, and an estimation device according to the invention,

[0032] [Fig. 2] schematically and functionally illustrates an example of an embodiment of a battery calculator comprising an estimation device according to the invention, and

[0033] [Fig. 3] schematically illustrates an example of an algorithm implementing an estimation method according to the invention. Detailed description of the invention

[0034] The invention aims in particular to provide an estimation method, and an associated DE estimation device, intended to allow the estimation of the current state of charge (or SoC (State of Charge)) of a BC cellular battery of an S system.

[0035] In what follows, system S is considered, by way of non-limiting example, to be a motor vehicle, such as a car, as illustrated in Figure 1. However, the invention is not limited to this type of system. It relates to any type of system comprising at least one cellular battery. Thus, it relates to vehicles (land, sea (or river), and air), mobile machinery (including those that perform a lifting function), electronic devices (possibly household appliances and / or possibly mobile), fixed or stationary installations (possibly industrial), as by For example, electrical power supply installations and buildings. As a purely illustrative example, the BC cellular battery of an S system can be connected to a renewable energy source (in particular photovoltaic or wind).

[0036] Furthermore, in what follows, we consider, as a non-limiting example, that system S (here a vehicle) comprises a powertrain (or PWM) of the all-electric type (and therefore whose propulsion is provided exclusively by at least one electric motor). However, the PWM could be of the hybrid type (thermal and electric) or purely thermal.

[0037] Furthermore, in the following, the BC cellular battery is considered, as a non-limiting example, to be a primary (or traction or power) battery capable of supplying electrical energy to at least one electric motor. However, the cellular battery under consideration could also be a service battery (possibly rechargeable via an inverter powered by a primary battery).

[0038] Figure 1 schematically represents a system S (here a vehicle) comprising an estimation device DE according to the invention and a transmission chain with electric GMP (and therefore with electric motor machine MME), an on-board network RB, a power supply group comprising a service battery BS and (here) a converter CV associated with a cellular battery BC (itself associated with a battery computer CB).

[0039] The RB on-board network is an electrical power supply network to which electrical (or electronic) equipment (or components) that consume electrical energy are coupled.

[0040] The service battery BS is responsible for supplying electrical power to the onboard electrical system RB, supplementing that supplied by the CV converter powered by the cellular battery BC, and sometimes replacing this CV converter (particularly when the engine is asleep and the CV converter is inactive). For example, this service battery BS can It must be arranged as a very low voltage battery (typically 12 V, 24 V, or 48 V). It is rechargeable, at least by the CV converter. In the following, for the sake of non-limiting example, the BS service battery is considered to be a 12 V lithium-ion type.

[0041] The transmission system has a powertrain that is purely electric, and therefore includes, in particular, an electric drive machine (MME), a drive shaft (AM), and a transmission shaft (AT). Here, "electric drive machine" refers to an electric machine arranged to provide or recover torque to move the system (S) (in this case, a vehicle). The operation of the powertrain is monitored by a control unit (CS).

[0042] The electric drive unit MME (here, an electric motor) is coupled to the battery cell BC to receive electrical power and, potentially, to recharge the battery cell BC (for example, during regenerative braking). It is coupled to the drive shaft AM to provide torque through rotation. This drive shaft AM is coupled to a reduction gear RD, which is also coupled to the transmission shaft AT. AT is itself coupled to a first set of wheels T1, preferably via a differential D1.

[0043] This first train T1 is located here in the front part PW of the vehicle S. But in a variant this first train T1 could be the one which is here referenced T2 and which is located in the rear part PRV of the vehicle S.

[0044] The MME driving machine is, here, also coupled to the CV converter which is also indirectly coupled to the BS service battery, in particular to recharge it with electrical energy from the BC cellular battery and converted.

[0045] This CV converter is an electrically coupled current converter, here, for example, to a CN charging connector of vehicle S. It is also responsible for supplying the RB on-board electrical network with electrical energy from the BC cellular battery, converted when the engine is running or when the engine is asleep but vehicle S is not. is in a phase of recharging its cellular battery BC, in addition to ensuring the recharging of the service battery BS.

[0046] For example, a BC cellular battery can include CE electrochemical energy storage cells. Each CE cell (electrochemical energy storage cell) can also be lithium-ion (or Li-ion), but this is not mandatory. It could be Ni-MH or Ni-Cd, for instance. Similarly, a BC cellular battery can be low-voltage (typically 450V, 600V, or 800V, for example), but it could also be medium-voltage or high-voltage.

[0047] As illustrated (though not exhaustively) in Figure 1, CE cells can be part of MC modules that are coupled together, for example in series, within the BC battery cell. Here, an "MC module" is defined as a group of at least one CE cell. When an MC module comprises several CE cells, these cells (CE) can be connected together in series and / or in parallel.

[0048] It should also be noted that the BC cellular battery is associated with a BB battery housing which includes, among other things, means for measuring voltage, current, and internal temperature (not shown) and the CB battery calculator. The latter (CB) centralizes current measurements, determined voltage measurements (including the initial uc1 voltages across the individual CE cells), the current flowing through the BC cellular battery, the open-circuit voltage (OCV) of the BC cellular battery, and internal temperature measurements (including those specific to each individual CE cell). Furthermore, the CB battery calculator can estimate parameters of the BC cellular battery based on these measurements, including its internal resistance, minimum voltage, and current capacity.

[0049] It should also be noted that in the example illustrated (non-exhaustively) in Figure 1, the vehicle S also includes a distribution box BD to which are coupled the auxiliary battery BS, the CV converter and the RB on-board network. This BD distribution box is responsible for distributing into the RB on-board network the electrical energy which is produced by the CV converter or stored in the BS auxiliary battery, for the supply of electrical components (or equipment) coupled to the RB on-board network, according to power requests received (in particular from the CS supervision computer of the GMP).

[0050] As mentioned above, the invention notably proposes an estimation method intended to allow the estimation of the SOCBR state of charge of the BC cellular battery of the S system (here a vehicle).

[0051] This estimation method can be implemented at least partially by the DE estimation device (illustrated at least partially in Figures 1 and 2), which comprises at least one PR1 processor, for example a digital signal processor (DSP), and at least one MD memory. This DE estimation device can therefore be implemented as a combination of electrical or electronic circuits or components (or "hardware") and software modules. For example, it could be a microcontroller.

[0052] The MD memory is random access memory (RAM) to store instructions for the PR1 processor to implement at least part of the estimation process. The PR1 processor may include integrated circuits (or printed circuit boards), or several integrated circuits (or printed circuit boards) connected by wired or wireless connections. An integrated circuit (or printed circuit board) is defined as any type of device capable of performing at least one electrical or electronic operation.

[0053] In the example illustrated (but not limited to) in Figures 1 and 2, the DE estimation device is part of the CB battery computer. However, this is not mandatory. The DE estimation device could comprise its own dedicated computer, which would then be coupled to the CB battery computer, or it could be part of another computer embedded in the S system, for example.

[0054] As illustrated, but not limited to, in Figure 3, the (estimation) method according to the invention comprises a step 10-120 which is put into works at least every time the vehicle V is used (and in particular every time its GMP is running) and an estimate of the state of charge SOCBC is requested, for example by the battery computer CB (but this could be requested by the supervisory computer CS or by a server that can access the system S via radio waves).

[0055] This step 10-120 includes a substep 10 in which one (for example the estimation device DE) begins by determining a predicted state of charge ecp by means of a chosen estimation algorithm and using a battery model as a function of estimated impedance parameters pie of the cells CE and voltages across each of the cells, and as a function of a cumulative current consumption (in Ah) by the cell battery BC.

[0056] This step 10-120 also includes a substep 80 in which a corrected state of charge ecc is determined by means of this estimation algorithm using the predicted state of charge ecp (determined in substep 10).

[0057] Furthermore, in step 10-120, one (for example, the DE estimation device) checks, firstly, whether the current use of the cellular battery BC meets a normality criterion, secondly, whether the estimated impedance parameters pie meet a chosen reliability criterion, and thirdly, whether the corrected state of charge ecc meets a chosen robustness criterion.

[0058] Then, if all three criteria mentioned above are met, step 10-120 also includes a substep 110 in which a new estimated state of charge SOCBR (for example, the DE estimation device) is provided, which is equal to the corrected state of charge ecc. The overall estimation of the new state of charge SOCBR is considered robust because it has successfully passed the three checks (by satisfying the three criteria).

[0059] However, if at least one of these three criteria is not met, step 10-120 also includes a substep 40 in which a new state of charge is provided (for example, by the DE estimation device). The estimated SOCBR is equal to the predicted state of charge ecp. The overall estimate of the new state of charge SOCBR is considered unrobust because at least one of the three checks was not passed successfully, and therefore the corrected state of charge ecc cannot be retained as the new state of charge SOCBC, which necessitates using the predicted state of charge ecp instead.

[0060] The invention offers several advantages, including:

[0061] - not taking into account the corrected state of charge ecc in the presence of abnormal use of the cellular battery BC (such as during a very high current or power dynamic) makes it possible to avoid the divergence of the state of charge SOCBC, and therefore to reduce battery failures;

[0062] - the guarantee of obtaining more accurate estimates of BC cellular battery parameters (such as the state of health (SoH)), due to the increased robustness of SOCBC state of charge estimates,

[0063] - the possibility of better matching the specific usage conditions of drivers, thanks to usage filtering;

[0064] - It allows for automatic verification of the activation conditions of the estimation algorithm,

[0065] - it allows for a more complete correction of SOCBC state of charge estimation errors,

[0066] - it helps to avoid unstable estimates and therefore to guarantee the stability of the SOCBR state of charge in all cases of use of the system (and in particular when it is a vehicle).

[0067] For example, in step 10-120, the estimation algorithm that is chosen may be a recursive algorithm using an extended Kalman filter (or EKF) with the chosen battery model.

[0068] In this case, and as illustrated non-exhaustively in Figure 3, step 10-120 may also include a substep 50 in which one (for example, the DE estimation device) can determine at each Sampling period k the voltage UBR(K) across the terminals of the cellular battery BC. Then, in a substep 60 of step 10-120 one (for example the estimation device DE) can predict an error u err(k) on this determined voltage UBR(K). Then, in a substep 70 of step 10-120, one (for example, the estimation device DE) can determine a covariance matrix P and a correction gain K, taking into account the update of the covariance matrix resulting from the determination of the corrected load state ecc(k - 1) during the previous sampling period (k - 1). Then, in substep 80 of step 10-120, one (for example, the estimation device DE) determines the corrected load state ecc(k) as a function of the predicted load state ecp(k) and the covariance matrix P and correction gain K determined in substep 70 (for example, ecc(k) = ecp(k) + K*Uerr(k)). Then, we update the covariance matrix P and return to perform substep 10 after incrementing the index k by one unit.

[0069] For example, in step 10-120, the battery model chosen (and used to determine the predicted state of charge ecp and the corrected state of charge ecc) can be an electrical circuit model of the so-called n-RC type (for each CE cell), with n > 1, well known to those skilled in the art. Thus, one can use a 1-RC equivalent electrical circuit model or, to improve accuracy, a 2-RC equivalent electrical circuit model, for example. A simple hysteresis model can also be added to an n-RC equivalent electrical circuit model (with n > 1) to increase the accuracy of the state of charge estimates, particularly in the presence of flat SoC = f(OCV) curves around 90% to 30% of SoC, for example, for LFP (Lithium Iron Phosphate or (LiFePO4)) type cells.

[0070] It is recalled that in an equivalent 1-RC electrical circuit model, the CE cell is electrically represented by impedance parameters (R0, R1, C1), where R0 is a resistor connected in series with a set including the resistor R1 connected in parallel with the capacitor C1. For example, these impedance parameters can be estimated for each CE cell using a Forgetting Factor Recursive Least Squares (FFRLS) method. of the forgetting factor). Two states SoC(k) and llp(k) can then be estimated with the following state equations:

[0071] SoC(k+1) = SoC(k) + l(k)*dt / capacity, where l(k) is the current consumed by the cellular battery BC, and

[0072] Up(k+1 ) = [Up(k)*(1 - (dt / (R1 *C1 )))] - [R1 *(l(k)*dt) / (R1 *C1 )].

[0073] As illustrated (though not exhaustively) in Figure 3, the verification of the normality criterion by the ongoing use of the BC cellular battery can be performed immediately after substep 10, which determines the predicted state of charge (ecp). Indeed, determining the corrected state of charge (ecc) is considered unnecessary if the current use of the BC cellular battery is abnormal.

[0074] But in an alternative embodiment (not illustrated) the verification of the normality criterion could be done after substep 80 of determination of the corrected load state ecc.

[0075] Also, for example, the normality criterion can be a variation over a chosen time period (dt) of the current (l(k)) flowing in the cellular battery BC or of the electrical power (P(t)) delivered or received by the cellular battery BC less than a chosen threshold.

[0076] In this case, as illustrated (non-exhaustively) in Figure 3, step 10-120 may also include a substep 20 in which the variation over a chosen period of the current flowing through the cell battery BC or the electrical power delivered or received by the cell battery BC can be determined (for example, by the DE estimation device). Then, in a substep 30 of step 10-120, this determined variation can be compared (for example, by the DE estimation device) to the chosen threshold. If the determined variation is less than the chosen threshold, the current use of the cell battery BC is normal, and therefore the corrected state of charge (ecc) can be determined. Conversely, if the determined variation is greater than or equal to the chosen threshold, the current use of the cell battery BC is abnormal, and therefore substep 40 is performed.

[0077] Also, for example, the reliability criterion can be membership in a first range of chosen values ​​of each of the estimated pie impedance parameters for the different CE cells.

[0078] As illustrated (though not exhaustively) in Figure 3, the reliability criterion can be verified using the estimated impedance parameters pi in substep 90 of step 10-120, immediately following substep 80 for determining the corrected state of charge ecc(k). This is because these estimated impedance parameters pi are used in the battery model, which is itself used by the estimation algorithm (here, EKF) to determine the corrected state of charge ecc(k). Indeed, the corrected state of charge ecc, determined in substep 80, cannot be retained if at least one of the estimated impedance parameters pi is unreliable.

[0079] But in an alternative implementation (not illustrated) the verification of the normality criterion could be done before substep 80.

[0080] For example, in substep 90, one (e.g., the estimation device DE) can determine whether the value of each estimated impedance parameter pie for a cell CE falls within the first range of selected values. If all the estimated impedance parameter values ​​pie fall within the first range of selected values, then they are all reliable, and therefore the robustness criterion can be verified. Conversely, if at least one of the estimated impedance parameter values ​​pie falls outside the first range of selected values, then they are considered unreliable, and therefore the corrected load state ecc, just determined in substep 80, cannot be retained. Consequently, substep 40 is performed.

[0081] Also, for example, the robustness criterion can be membership in a second interval of chosen values ​​of the corrected load state ecc(k).

[0082] It should be noted, as illustrated (but not limited to) in Figure 3, that the verification of the robustness criterion by the corrected load state ecc(k) can This is done in substep 100 of step 10-120, immediately following substep 90, which verifies the reliability criterion. Indeed, the corrected load state ecc, determined in substep 80, cannot be retained if it falls outside the second range of selected values, even if the estimated impedance parameters pie are reliable.

[0083] But in one alternative implementation (not illustrated) the verification of the normality criterion could be done just before the verification of the reliability criterion.

[0084] For example, in substep 100, one (e.g., the DE estimation device) can determine whether the corrected charge state ecc(k) falls within the second range of selected values. If so, it is considered robust and can therefore become the new charge state SOCBR in substep 110. Conversely, if it is not (ecc(k) not within the second range of selected values), it is considered not robust and therefore cannot be retained. Consequently, substep 40 is performed.

[0085] For example, and as illustrated (non-exhaustively) in Figure 3, step 10-120 may also include a substep 120 in which, after a full charge of the BC cell battery, the new estimated state of charge SOCBR (for example, the estimating device DE) can be corrected based on a target state of charge ecb, which is a function of the full charge and an estimated maximum state of charge SoCmax of one of the CE cells. This correction is only performed when current values ​​of at least two selected parameters of the BC cell battery and a charging potential current used during its full charge meet normality criteria. When either of these two normality criteria is not met, the new estimated state of charge SOCBC is not corrected and is therefore retained as is.

[0086] To perform this correction after a full charge, one can, for example, proceed as follows. First, the SoCmax variable is Initialized with the raw SoC value of the first CE cell, the SoC of each subsequent CE cell is then compared to the SoCmax variable at each iteration to find the CE cell with the maximum SoC in the entire BC battery. Once this maximum SoC is found, the SoC corresponding to the found CE cell is set to the target state of charge (ecb) value. The SoCs of the remaining CE cells are then adjusted proportionally to the maximum SoC. However, other methods can be used, such as adjusting the SoC of each cell relative to the target state of charge (ecb).

[0087] This ability to reset the SoCmax variable error is particularly useful when cell chemistry results in a relatively flat OCV = f(SoC) curve, as it helps to start a WLTP (Worldwide Harmonized Light Vehicle Test Procedure) cycle without any risk of underestimation or overestimation. It is worth noting that overestimation is problematic for the authorization of regenerative (or recuperative) energy, leading to higher energy consumption (kWh / km), while underestimation is detrimental due to safety and sustainability risks, as it results in a higher-than-expected regeneration current and therefore a greater risk of lithium metal deposition (Li-plating).

[0088] It should also be noted, as illustrated (but not limited to) in Figure 2, that the battery calculator CB (or the dedicated calculator of the DE estimation device) may also include a mass storage MM1, specifically for the temporary storage of each quantity of ampere-hours (Ah) consumed, the estimated impedance parameters pi, each open-circuit voltage, each current capacity, voltage, current, and possibly internal temperature measurements, and any intermediate data involved in all its calculations and processing. Furthermore, this battery calculator CB (or the dedicated calculator of the DD estimation device) may also include an input interface IE for receiving at least each quantity of ampere-hours (Ah) consumed, the estimated impedance parameters pi, each open-circuit voltage, each The current capacity, voltage, current, and possibly internal temperature measurements, are used in calculations or processing, possibly after being shaped, demodulated, and / or amplified, in a manner known per se, by means of a PR2 digital signal processor. Furthermore, this CB battery calculator (or the dedicated calculator of the DD estimation device) may also include an IS output interface, notably to deliver each new (possibly corrected) state of charge SOCBC.

[0089] It should also be noted that the invention also proposes a computer program product (or computer program) comprising a set of instructions which, when executed by processing means of the type of electronic circuits (or hardware), such as for example the PR1 processor, is suitable for implementing the estimation process described above to estimate the state of charge SOCBR of the BC cellular battery of the S system.

Claims

DEMANDS

1. A method for estimating the state of charge of a rechargeable battery (BC) comprising cells (CE) and equipping a system (S), said method comprising a step (10-120) in which a predicted state of charge is determined by means of a chosen estimation algorithm and using a battery model as a function of estimated impedance parameters of said cells (CE) and the voltages across each of said cells (CE), and as a function of a cumulative current consumption by said rechargeable battery (BC), and then a corrected state of charge is determined by means of said estimation algorithm, characterized in that in said step (10-120) it is verified i) whether a current use of said rechargeable battery (BC) satisfies a normality criterion, ii) whether said estimated impedance parameters satisfy a chosen reliability criterion, and iii) whether said corrected state of charge satisfies a chosen robustness criterion, and then,If all of the aforementioned criteria are met, a new estimated load state equal to the corrected load state is provided, or, if at least one of the aforementioned criteria is not met, a new estimated load state equal to the predicted load state is provided.

2. Method according to claim 1, characterized in that in said step (10-120) said estimation algorithm uses an extended Kalman filter.

3. Method according to claim 1 or 2, characterized in that in said step (10-120) said battery model is an electrical circuit model of type said n-RC, with n > 1.

4. A method according to any one of claims 1 to 3, characterized in that in said step (10-120) said normality criterion is a variation over a chosen time of a current flowing in said rechargeable battery (BC) or of an electrical power delivered or received by said rechargeable battery (BC) below a chosen threshold.

5. A method according to any one of claims 1 to 4, characterized in that in said step (10-120) said reliability criterion is membership in a first range of chosen values ​​of each of said estimated impedance parameters.

6. A method according to any one of claims 1 to 5, characterized in that in said step (10-120) said robustness criterion is membership in a second range of selected values ​​of said corrected load state.

7. A method according to any one of claims 1 to 6, characterized in that in said step (10-120), after a complete recharge of said rechargeable battery (BC), said new estimated state of charge is corrected as a function of a target state of charge as a function of said complete recharge and an estimated maximum state of charge of one of said cells (CE), when current values ​​of at least two selected parameters of said rechargeable battery (BC) and a charging potential current used during said complete recharge respectively satisfy normality criteria.

8. Product computer program comprising a set of instructions which, when executed by processing means, is suitable for implementing the estimation method according to any one of claims 1 to 7, in a system (S) comprising a rechargeable battery (BC) comprising cells (CE), to estimate a state of charge of said rechargeable battery (BC).

9. Estimating device (ED) suitable for being part of a system (S) comprising a rechargeable battery (BC) comprising cells (CE), and comprising at least one processor (PR1) and at least one memory (MD) arranged to perform the operations of determining a predicted state of charge by means of a selected estimation algorithm and using a battery model as a function of estimated impedance parameters of said cells (CE) and voltages across each of said cells (CE), and as a function of a cumulative current consumption by said rechargeable battery (BC), and then determining a corrected state of charge by means of said estimation algorithm, characterized in that said processor (PR1) and memory (MD) are further arranged to perform the operations of verifying i) whether a current use of said rechargeable battery (BC) satisfies a normality criterion,ii) if said estimated impedance parameters satisfy a chosen reliability criterion, and iii) if said corrected load state satisfies a chosen robustness criterion, then, if all said criteria are satisfied, to, provide a new estimated state of charge equal to said corrected state of charge, or, if at least one of said criteria is not met, provide a new estimated state of charge equal to said predicted state of charge.

10. System (S) comprising a rechargeable battery (BC) comprising cells (CE), characterized in that it further comprises an estimation device (DE) according to claim 9.

Images