Method for determining the aging state of an electrochemical cell

Abstract

The present invention relates to a method for estimating the aging state of an electrochemical cell based on parameter estimation, the parameter representing the autocatalytic speed during relaxation. According to the present invention, the method comprises the following steps: continuously calculating (203) a logarithmic signal of the voltage during relaxation during the relaxation of the cell; calculating (207) a model function of the logarithmic signal; calculating (208) a second parameter representing the speed of change of the model function of the logarithmic signal during relaxation; and estimating (209) an aging state parameter based on the second parameter and a predetermined graph.

Description

Technical Field

[0001] The field of the invention relates to a method for determining a parameter of an aging state of an electrochemical cell and to a control unit comprising components for implementing said method. Background Art

[0002] In the motor vehicle industry, hybrid and electric vehicles are equipped with power battery systems with electrochemical accumulators (also called battery cells) intended to power the vehicle's electric motor. These systems now play an important role in improving the energy performance of the vehicle in terms of reducing fuel consumption, electric driving range and transmission power. In addition, multiple transmission functions of the powertrain depend on the values ​​of state indicators of the battery system, such as the charge state level and aging state of the battery system (respectively expressed as the English abbreviations SOC (State of Charge) and SOH (State of Health). For example, the decision to start the internal combustion engine, the choice between hybrid or all-electric driving mode, or the route planning navigation function are usually dependent on the charge state level of the battery system. The accuracy and updating of the state indicators are crucial for the operation and driving experience of the vehicle. The task of the control unit of the battery system (usually referred to as the English abbreviation BMS (Battery Management System)) is to estimate these state indicators in real time.

[0003] Among these indicators, the aging state is essential for the management of the battery system. In fact, the aging state makes it possible to quantify the maximum amount of energy that a battery can deliver after a full charge. From the prior art, the document WO2015049300A1 is known, in which the current drop characteristic during recharging at a constant voltage varies linearly with the aging state of the battery. The current decreases exponentially over time. In addition, the document WO2017050944A1 describes a method that exploits the linear relationship between the open circuit voltage and the aging state. However, these solutions have the following disadvantages.

[0004] As is known, among these indicators, the open circuit voltage (also known by the English acronym OCV (Open Circuit Voltage)) curve is very important, since many electrical characteristics are derived based on the value of the open circuit voltage. However, the voltage OCV should be measured at the thermodynamic equilibrium of the electric energy storage device, and this equilibrium can only be reached after a sufficiently long time (usually more than one hour). The use in the field of motor vehicles reduces the chances of this continuous time and prevents estimation and updating in real time during driving. In addition, future "smart" applications of battery systems are designed to intervene in the management of the household electrical network, which will reduce the frequency of relaxation phases.

[0005] In addition, the method proposed in WO2015049300A1 requires a complete recharge to determine the indicators necessary for diagnosing the battery aging state. The method is therefore not applicable to partial recharge. Similarly, for WO2017050944A1, its method is only applicable when the battery is fully charged or fully discharged. Summary of the invention

[0006] There is therefore a need to overcome the above mentioned problems. The main object is to improve the determination method for determining the state of aging of a battery system. In particular, the invention aims to reduce the calculation time and to be able to determine the state of aging regardless of the state of charge level of the battery.

[0007] More precisely, the invention relates to a determination method for determining a first parameter representative of an aging state of an electrochemical cell of a battery system, the determination method being executed by a control unit of the battery system and comprising the following successive steps:

[0008] - continuously measuring the voltage signal Uacc(t) at the terminals of the cell during relaxation as a function of the elapsed time,

[0009] - estimating a first aging condition parameter, said first aging condition parameter being dependent on said measured voltage signal.

[0010] The method is characterized in that it also includes:

[0011] - continuously calculating a logarithmic signal of the voltage during the relaxation of the battery from values ​​of logarithmic form of the time elapsed since a first moment, the logarithmic signal being dependent on the voltage signal Uacc(t),

[0012] - a step of calculating a model function of said logarithmic signal as a function of the values ​​of said logarithmic signal,

[0013] - calculating a second parameter representing the speed of change of the model function of the logarithmic signal during relaxation, the speed of change representing the autocatalytic speed of the battery during the relaxation phase,

[0014] - estimating the first aging status parameter as a function of the second parameter and a predetermined graph, the graph being stored in a memory of the control unit and providing a predetermined value of the first aging status parameter as a function of the value of the second parameter.

[0015] According to a variant, the calculation of the model function involves adjusting a parameter of the model function, and the second parameter is calculated as a function of the adjusted parameter of the model function.

[0016] According to a variant, the model function is a function g of any of the following relations:

[0017] Formula 1

[0018] g:x→a*tanh(b*x), or

[0019] Formula 2

[0020] g:x→a*arcsinh(b*x), or

[0021] Formula 3

[0022] g:x→a*sigmf(b*x)-0.5,

[0023] Where a and b are adjusted parameters, and the second parameter Vac is calculated according to the following relationship:

[0024] Formula 4

[0025] Vac=a*b.

[0026] According to a variation of the method, the method further comprises a step of detecting an inflection point of the logarithmic signal of the voltage, and calculating the logarithmic model function symmetrically with respect to the inflection point based on a measurement value of the logarithmic signal before the inflection point.

[0027] More precisely, said predetermined pattern is generated during an ageing phase of the electrochemical cell.

[0028] More specifically, said predetermined graph provides a plurality of predetermined values ​​of the aging state according to a plurality of predetermined values ​​of said second parameter.

[0029] The present invention also provides a control unit of a battery system, the battery system includes an electrochemical battery, and the control unit also includes a memory storing a predetermined graph, the predetermined graph is suitable for providing a predetermined value of a first aging state parameter of the battery based on a predetermined value of a second parameter, the second parameter represents the autocatalytic speed of the battery during the relaxation stage, and the control unit includes a component for implementing the determination method described in any one of the above embodiments.

[0030] The invention also provides a motor vehicle comprising a battery system, the battery system comprising the control unit.

[0031] The invention also relates to a computer program product comprising instructions which, when executed by a control unit of the battery system, cause the battery system to implement any one of the embodiments of the method for determining an aging state according to the invention.

[0032] The method according to the invention requires few computing resources and is therefore easy to integrate in on-board applications, such as vehicle battery systems and portable devices. In addition, the results of the method are not affected by the state of charge level of the battery. This feature increases the frequency of implementation of the method and improves the management of the state indicators of the battery system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Other features and advantages of the present invention will become more apparent upon reading the following detailed description of embodiments of the invention provided by way of illustrative example only and the accompanying drawings, in which:

[0034] - Figure 1 A battery system is shown, which comprises an electrochemical cell and a control unit which is configured to carry out the method according to the invention.

[0035] - Figure 2 The flowchart of FIG. 1 shows an algorithm of the method according to the invention, which algorithm is able to characterize the autocatalytic speed of the reactants of the battery during relaxation as well as the aging parameters.

[0036] - Figure 3 The graph shows the changes in the voltage curve of the battery during relaxation and the changes in the logarithmic signal, which can characterize the autocatalytic speed of the battery during relaxation.

[0037] - Figure 4 The graph of φ shows a predetermined pattern stored in a memory of the control unit, which enables the determination of the aging state based on characteristic parameters of a model of the logarithmic signal of the voltage. DETAILED DESCRIPTION

[0038] The invention relates to a method for determining aging parameters of an electrochemical cell of a battery system, which is often also referred to as a battery module or in English as a "pack". The invention makes use of the fact that the rate of change of the voltage signal of the battery during relaxation (which is expressed in the form of a logarithmic signal) is proportional to the rate of change of the volume (or autocatalytic rate) of the phase transition of the reactants during relaxation. In fact, the volume concentration of the reactants changes with respect to time in an S-shaped or "sigmoidal" manner. ". The method aims to model this temporal variation by parameter adjustment based on the measurement of the voltage signal and the associated logarithmic signal. Finally, the method enables the estimation of the aging state during the use of the battery, based on the estimation of the parameter representing the autocatalytic speed and a predetermined graph associating this parameter with the aging state. This association is predetermined and stored in a graphical form during the aging operation in order to implement the method during the use of the battery.

[0039] Figure 1A battery system 100 intended to be mounted on an electric motor vehicle is schematically shown, the battery system comprising a plurality of electrochemical accumulators. The battery system 100 comprises an electrochemical accumulator module 10 and a control unit 20 for managing the module 10. Typically, in the case of an electric hybrid or fully electric powertrain application, the battery system of the drive machine may be equipped with a plurality of modules capable of storing and recovering electrical energy. The modules 10 may be arranged individually by the battery system 100. The module 10 comprises two connection terminals 15, 14, which can provide a determined voltage and current depending on the number of battery cells 13 assembled in series and / or in parallel. For example, the module 10 of the single battery cells 13 comprises two rows 11, 12 or more rows assembled in parallel, each row consisting of single battery cells 13 connected in series.

[0040] The electrochemical battery system 100 is of lithium, lithium-ion or other technology type. It should be understood that the battery system 100 and method according to the present invention can be applied to other applications and more generally to any electric or electronic device requiring an electrical supply consisting of an electrochemical energy storage device. Portable devices, computers, electrical supply facilities, satellites, and test and calibration workbenches are all examples of where the present invention can be applied, but the list is not exhaustive.

[0041] The method may be performed to determine an aging parameter of each of individually available battery modules of a battery system via electrical connection terminals of the battery modules, or to determine an aging parameter of the system as a whole or an aging parameter of an individually available single battery cell.

[0042] The control unit 20 is an integrated circuit electronic computer equipped with a processor and an electronic memory to execute the determination method for determining the aging parameters of the electrochemical battery module. The control unit 20 comprises: a measuring component (voltage and current sensor) for measuring the voltage and current of the module 10; a detection component for detecting the relaxation phase of the battery module (more precisely, detecting the zero current at the terminals of the module); and a calculation component for operating the calculation steps of the method described below. The control unit comprises a component for saving the state indicators of the module (for example, the open circuit voltage indicator OCV, the aging state indicator SOH and the charge state indicator SOC for managing the battery system during the service life of the vehicle). In addition, according to the application, the control unit is an on-board battery management system (for example, a BMS of a motor vehicle) or any type of battery control device (computer, remote management application or test bench computer).

[0043] According to the invention, the control unit 20 comprises a calculation module 21 for calculating the aging state of the battery module 10. To this end, the calculation module 21 comprises a memory in which a predetermined graph 22 is stored, which is able to provide a predetermined value of the aging state parameter of the battery as a function of a parameter representing the speed of autocatalysis carried out during the relaxation reaction of the battery. Thus, as a function of the value of the parameter representing the speed of autocatalysis, which is calculated during the method and provided as input data of the graph, the control unit reads the value of the aging state parameter.

[0044] More precisely, the parameter representative of the autocatalytic speed is calculated based on the parameter of the model function characterizing the voltage variation curve of the battery during the relaxation phase. The predetermined graph 22 is generated in advance during an aging operation of the battery. During this aging operation, for a plurality of known aging states, calculation steps of the parameter representative of the autocatalytic speed are performed. These calculation steps will be described more precisely in the description of the method below. The graph is configured to correspond to aging states between 30% and 100% of the nominal state of charge.

[0045] exist Figure 2 In the present invention, a determination method for determining the aging state of a battery will now be described. In a first step 200, a control unit of an accumulator controls a recharging or discharging operation. The current at the terminals of the accumulator is non-zero. As long as the current is non-zero, the method remains in step 200.

[0046] In step 201, when zero current is detected at the terminals of the energy storage device, the control unit controls a measuring step 202, which continuously measures a voltage signal Uacc(t) by means of a voltage sensor from time t0 for times greater than t0. The value of the voltage signal Uacc(t) is stored in a memory of the control unit.

[0047] Then, in a step 203 , the control unit continuously calculates a logarithmic signal SF of the voltage during the relaxation of the energy storage device as a function of the logarithmic value of the time elapsed from the instant t0 , which logarithmic signal SF depends on the voltage signal Uacc(t).

[0048] exist Figure 3, the value of the voltage signal Uacc(t) is shown in the upper part of the graph. The graph shows the voltage signal Uacc in volts measured at the terminals of the electrochemical accumulator according to the time t in seconds (from the moment t0 at the beginning of the relaxation phase after a recharging operation of the accumulator or a discharging operation of the accumulator) during the relaxation phase. As a schematic example, the measured voltage values ​​are represented by circles until an inflection point is detected at the moment t1. The points indicated by the crosses show the expected changes in the relaxation curve of the voltage Uacc. U0 is the value of the voltage Uacc at the beginning of the relaxation phase. Then, after a relaxation period of more than about 1000s, the electrochemical accumulator reaches thermodynamic equilibrium, from which the voltage Uacc tends to stabilize at the value of the open circuit voltage OCV.

[0049] In addition, the value of the logarithmic signal SF is schematically indicated in the lower part of the graph by a cross. The logarithmic signal SF is characterized in the form of an S-shaped curve due to the asymptotic behavior at infinity and the symmetry with respect to the inflection point PFL which can be detected at the time t1. The logarithmic signal SF characterizes the speed of change of the phase transition of the electrochemical reaction during the relaxation of the electrochemical energy storage device. The signal SF thus allows the determination of the autocatalytic speed, which depends on the aging state of the battery.

[0050] Furthermore, as long as the current at the terminals of the electrochemical energy accumulator is zero, the measuring step 202 is carried out until thermodynamic equilibrium is reached or until a point of inflection is detected. The latter variant makes it possible to reduce the calculation time for determining the logarithmic signal SF by utilizing the symmetry of the logarithmic signal with respect to the point of inflection.

[0051] More precisely, at least three variants of the logarithmic signal SF can be calculated by the control unit. According to a first representation f1 of the voltage signal Uacc(t), the logarithmic signal SF is a representation of the voltage signal Uacc(t) in logarithmic form according to the time value, wherein

[0052] Formula 5

[0053] f1:log(t)→Uacc(t).

[0054] According to another variant, the logarithmic signal SF is a second representation f2 of the voltage signal Uacc(t) in the first logarithmic form as a function of time and in logarithmic form, where:

[0055] Formula 6

[0056] f2:log(t)→log(Uacc(t)).

[0057] Alternatively, according to another variant, the logarithmic signal SF is a third representation f3 in logarithmic form according to the time value of the voltage signal Uacc(t) in the second logarithmic form, wherein:

[0058] Formula 7

[0059] f3: log(t)→log(log(Uacc(t))).

[0060] Function f3 is preferably selected during calculation step 203 because the logarithmic signal SF of the voltage signal is proportional to the volume of the reactant undergoing a phase transition during relaxation.

[0061] Furthermore, in order to verify the presence of an inflection point on the curve of the logarithmic signal SF, the control unit calculates the sign of the second derivative of the logarithmic signal SF in step 203. Of course, other calculation variants of inflection point detection are also possible.

[0062] Then, in step 204 of the method, the control unit saves in a memory the value of the measured voltage signal Uacc(t), the value of the logarithmic signal SF of the voltage signal and, if necessary, the value of the sign of the second-order derivative.

[0063] In step 205, the control unit continuously verifies whether the logarithmic signal SF has an inflection point. This detection is performed by verifying the sign of the second derivative. In the event of a change of sign, the control unit has detected an inflection point. As long as no inflection point is detected, the control unit continuously implements the successive steps 202, 203, 204. Furthermore, when the control unit detects a non-zero current in step 211, the method terminates and returns to step 200.

[0064] refer to Figure 3 At time t1, the control unit detects the inflection point PFL. In step 206, the control unit stores the values ​​of the voltage signal Uacc and the logarithmic signal SF for the detected inflection point in a memory.

[0065] Then, in step 207, the method comprises a calculation step of a model function MSF of the logarithmic signal SF according to the value of the logarithmic signal SF. The model function is Figure 3 Indicated by dotted line.

[0066] The calculation step 207 involves adjusting the parameters of a model function of an S-shape (e.g. a model function of the “sigmoid”, hyperbolic tangent, inverse hyperbolic sine type, denoted g) based at least on the value of the logarithmic signal SF. In fact, during relaxation, the main reaction is dominated by a phase transition in the active species, i.e. an autocatalytic nucleation reaction, which is characterized by a “sigmoid” speed. When the control unit aims at building a model to identify the parameters of the “sigmoid” speed of the autocatalytic reaction, the “sigmoid” speed is characterized by the phase transition in the active species, i.e. an autocatalytic nucleation reaction. ” is equivalent to the hyperbolic tangent function.

[0067] The control unit applies a control method by means of parameter optimization by reducing the squared error between the measured logarithmic signal SF and the model function g. As a non-limiting example, the adjustment method utilizes a parameter optimization algorithm known to those skilled in the art, such as the simplex method cited in the reference "Lagarias, JC, JA Reeds, MH Wright, and PE Wright," Convergence Properties of the Nelder-Mead Simplex Method in Low Dimensions." SIAM Journal of Optimization. Vol. 9, Number 1, 1998, pp. 1 12-147," or the least squares method - for example, the method cited in the reference "Marquardt, D." An Algorithm for Least-squares Estimation of Nonlinear Parameters." SIAM Journal Applied Mathematics, Vol. 11, 1963, pp. 431-441.", or through a model function of a data analysis software, which is equivalent to the "Spline" function of the Matlab software.

[0068] Next, a preferred variation of the calculation step 207 is described, which includes the following consecutive sub-steps. In order to reduce the number of parameters to be determined for the parameter adjustment calculation, the calculation 207 includes a first sub-step, which translates the measured logarithmic signal SF to make the inflection point PFL centered on the origin coordinate (0, 0) of the calculation coordinate system. The advantage of this step is that the number of parameters to be calculated for the model function is reduced, which is two here.

[0069] Then, the second sub-step involves calculating the adjustment parameters of a model function g based on the translated logarithmic signal SF, said model function being symmetrical with respect to the center of the coordinate system under investigation (with respect to the coordinates 0,0) and being S-shaped. As an example, the model function g is one of the following functions:

[0070] Formula 8

[0071] g: x→a*tanh(b*x),

[0072] or:

[0073] Formula 9

[0074] g:x→a*arcsinh(b*x),

[0075] or:

[0076] Formula 10

[0077] g:x→a(sigmf(b*x)-0.5),

[0078] Where a and b are the parameters to be adjusted in the calculation step 207, tanh is the hyperbolic tangent function, arcsinh is the inverse hyperbolic sine function, and sigmf is the "sigmoid" function. Figure 3 Above, the model function g shown is a hyperbolic tangent function.

[0079] Furthermore, according to a variant of the method, the logarithmic model function MSF is calculated symmetrically with respect to the inflection point PFL from the measured values ​​of the logarithmic signal before the moment of the inflection point. In fact, based on the value of the inflection point and the symmetry characteristics of the logarithmic signal SF, it is possible to estimate the variation of the curve SF after the inflection point without having to reach thermal equilibrium.

[0080] Other variations of the calculation step 207 are possible. For example, a calculation method can be cited which involves determining, during a first sub-step, symmetrically with respect to the inflection point PFL, a finite point of the logarithmic signal SF after the instant t1 of the inflection point. Then, a second calculation step involves calculating the model function of the logarithmic signal over the instants from 0 to infinity by interpolation-extrapolation of the points measured before t1 and the points calculated symmetrically after t1.

[0081] Another variant involves measuring the voltage until the thermodynamic equilibrium of the battery, and involves calculating the model function MSF based on samples of the voltage signal and the logarithmic signal. During the execution of this variant, the detection step 205 of detecting the inflection point and the saving step 206 of saving the inflection point are not performed, and the method switches from step 204 to 207 when thermodynamic equilibrium is reached (for example after a period of one hour).

[0082] Then, during step 208 of the method, the control unit calculates a parameter Vac representing the speed of change of the model function MSF of the logarithmic signal SF during the relaxation period. This speed of change represents the speed of autocatalysis of the battery during the relaxation phase. According to the model function g described above, the parameter Vac is calculated according to the following relationship:

[0083] Formula 11

[0084] Vac=a*b,

[0085] a and b are the adjusted parameters of one of the functions g described above during calculation step 207. In practice, the product of the adjusted parameters characterizes the change in the autocatalytic speed and thus the aging state of the electrochemical energy storage device.

[0086] Then, according to step 209 of the method, the control unit estimates the aging parameter SOH of the electrochemical energy storage device by reading the value of a predetermined pattern stored in a memory of the control unit. This value is read as a function of the parameter Vac calculated in step 208 .

[0087] Finally, according to the method, in the next step 210 , when the control unit detects that the current is non-zero, the method returns to the initial step 200 , and if the current is zero, the method returns to step 202 .

[0088] Figure 4 A predetermined graph intended to be stored in the memory of the control unit is schematically shown. The graph relates a predetermined value of a parameter Vac representative of the autocatalytic speed to a predetermined value of the state of aging of the battery. The relationship between the parameter Vac and the parameter SOH is a linear function that can be exploited by the method to determine the value of the state of aging based on the value of the parameter Vac.

[0089] For a known aging state of the battery, the graph is generated by executing steps 202 to 209 of the method. The graph is then stored in a memory of a calculation module of the onboard battery system for calculating the aging state.

Claims

1. A method for determining a first parameter SOH, which is representative of the state of aging of an electrochemical cell of a battery system, the method being performed by a control unit of the battery system and comprising the following consecutive steps: - continuously measuring (202) the voltage signal Uacc(t) at the terminals of the battery during relaxation as a function of the elapsed time, - estimating (209) a first aging state parameter, said first aging state parameter being dependent on said measured voltage signal, It is characterized in that The determination method further comprises: - continuously calculating (203) a logarithmic signal SF of the voltage during the relaxation of the battery (10) from values ​​in logarithmic form of the time elapsed since a first instant (t0), the logarithmic signal being dependent on the voltage signal Uacc(t), - a step of calculating (207) a model function MSF of said logarithmic signal SF as a function of the value of said logarithmic signal SF, - calculating (208) a second parameter Vac, said second parameter representing the speed of change of the model function MSF of said logarithmic signal SF during the relaxation period, said speed of change representing the autocatalytic speed of said battery during the relaxation phase, - estimating (209) the first aging status parameter according to the second parameter Vac and a predetermined graph, the graph being stored in a memory of the control unit and providing a predetermined value of the first aging status parameter according to the value of the second parameter Vac, The predetermined pattern is generated during an aging phase of the electrochemical cell and provides a plurality of predetermined values ​​of the aging state as a function of a plurality of predetermined values ​​of the second parameter.

2. The determination method according to claim 1, characterized in that: The calculation (207) of the model function involves adjusting parameters of the model function MSF, and the second parameter Vac is calculated based on the adjusted parameters of the model function.

3. The determination method according to claim 2, characterized in that: The model function MSF is a function g of any one of the following relations: g:x→a*tanh(b*x), or g:x→a*arcsinh(b*x), or g:x→a*sigmf(b*x)-0.5, Where a and b are adjusted parameters, and the second parameter Vac is calculated according to the following relationship: Vac=a*b.

4. The determination method according to any one of claims 1 to 3, characterized in that: The determination method further comprises a step of detecting (205) an inflection point of the logarithmic signal SF of the voltage, and calculating a model function MSF of the logarithmic signal symmetrically with respect to the inflection point based on a measurement value of the logarithmic signal before the inflection point.

5. A control unit of a battery system, the battery system comprising an electrochemical battery, characterized in that: The control unit also includes a memory storing a predetermined graph, the predetermined graph being suitable for providing a predetermined value of a first aging state parameter of the battery according to a predetermined value of a second parameter Vac, the second parameter representing the autocatalytic speed of the battery during the relaxation phase, and the control unit includes a component for implementing a determination method according to any one of claims 1 to 4.

6. A motor vehicle comprising a battery system comprising a control unit according to claim 5.

Images