FR3100670A1 - PROCESS FOR CONTROL OF A BATTERY ALLOWING THE BALANCING OF THE CELLS OF A CURRENT LINE BY A CHARGE AND DISCHARGE CURRENT - Google Patents

Abstract

The present invention relates to a method for controlling a battery (BAT) comprising elementary modules (MEk) of electrochemical cells (CEk) allowing balancing of the state of charge of the cells (CEk). More precisely, the method comprises the following successive steps: determining the state of charge of each elementary module (MEk) of the plurality n, determining a ranking of the n elementary modules (MEk) according to an order depending on l 'state of charge of each elementary module, a control step consisting in connecting to the current line (LT1) the cells of said group of q elementary modules (MEk) having the highest state of charge in the classification (13) in case of detection of a discharge line current, and in connecting the cells of said group of q elementary modules (MEk) having the lowest state of charge in the classification (13) in case of detection of a current load line. The invention applies to the fields of electromobility and stationary energy storage. Figure 1

Description

METHOD FOR CONTROLLING A BATTERY FOR BALANCING THE CELLS OF A CURRENT LINE BY A CHARGE AND DISCHARGE CURRENT

The field of the invention relates to a method for controlling an electrochemical battery allowing the balancing of the state of charge of electric cells for electromobility and stationary energy storage applications. The invention applies, but not exclusively, to Lithium-ion (Li-ion) batteries.

Nowadays, lithium-ion technology electrochemical cells are particularly suitable for electric traction powertrains because they have an interesting ratio of energy density and kWh cost. As we know, the energy cells constituting the batteries have different self-discharge rates resulting in a natural lag between them. This is a slow phenomenon which over time, if not compensated for, can limit the useful capacity of the battery. To correct this phenomenon, it is known to operate a balancing mechanism consisting in aligning all the cells of the battery on the same level of state of charge. However, when the cells are connected in series in order to increase the battery voltage, the same charging or discharging current passes through all the cells, which does not allow the state of charge of each cell to be adjusted individually. .

For these architectural cases, those skilled in the art know specific balancing mechanisms. A first technique called passive balancing consists of discharging the most charged cells by temporarily connecting resistors to their terminals. This technique has several shortcomings. It causes a loss of energy on board the battery by heat dissipation in the balancing resistors. The balancing currents generally limited to low values (a few tens or even hundreds of milliamperes) to avoid excessive heating of the resistors consequently increases the balancing time when the battery is highly unbalanced. A second technique called active balancing consists of transferring energy from the most charged cells to the least charged cells by driving charge transfer currents between the cells. These balancing currents have slightly higher values than the passive solution and are around a few amperes. The advantage of this technique is that it aims to conserve energy, however its cost is much higher than the passive balancing technique due to the fact that it requires dedicated power electronics (transformers, buffer capacitors ) to operate the energy transfer. In addition, its implementation nevertheless presents significant transfer losses due to an energy efficiency of about 50% to 80%. Since passive systems are significantly less expensive, car manufacturers generally opt for the latter.

In a classic battery module, also referred to as a "pack", all the cells are connected in series and in parallel to form a high voltage battery. This battery module is electrically connected to an external converter of the inverter type to deliver the desired voltage to the electric traction machine.

Also known from the state of the art are the documents WO2012117110A2 and WO2014145756A1 describing architectures of battery modules with integrated power control comprising modules of cells electrically connected in series and parallel and describing battery management strategies allowing the balancing the cells in the charging and discharging phase.

Furthermore, the applicant recently filed patent applications WO2018193173A1 and WO2018154206A1 concerning an alternative battery architecture where the innovation consists in proposing an integrated inverter control architecture within the high voltage battery module itself and in which each electric cell , having at its terminals a nominal elementary voltage of between 2V and 4V (approximately 2.2V or 3.3V to 3.7V depending on the chosen technology, for example NMC/LTO or NMC/Graphite), is associated with a power structure individual allowing the cell to be individually controlled to provide a line output voltage generated in stages of amplitude equal to the elementary voltage. The control principle of this battery architecture consists at all times in connecting to the current line q cells in series among a plurality of n cells in total, where q is the ratio between the reference voltage setpoint and the elementary voltage d 'a cell.

This architecture has many advantages cited below non-exhaustively. It is promising in terms of performance and autonomy and allows direct connection of the battery module with a single-phase or three-phase network without the need for an external inverter. It allows individual cells to be connected or bypassed (i.e. disconnected and shunted) so as to disconnect a defective cell if necessary without necessarily replacing an entire battery module. The integrated control structure enables full-wave control which causes less degradation in the windings of an electrical machine, reduces common mode currents and eliminates leakage currents generally present in traditional architectures. In the case of a motor vehicle, this architecture allows degraded operation without immobilizing the vehicle and improves management in recycling or second life of the cells because it is possible to know the state of health of each cell. In a stationary energy storage application, it is possible to replace a defective cell without necessarily stopping the operation of the station.

However, depending on the reference voltages requested, it is possible that certain cells of the system are isolated for more or less long periods. This architecture therefore tends to naturally unbalance cells. Moreover, certain situations in the life of motor vehicles require at the same time a low voltage level and a high current demand, such as, for example, climbing the sidewalk. A large part of the cells is therefore bypassed during the realization of the reference voltage. Cells can quickly become unbalanced for a short time.

One objective of the invention is to improve this last battery architecture proposed by the applicant. More precisely, the aim of the invention is to prevent an imbalance of the cells during the generation of a line voltage. The invention also aims to improve the control of the battery to allow balancing of the cells at any instant of operation of the battery, in electrical charge and discharge, for electromobility and stationary applications. The invention also proposes a control method guaranteeing the homogeneous charging and discharging of the cells for specific life situations where the voltage set point has a low voltage level.

More specifically, the invention relates to a method for controlling a battery with electrochemical cells allowing the state of charge of the cells to be balanced, the battery comprising at least one current line, formed by a plurality n of elementary modules of cells connected in series, capable of delivering a voltage waveform, each elementary module comprising two connection terminals, at least one cell delivering an elementary voltage Vcell and switching means capable of driving the elementary module in the current line function of three different control states making it possible respectively to deliver to said terminals said elementary voltage Vcell, a zero voltage and said inverted voltage Vcell so as to produce a voltage waveform, the method comprising a step of controlling the control signals elementary modules so as to provide the voltage waveform from a selection of a group of q elementary modules as a function of a reference voltage setpoint Vref, where Vref=qVcell. According to the invention, the method further comprises:

- The determination of the state of charge of each elementary module of the plurality n,

- The determination of a classification of the n elementary modules according to an order depending on the state of charge of each elementary module,

- And in that the control step consists in connecting to the current line the cells of said group of q elementary modules having the highest state of charge in the classification in the event of detection of a discharge line current , and to connect the cells of said group of q elementary modules having the lowest state of charge in the classification in the event of detection of a load line current.

According to a variant, the piloting step comprises the following sub-steps:

- When the current of the current line and the reference voltage setpoint Vref have the same sign, the control of the elementary modules so as to connect to the current line the cells of the elementary modules whose position nk is between 1 and q,

- When the current of the current line and the reference voltage setpoint Vref are of opposite sign, the control of the elementary modules so as to connect to the current line the cells of the elementary modules whose position nk is between n-q and not.

According to a variant, the determination of the state of charge of each elementary module is triggered at an instant of switching of the piloting state of said each elementary module and said determination consists in estimating the variation of state of charge of said each module between a first and a second instant of switching based on the control state of said each module between the first and the second instant of switching and the value of the mean current of the current line.

According to a variant, the method comprises a diagnostic step, for example calibration of the state of charge value of an elementary module, comprising the following successive sub-steps:

- In the event of detection of a diagnostic request, the assignment to said elementary module of an exclusion position in the classification so as to exclude said elementary module from said group of q modules for a predetermined exclusion period,

- The execution of the diagnostic of the elementary module during the exclusion period.

According to a variant, the step of determining the classification is triggered according to a period whose value is dependent on the rate of discharge and charge of the battery over a predetermined duration.

According to a variant, the method further comprises, in the event of detection that each elementary module of the plurality n has a difference in state of charge, with respect to the other elementary modules, less than a predetermined minimum threshold, the processing of the classification of the plurality n according to a circular permutation of the positions of the elementary modules so that each elementary module of the plurality n participates in producing the voltage waveform.

According to a variant, the circular permutation is executed at a permutation frequency greater than the frequency of the reference voltage setpoint.

According to a variant, the processing of the classification according to the circular permutation is executed in the event of detection in addition that the reference voltage setpoint Vref is lower than a predetermined voltage threshold.

According to a variant, the method further comprises, in the event of detection that at least one elementary module of the plurality n has a difference in state of charge with respect to the other elementary modules which is greater than a predetermined maximum threshold, the stop round-robin sorting processing.

According to a variant, the method further comprises, in the event of detection that the remaining capacity of the battery is less than a predetermined minimum threshold, stopping the processing of the classification according to the circular permutation.

The invention also provides a control unit for an electrochemical battery comprising at least one current line, formed by a plurality n of elementary cell modules connected in series, capable of delivering a voltage waveform, each elementary module comprising two connection terminals, at least one cell delivering an elementary voltage Vcell and switching means capable of driving the elementary module in the current line according to three different driving states respectively making it possible to deliver said elementary voltage Vcell to said terminals , a zero voltage and said voltage Vcell inverted so as to produce the voltage waveform. According to the invention, the control unit is configured to implement the control method allowing the state of charge balancing of the cells according to any one of the preceding embodiments.

An electrochemical battery system including said control unit is contemplated.

A motor vehicle is contemplated comprising an electric traction module and said battery system electrically connected to the traction module.

A stationary electrical energy storage unit comprising said battery system is also envisaged.

The invention also relates to a program-computer product comprising instructions which, when the program is executed by a control unit of the battery system, lead the latter to implement any one of the embodiments of the control method allowing the balancing of the cells of the battery.

This control of the voltage wave on the current line according to the method makes it possible to operate a dynamic balancing of the cells individually during the operation of the battery in charging and discharging during acceleration phases, regenerative braking and in phase of charging and discharging on the electrical mains when the vehicle is connected to an external charging station. The method also applies to fixed stations connected to the electrical network or operating in island mode, that is to say disconnected from the electrical distribution network.

Moreover, the balancing is carried out under the action of the charging and discharging current of the battery, which improves the balancing dynamics with regard to passive and active balancing solutions while avoiding losses by Joule effect and reducing yield losses.

Furthermore, the electrical architecture of the battery with integrated inverters makes it possible to cohabit cells of different capacities in the battery, to reduce switching losses, to reduce common mode currents in an electrical machine, to eliminate leakage which generate in conventional architectures mechanisms of degradation of bearings and gears by corrosion ("pitting"), to control degraded operation without immobilizing the vehicle and to control a full-wave control which causes less degradation in the machine windings. The architecture also has high efficiency at low power and partial power compared to a conventional structure, charging in direct connection with a single-phase and three-phase network with dynamic cell balancing and dynamic balancing carried out in masked time. There is therefore no longer any need for life phases linked to the static balancing of the cells, generally carried out at the end of the battery charge.

Other characteristics and advantages of the present invention will appear more clearly on reading the following detailed description comprising embodiments of the invention given by way of non-limiting examples and illustrated by the appended drawings, in which:

schematically represents a battery system comprising an integrated inverter architecture allowing individual control of an electric cell able to implement the control method according to the invention.

represents an example of a voltage waveform that can be generated by the battery system, in particular for supplying an electrical machine with alternating voltage of sinusoidal form.

schematically represents a functional module of the battery system control unit whose function is to ensure the balancing of the cells in the charging and discharging situation of the battery.

represents an embodiment of a control algorithm of a battery in accordance with the method according to the invention allowing the balancing of the cells of a current line.

is a graph representing the instantaneous current seen by a cell of an elementary module during the implementation of the balancing control method.

represents an example of a motor vehicle electric traction powertrain system comprising a battery system with integrated inverter and more specifically the control chain of the battery system implementing the method according to the invention.

represents an example of an electrical energy storage station system for a high-voltage network comprising a battery system with integrated inverter and more precisely the control chain of the battery system implementing the method according to the invention.

The invention applies to battery systems consisting of electrochemical cells, for example of the lithium-ion type, where each cell is individually controllable according to the architecture proposed by the applicant in the international patent applications WO2018193173A1 and WO2018154206A1 mentioned above. This architecture is able to generate complex output voltages that can take the form of several types without the need for an additional voltage converter external to the battery system, for example an alternating voltage of sinusoidal form electrically connected to an asynchronous or synchronous electrical machine. , or to a single-phase or three-phase electrical network, or even a constant voltage electrically connected to a direct current electrical machine, a DC/DC type voltage converter or a charging interface.

FIG. 1 schematically represents an embodiment of the battery system configured to apply the control method according to the invention. The battery system comprises a battery module BAT having at least one current line LT1 and a control unit UMI whose function is at least to control the voltage waveform of the line LT1 according to a setpoint reference Vref. According to the embodiment of FIG. 1, the battery BAT comprises three current lines LT1, LT2, LT3 capable of producing three three-phase voltages offset by 2π/3. The control of each line of current is similar, differing only by a shift of 2π/3 between them. The purpose of the control method according to the invention is to enable the balancing of the electric cells of the same current line LT1, LT2 or LT3.

As those skilled in the art know, an elementary electrical cell of a battery is an electrical energy accumulator having two terminals, a positive electrode and a negative electrode, and having a nominal voltage of a few volts, most often between 2V and 4V, generally around 2.2V or 3.7V. More precisely, a Lithium-ion cell is also composed of a separator, an electrolyte and current collectors deposited on the electrodes. The operating principle of a lithium-ion cell is based on the reversible exchange of lithium ions between the two electrodes. During discharge, the cell provides electrical energy, the lithium ions disintegrate from the negative electrode, releasing an electron, then migrate and are transported through the electrolyte to the positive electrode. The electrons produced at the negative electrode and consumed at the positive electrode circulate through a circuit external to the cell and generate an electric current, which will supply the loads connected to the terminals of the cell. The process is reversed during charging. The control process applies to other technologies, such as Lithium polymer, Nickel Cadmium or Nickel-Metal-Hydride type cells.

A battery system (or pack) consists of several elementary cells arranged in such a way as to present, at the terminals of a current line, a voltage higher than that of an elementary cell. The battery system BAT according to FIG. 1 presents at its terminals a voltage of several hundreds of volts, preferably 250V or more, but can, depending on electrical requirements, present only several tens of volts (24V, 36V, 48V for example), and addresses in particular applications for the power supply of electric vehicle traction machines, in particular motor vehicle equipped with an asynchronous, synchronous or direct current electric machine, but also for fixed energy storage stations for the high power grid tension.

In the case of a vehicle powertrain, the output of each current line is intended to be connected directly (without voltage converter) to the current line of one or more electrical machine(s) , and to the power line of a vehicle charging interface that can operate at constant voltage or alternating voltage.

The battery system BAT further comprises a plurality n of elementary modules ME1, MEk-1, MEk, MEn of electric cells, said modules being connected in series between two terminals B1, B2 to form each current line LT1, LT2, LT3 . The n elementary modules are referenced by an index k, where k is a natural integer between 1 and n. For example, n can be equal to 80 to form a line voltage which can reach an average peak value of approximately 290V on each current line LT1, LT2, LT3, or a lower voltage, for example approximately 240V, at from a group q of elementary modules, where q is less than n. For the sake of simplification, a single elementary module is shown in detail.

More specifically, each elementary module MEk comprises two connection terminals Bk1, Bk2, at least one electrical elementary cell CEk delivering an elementary voltage Vcell connected between said terminals, switching means comk1, comk2 capable of configuring the elementary module MEk in three states different respectively delivering said elementary nominal voltage Vcell, a zero voltage and said inverted voltage Vcell to said connection terminals Bk1, Bk2. Each elementary module MEk comprises a single elementary cell.

As a variant, two or more elementary cells are electrically connected within the same elementary module, directly and permanently in parallel and/or in series with each other. In this way, such a group of elementary cells has two connection terminals electrically connected to the switching means comk1, comk2 so as to deliver the voltage to the terminals of the group of elementary cells, a zero voltage and said inverted voltage to the connection terminals Bk1 , Bk2 of the elementary module. In the event that several elementary cells are connected in series in a single elementary module MEk, this elementary module also comprises a means of conventional balancing of said cells in series of said elementary module.

The switching means is for example made up of two elementary switching modules comk1, comk2 forming an H bridge which can be controlled in the three different states by a control signal uik from the control unit UMI of the battery BAT, performing the function of "DRIVER" according to the English term. The states are represented by a control variable uik which can take for example the values 1, 0, -1 representing the three different states respectively controlling said elementary voltage Vcell, a zero voltage and said inverted voltage -Vcell at said connection terminals Bk1, Bk2 of the elementary module indexed k. As illustrated in FIG. 1, each elementary switching module comprises two electronic components, such as power transistors, possibly of the MOSFET or HEMT ("High Electron Mobility Transistor" in English) type, driven by the signals ui1 to uin of the UMI control unit. Thus, the voltage vik at the terminals Bk1, Bk2 of each elementary module MEk can be controlled according to the control signal uik according to the following relationship:

In figure 2, an example of a voltage waveform that can be delivered by the battery BAT on the current line LT1 has been shown. In this non-limiting example, the reference voltage setpoint Vref is of sinusoidal form and the phase voltage VM1 of the current line LT1 is formed in stages of amplitude equal to the elementary voltage Vcell. Any voltage waveform can be controlled by the control unit, for example sinusoidal with a frequency of 50hz or constant voltage for example.

We now describe the functional means of the UMI control unit allowing the implementation of the control method according to the invention to ensure the balancing of the state of charge of the cells of the same current line. This implementation can be done by means of the UMI control unit of the BAT battery, which can be integrated or constituted the battery management computer BMS ("Battery Management System" in English) in a decentralized architecture, as illustrated non-exhaustively in FIG. 1. But this is not obligatory. Indeed, the method could be implemented in a module external to the BMS, for example within an elementary module MEk according to a decentralized architecture, while being coupled to the latter. In the latter case, the external module can itself be arranged in the form of a dedicated computer comprising a possible dedicated program. Consequently, the module for implementing the battery control method, according to the invention, can be produced in the form of software (or computer (or even “software”)) modules, or electronic circuits (or "hardware"), or a combination of electronic circuits and software modules.

In figure 3, a functional module of the control unit UMI of the battery BAT has been shown more precisely, the function of which is to control the control signals uik of each elementary module MEk.

The control unit UMI is capable of receiving as input parameter a reference voltage setpoint Vref and a current value Ia of the current line (a setpoint or a measurement). In response to the voltage setpoint Vref, a line current is generated passing through each elementary module connected to the current line. The line current can be a charging current and a discharging current (with active and/or reactive component). In the case of an alternating reference voltage, the line current can be in phase or out of phase with the reference voltage by a phase shift value between -90° and 90° depending on the electrical behavior (resisto-inductive) of the electrical network (mains or on-board network of the vehicle) connected to the, or each, current line of the battery.

A first module 11 is able to determine at each instant the number q of elementary modules necessary among the plurality n to produce the voltage waveform for each quantized voltage phase ViM requested by the setpoint Vref and where all the elementary cells have a same elementary voltage Vcell according to the following relationship:

A second module 12 is capable of assigning to each elementary module MEk a position nk in a classification 13 (table, list or index) recorded in the memory of the control unit UMI. This classification 13 is dynamic and modifiable by the control unit UMI and records the position nk of each of the elementary modules MEk at any time. Thus, thanks to the dynamic classification it is possible to obtain the same voltage from different combinations of commands uik of the elementary modules. The object of the invention is to modify this classification 13 during the production of a voltage wave of a current line in charge and discharge of the battery so as to operate the balancing of the cells.

To this end, the position nk of each elementary module MEk is at least dependent on the state of charge level SOC of each elementary module. Classification 13 has the function of recording a list of elementary modules in ascending or descending order of SOC state of charge. For the ranking in descending order, the elementary module Mek having the highest state of charge, the module 12 assigns the position nk=1, and so on until the position nk=n for the elementary module having the lowest state of charge.

In addition, each elementary module MEk is individually controllable by a module 14 in said three different states (+1, 0, -1), and in accordance with the method the module 14 is configured to control the elementary modules so as to connect to the line current a group of q elementary modules MEk having the highest state of charge SOC in classification 13 in the event of detection of a discharge current crossing the current line LT1, and in connecting said group of q elementary modules MEk having the lowest state of charge in the ranking in the event of detection of a charging current crossing the current line LT1. The q modules have successive positions in the ranking 13.

By convention, the following terms are defined:

The line current is a discharge current when the current and the voltage have the same sign, i.e. at a given instant the reference voltage and the line current are both of positive sign or of negative sign .

The line current is a load current when the current and the voltage are of opposite sign, i.e. at a given instant, the reference voltage is of positive sign and the line current is negative, where the reference voltage has a negative sign and the line current is positive.

A line current is positive when the current is outgoing to the electrical circuit and the line current is negative when the current is incoming to the battery.

The sign of the voltage is referenced in relation to terminals B1, B2 of the battery current line where terminal B1 is connected to a phase terminal of the electrical circuit (electrical machine winding terminal or electrical network phase terminal) and the terminal B2 to a neutral terminal. In the case of a DC voltage circuit, terminal B1 is the positive terminal and terminal B2 is the negative or ground terminal of the battery.

This mode of controlling the voltage wave on the current line makes it possible to operate a dynamic balancing of the cells individually during the operation of the battery by controlling a charge and discharge current crossing the current line. In addition, the balancing is carried out under the action of the charging and discharging current of the battery of the order of several tens of amperes, or even several hundreds, which improves the balancing dynamics with regard to the known solutions of passive and active balancing where the balancing currents are only a few amperes, which they present losses by Joule effect and therefore losses of efficiency. The charge/discharge current is a current crossing the current line of the battery when the latter is connected to the electrical network of the vehicle or the mains.

According to a battery control variant, the module 14 is configured so that the classification consists in assigning a position nk to each elementary module MEk comprised between 1 and n in a list in descending order of state of charge, and to each instant the step of piloting the elementary modules comprises the following sub-steps:

- The determination of the line current Ia of the current line LT1,

- When the line current Ia and the reference voltage setpoint Vref have the same sign, the control of the elementary modules MEk so as to connect to the current line the elementary modules MEk whose position nk is between 1 and q,

- When the line current and the reference voltage setpoint are of opposite sign, the control of the elementary modules MEk so as to connect to the current line the elementary modules whose position nk is between n-q and n.

It is recalled that at each instant, the control unit UMI connects q cells (q ≤ n) where q is the ratio between the reference voltage Vref desired at this instant at the line output and the elementary cell voltage Vcell.

Thus, the state of each control variable of the elementary modules can be controlled by the module 14 according to the following algorithmic relationship where the classification is indexed in decreasing order of state of charge:

Where nk is the position of each module in the classification, Vcell is the elementary voltage of a cell and V _refi and i _i are the reference voltage setpoint and the line current (ampere) of each current line referenced i, or i is for example between 1 and 3 for each current line of a three-phase voltage respectively.

The person skilled in the art will of course be able to configure the classification by order of increasing state of charge and modify the algorithmic relationship allowing the control of each elementary module MEk accordingly.

Furthermore, it is possible for the module 12 to be able to assign a position nk as a function, in addition to the state of charge SOC, of a criterion which is dependent on the state of aging of each elementary module or on the residual capacity of each elementary module. For example, for modules with an equivalent state of charge, it is possible to carry out the classification by taking into account the state of aging in order to activate in priority the younger modules or those with a higher total residual capacity compared to the modules presenting a more advanced fatigue.

In addition, a module 15 is able to determine the state of charge SOCk of each elementary module MEk at any time. The state of charge SOCk corresponds to an electrical capacity contained in the cell of an elementary module MEk, expressed in A.s, A.h (Ampere per second or Ampere per hour) or even usually in percentage of charge compared to a total capacity real at time t.

The determination of the state of charge SOCk is calculated by a first coulometric method. As is well known to those skilled in the art, the estimation by coulometric measurement is carried out from a known initial state of charge of reference, for example complete state of charge corresponding to 100% of the SOC for a capacity total known at the instant of evaluation for a known state of aging, and from the load variation measured by a current sensor of the current line. The algorithms for evaluating the state of health of an electrochemical cell are well known to those skilled in the art and can be implemented by the control unit of the battery system BAT.

Advantageously, the module 14 is configured to determine the state of charge SOC of each elementary module MEk at a time of switching of the piloting state of each elementary module. To this end, the module 14 estimates the variation in the state of charge of each module MEk between a first and a second instant of switching from the control state uik of each module MEk and the value of the average current of the current line. Thanks to this method, it is thus not necessary to equip the battery with a current sensor for each cell. The control logic variable uik is known at all times and thus makes it possible, solely from the measurement of the line current, to estimate the state of charge by an integration calculation over the period between the two switching times . This period can be fixed or variable. Control state switching means the control command corresponding to a refresh of the control state of an elementary module MEk and covers the situation of change of a first control state among the states +1, 0, -1 to a second different state among these same states or, from a control state among the states +1, 0, -1 to this same state.

This first coulometric method nevertheless has a disadvantage, because over time it is likely to lead to an error in the estimation of the SOC increasing as the calculations progress. Provision is therefore also made for the control unit UMI to include a module 18 capable of performing a readjustment or calibration of the estimate of the state of charge of each cell of the elementary modules, or even any other type of diagnosis such as for example an assessment of the state of aging.

According to a variant consisting in calibrating the state of charge when the precision is no longer guaranteed by the coulometric method, the module 18 is capable of estimating the no-load voltage of a cell during a relaxation phase of the cell ( principle based on Kalman filters or possibly tabulated values). No-load voltage estimation techniques are well known to those skilled in the art and are not precisely the subject of the invention. As we know, some methods require several tens of minutes to observe the stabilization of the voltage at rest and to obtain a reliable value of the no-load voltage and therefore of its state of charge SOC. Other modeling methods can run in less time but require more computational resources.

Whatever the method for estimating the off-load voltage, the control unit UMI comprises a module 17 configured to assign, at any time, to each elementary module MEk an exclusion position in the classification 13 so as to exclude said elementary module of said group of q modules during a predetermined exclusion period. Thus, thanks to the module 17, it is possible to temporarily bypass an electrical cell individually, independently of the value of its state of charge, to carry out a diagnosis or calibration in masked time, without affecting the normal operation of the battery. It is conceivable that the action of exclusion of the module 17 be triggered for any type of diagnosis requiring the deactivation of the elementary module concerned, whether it is the calibration of the state of charge or the evaluation of the state of aging of a cell for example.

It should be noted that the duration of exclusion is dependent on the time required to make the diagnosis. Once the exclusion position is assigned, nk greater than q in discharge and nk greater than n-q in charge, module 18 is authorized to execute the no-load voltage evaluation method for the associated elementary module. It will be noted that the action of the module 17 is a function of the state of a flag or indicator ("flag" in English) activating when the accuracy of the state of charge is not guaranteed by the unit of order. The indicator is activated, for example high state, to request a readjustment of the SOC. The indicator deactivates, low state, when the recalibration is performed.

Furthermore, a module 16 is optionally provided, the function of which is to operate a processing of the classification of the plurality n according to a circular permutation of the positions nk of the elementary modules MEk so that each elementary module of the plurality n participates in carrying out the voltage waveform for an equal duration for each module. The permutation consists in shifting the position of the elementary modules in the classification. Since the number q of elementary modules necessary to produce the waveform is less than the total number n of modules available, at each instant some modules are shunted while others are crossed by the line current. This classification cycling operates an action of chopping the current passing through each cell. Cycling is particularly interesting when the cells have reached a state of charge equilibrium between them over a given period to equalize the average currents passing through them. Similarly, cycling makes it possible to reduce the impact of strong current stresses and when few cells are stressed (low q compared to n) to produce the voltage waveform, in particular in the event of a sidewalk climb for a motor vehicle. This avoids accentuating any imbalance that already exists.

Thus, the triggering of cycling by the module 16 according to the circular permutation can be dependent on a difference in the state of charge between the elementary modules MEk with respect to a predetermined threshold, or even according to a setpoint value of current in relation to the value of the requested voltage level.

Specifically, module 16 modifies the classification table 13 according to the following algorithmic relationship:

The cycling of the positions nk in the total list of the n elementary modules takes place when q is less than n and at permutation frequencies greater than the frequency of the signal Vref, for example approximately 1 kHz for a reference voltage Vref of frequency 50 Hz. The ratio between the switching frequency and the frequency of the voltage signal can be different and will be chosen according to the switching joule losses with regard to the drop in internal impedance of the battery. The permutation frequency is configured to ensure the complete permutation of the n modules over a predetermined period, i.e. each module has been positioned in the ranking at each of the positions successively over the said period. The predetermined duration equal to or less than the period of the electrical voltage of the current line.

It has been observed that cycling makes it possible to reduce the internal impedance value of an electric cell, by around 30% for a cell using Lithium Iron Phosphate technology. It is thus possible to adapt the switching frequency to the technology of the cells chosen. Thus, as a variant, the permutation frequency of the circular permutation can be dependent on the value of the internal impedance of the cell of each elementary module.

It will be added that the classification 13 of the elementary modules by order of state of charge is active at any moment of the operation of the battery, in charge, discharge, so as to guarantee the balancing of the cells of a current line without protocol of dedicated monitoring. The frequency of charging state ordering is configured according to a predetermined duration or according to the battery discharge/charge speed, for example in a range between 1 second and 5 minutes, or for a variation of predetermined proportion of the total capacity of the battery, for example each 1% of SOC discharged/charged (ie each hundredth of the capacity with respect to the total capacity of the battery).

In Figure 4, an algorithm block diagram is described describing an embodiment of the method for controlling a battery allowing the balancing and maintenance of the balancing of the cells of a current line. The process includes:

- a step 31 for determining the state of charge of each elementary module MEk of the plurality n,

- then a determination step 32 of a classification of the n elementary modules according to an order depending on the state of charge of each elementary module,

- then a step 33 of controlling the control signals of the elementary modules MEk so as to provide, at any time t, the voltage waveform of the current line from a selection of a group of q modules elements according to the reference voltage setpoint Vref, where Vref=qVcell.

More specifically, according to the method, this control step 33 consists in connecting to the current line said group of q elementary modules, successive in the classification, having the highest state of charge in the classification in the event of detection of a discharge line current, and connecting said group of q elementary modules having the lowest state of charge in the classification in the event of detection of a charging line current.

This step is carried out thanks to the invention during the charging and discharging battery operating phase under the action of the charging and discharging current, thus guaranteeing a balancing dynamic superior to the passive and active solutions described in the state of the art. The line current and the voltage setpoint are alternating signals of sinusoidal form taking positive and negative values over a period. The method thus preserves the balancing of the cells at any moment of operation of a vehicle due to the consideration of the direction of the current to detect the phases of charge and discharge over at least one phase period, in particular during the succession acceleration and regenerative braking phases. The balancing control is of course applicable identically for a vehicle battery connected to an electrical network via the charging station or for a stationary application, in particular in the case of a so-called "Smart Grid" control process for the management of a network where charging and discharging phases are controlled.

Thus, over a period of the reference voltage, an elementary module with a high load relative to the other modules (for example nk=1) will be subjected to a discharge current only, while a module with a low load compared to the other modules (eg nk=n) will be subject to load current only.

To illustrate this example, FIG. 5 shows the instantaneous current Ia of a cell of an elementary module of position nk=10 during the execution of the control method for an example of battery architecture. On the ordinate axis is represented the value of the charging current in amperes and on the abscissa axis the time t in milliseconds.

In this example, the current line is equipped with 80 elementary modules mounted in series, the unit voltage Vcell is equal to 3.3V, the sinusoidal reference voltage has an amplitude of 264V and the sinusoidal current is equal to 100A. The voltage and current have a phase shift of 45° considering the inductive behavior of the machine windings and its operating point. As seen in this figure, the current seen by the cell is partially sinusoidal in shape and remains of positive sign over the same period, because the cell of the elementary module is connected to the line only when the line current is a current of dump. Indeed, its relatively low position in the list (high state of charge) exposes it to discharge currents only. It should be noted that the current seen by an elementary module therefore depends on the state of charge value of the cell of the elementary module, and on the phase shift between the voltage setpoint and the line current. In the event of zero phase shift, the current will be of continuous value without cuts.

Then, according to the method, in the event of detection, at a step of 34, of activation of a classification rescheduling condition, duration since the previous scheduling greater than a predetermined duration, or possibly a variation in the state of charge since the previous scheduling greater than a predetermined threshold, steps 31 and 32 are executed again. The predetermined duration can be between one or several minutes for example, in order to reduce the energy losses linked to the switching actions of the transistors.

According to a variant of the method, represented in FIG. 4, the classification step 32 consists, in a sub-step 320, in assigning a position nk to each elementary module between 1 and n in a list in decreasing order of state of charge and the piloting step 33 comprises the following successive sub-steps:

- The determination 330 of the sign of the current of the current line to then compare in a step 331, the sign of the current and the sign of the reference voltage to detect whether the line has a charge or discharge current.

- When the current and the reference voltage setpoint have the same sign (that is to say discharge current), the control 332 of the elementary modules so as to connect to the current line the elementary modules whose position nk is between 1 and q.

- When the current and the reference voltage setpoint are of opposite sign (that is to say load current), the control 333 of the elementary modules so as to connect to the current line the elementary modules whose position nk is between n-q and n.

It is further envisaged that the classification by order of state of charge carried out by the module 12 can be deactivated if the control unit UMI detects that the cells are balanced, that is to say that the state difference load between each cell is less than a predetermined minimum threshold.

At a step 35, if the control unit detects such a situation, the module 16 is activated to operate at a step 36 the circular permutation of the classification so as to guarantee the same average current demand for all the cells. The circular permutation is active as long as the cells are balanced and is carried out at a permutation frequency greater than the frequency of the reference voltage set point Vref, for example 1 kHz for a reference voltage of 50 Hz.

As a variant of step 35, the module 16 is activated in the event of detection in addition that the reference voltage setpoint is lower than a predetermined voltage threshold and in the event of simultaneous detection of a current setpoint higher than a threshold of predetermined current. This strategy aims to respond specifically to high current demands when a small number of cells is called upon to produce the voltage waveform. This activation aims to prevent an imbalance of the cells.

SOC ordering 32 may be activated simultaneously with round robin processing 36, or either only during a given period.

Finally in step 35 of the method, in the event of detection of a state of charge imbalance and of a request to deactivate the circular permutation, the method returns to step 31. This situation is detected when a difference in state of charge between at least one of the cells of the other cells is greater than a predetermined maximum threshold.

In a variant, when the remaining capacity of the battery (or residual autonomy of the vehicle for example) is less than a predetermined threshold, the classification by order of state of charge is activated. In this case, the classification by order of state of charge and the associated control is kept active until the remaining capacity of the battery reaches a minimum limit, for example less than 10% or a few percent of the total capacity or even still zero capacity. In this way, all the cells reach the minimum authorized state of charge at the same time.

Additionally, as mentioned previously, grading allows one or more cells to be temporarily bypassed for diagnostic or state-of-charge calibration purposes while maintaining normal battery system operation. In Figure 4, the sequence of these steps 40, 41, 42 has also been shown where:

In the event of detection 40 of a diagnostic request (calibration of state of charge for example), the method comprises a step 41 of assigning to an elementary module an exclusion position in the classification so as to exclude said elementary module of said group of q modules during a predetermined exclusion period, then a step 42 of executing said diagnosis of the elementary module during the exclusion period.

In FIG. 6, there is shown a first application of the control method according to the invention in the field of electromobility for a motor vehicle powertrain comprising a battery system in accordance with the architecture with integrated inverters as described in FIG. 1. The vehicle may be a hybrid or electric vehicle fitted with one or more electric traction machines (asynchronous, synchronous or direct current) where the, or each, machine is electrically connected to the battery system. The powertrain 500 comprises a module 50 for generating a current setpoint 51 sent to a current servo loop. The generation module 50 delivers a current setpoint according to a torque request from the accelerator pedal of the vehicle or from an automatic piloting module, such as an automatic speed regulation function from a speed and torque measurement of the driving wheelset. Conventionally, the current servo loop receives said setpoint 51 and the measurement of the current generated 58 on at least one of the current lines of the battery 56. The battery 56 is electrically connected directly (without electrically interposed voltage converter) to at least one electric traction machine 57 motorizing a wheel set 59 of the vehicle, as well as to a charging interface (not shown) on an electrical network external to the vehicle, in this example in three-phase voltage. The battery architecture does not require the use of an inverter at the battery line output.

The servo loop further comprises a corrector 52 delivering a reference voltage setpoint 53 as a function of the setpoint 51 and the measurement 58, to a control unit 54 of the battery 56. The control unit 54 delivers control setpoints 55 controlling each elementary module of the battery in accordance with the control method described above as a function of the voltage setpoint to maintain the balancing of the cells of the battery.

In Figure 7, there is shown a second application of the control method according to the invention in the field of stationary batteries where a battery system in accordance with the architecture with integrated inverters as described in Figure 1 is connected to the electrical network. The battery system 600 receives a power setpoint 60, resulting from active power (P) and reactive power (Q) demands to be injected into the electrical network 70. The system further comprises a module 61 for determining the power setpoint. current 62 in amplitude and phase as a function of setpoint 60 and a voltage measurement 69 on the current line at the output of battery 67, and a current servo loop comprising a corrector 63 delivering a voltage setpoint reference 64 as a function of a current measurement 68 on a current line (here three-phase) at the output of the battery 67, and of the current setpoint 62. The control unit 65 of the battery 67 delivers setpoints of control for each elementary module of the battery in accordance with the control method described previously to maintain the balancing of the cells according to the reference voltage setpoint

In this type of application, the 600 system can be controlled to charge and discharge the battery from the electrical network.

We now summarize the following various technical advantages of the ordering process:

- Dynamic balancing of the cells during the life phases of the vehicle: traction, charging and discharging on the mains electricity network,

- Balancing of cells that may present different states of aging,

- Balancing of cells with different capacities,

- Dynamic balancing without using a specific balancing device but by reusing the local inverters of each elementary module,

- High balancing efficiency (> 90%),

- Balancing current equal to the battery cycling current (charging and discharging current of the battery in operation),

Claims (10)

Method for controlling a battery (BAT) with electrochemical cells (CEk) enabling the state of charge of the cells (CEk) to be balanced, the battery (BAT) comprising at least one current line (LT1), formed by a plurality n of elementary modules (MEk) of cells connected in series, capable of delivering a voltage waveform, each elementary module (MEk) comprising two connection terminals (Bk1, Bk2), at least one cell (CEk) delivering an elementary voltage Vcell and a switching means (comk1, comk2) capable of driving the elementary module (MEk) in the current line (LT1) according to three different driving states respectively making it possible to deliver said elementary voltage Vcell to said terminals , a zero voltage and said voltage Vcell inverted so as to produce a voltage waveform (VM1), the method comprising a step of driving (33) the control signals (uik) of the elementary modules (MEk) so as to providing the voltage waveform (VM1) from a selection of a group of q elementary modules (MEk) as a function of a reference voltage setpoint Vref, where Vref=qVcell, the method being characterized in what it additionally features:

• The determination (31) of the state of charge of each elementary module of the plurality n,
• The determination (32) of a classification (13) of the n elementary modules according to an order depending on the state of charge of each elementary module (MEk),

• And in that the control step (33) consists in connecting to the current line (LT1) the cells of said group of q elementary modules (MEk) having the highest state of charge in the classification (13) in case of detection of a discharge line current, and to connect the cells of said group of q elementary modules (MEk) having the lowest state of charge in the classification (13) in case of detection of a current of load line.

Method according to claim 1, characterized in that the piloting step (33) comprises the following sub-steps:

• When the current (Ia) of the current line (LT1) and the reference voltage setpoint Vref have the same sign, the control of the elementary modules (MEk) so as to connect to the current line (LT1) the cells of the elementary modules whose position nk is between 1 and q,
• When the current (Ia) of the current line (LT1) and the reference voltage setpoint Vref have opposite signs, the control of the elementary modules (MEk) so as to connect to the current line (LT1) the cells of the elementary modules (MEk) whose position nk is between nq and n.

Method according to Claim 1 or 2, characterized in that the determination (31) of the state of charge of each elementary module (MEk) is triggered at an instant of switching of the control state of said each elementary module (MEk) and in that said determination (31) consists in estimating the variation in the state of charge of said each module between a first and a second instant of switching from the driving state of said each module between the first and the second instant of switching and the value of the average current of the current line. Method according to any one of Claims 1 to 3, characterized in that it comprises a diagnostic step, for example calibration of the state of charge value of an elementary module, comprising the following successive sub-steps:

• In the event of detection of a diagnostic request (40), the assignment (41) to said elementary module of an exclusion position in the classification (13) so as to exclude said elementary module from said group of q modules during a predetermined exclusion duration,
• The execution of the diagnostic (42) of the elementary module (MEk) during the exclusion period.

Method according to any one of Claims 1 to 4, characterized in that the step of determining the classification (13) is triggered according to a period whose value is dependent on the rate of discharge and charge of the battery (BAT) on a predetermined duration. Method according to any one of Claims 1 to 5, characterized in that it further comprises, in the event of detection (35), that each elementary module of the plurality n exhibits a difference in state of charge, with respect to the others elementary modules, below a predetermined minimum threshold, the processing (36) of the classification of the plurality n according to a circular permutation of the positions (nk) of the elementary modules (MEk) so that each elementary module (MEk) of the plurality n participates to realize the voltage waveform. Method according to Claim 6, characterized in that the circular permutation is carried out at a frequency greater than the frequency of the reference voltage setpoint. Method according to Claim 6 or 7, characterized in that the processing (36) of the classification (13) according to the circular permutation is carried out in the event of detection in addition that the reference voltage setpoint Vref is lower than a predetermined voltage threshold . Method according to any one of Claims 6 to 8, characterized in that the method further comprises, in the event of detection that at least one elementary module of the plurality n has a difference in state of charge with respect to the other modules which is greater than a predetermined maximum threshold, stopping the processing (36) of the classification according to the circular permutation. Method according to any one of Claims 6 to 9, characterized in that it further comprises, in the event of detection that the remaining capacity of the battery (BAT) is less than a predetermined minimum threshold, stopping the processing ( 36) of the classification (13) according to the circular permutation.