Method and device for detecting a failure of a Lithium-Ion battery

Abstract

Method (100) for detecting a failure of a Lithium-Ion battery. The method comprises, for at least one battery relaxation phase, and for each battery cell: a collection (110) of several cell voltage measurements, a formation (120) of a relaxation signal in the form of a logarithm of a normalized cell voltage value, a decomposition (130) into empirical modes (EMD) of the relaxation signal into intrinsic components. The method (100) also comprises: a determination (150) of an incidence value representative of a battery health status from the intrinsic components obtained for the different cells, an evaluation (160) of a battery failure detection criterion based on the incidence value. Figure for the abstract: Fig. 2

Description

Title of the invention: Method and device for detecting a failure of a Lithium-Ion battery Scope of the invention

[0001] The present invention relates to the field of Lithium-Ion battery management. More particularly, a method and a device are proposed for detecting a failure of a Lithium-Ion battery. State of the art

[0002] The energy storage sector, particularly with regard to the use of lithium-ion batteries, is currently experiencing rapid growth. This growth is largely driven by the development of electric vehicles, which generally incorporate so-called first-life batteries (batteries with a state of health, or SOH, close to 100% at the start of their use), but also by the stationary energy storage sector, especially for grid support applications, using new batteries as well as so-called second-life batteries (batteries with a lower state of health at the start of their use, for example, an SOH below 80%). Second-life batteries used for grid regulation often come from the reuse of electric vehicle batteries.

[0003] Batteries undergo degradation during their lifespan. It is important to be able to detect the signs of this degradation in order to anticipate potential failures or safety issues. This also makes it possible to determine whether a battery can be reused for a second life. Reusing a battery for a second life is possible if it has not suffered too many failures during its first life.

[0004] There are different methods for detecting failures in lithium batteries.

[0005] Some methods are based on measurements of the mechanical properties of the battery (vibrations, acoustic waves, etc.). These measurements can be used to identify early signs of failure, such as cracks, delaminations, and manufacturing defects.

[0006] Some methods are based on measurements of the battery's thermal properties, such as temperature, thermal conductivity, and thermal capacity. These measurements can be used to identify abnormal battery heating. A drawback of these methods is that the failure is usually detected rather late.

[0007] Other methods are based on measurements of the battery's electrochemical properties, such as voltage, current, resistance, or capacity. In particular, many methods are based on voltage signals measured at the battery. Examples include US patent applications 2017 / 146608 Al, US 2022 / 0381849 Al and EP 3324197 Al.

[0008] However, the reliability of existing methods is not always entirely satisfactory. Furthermore, battery management systems (BMS) sometimes provide a rather limited amount of data, and it is necessary to find a sufficiently simple fault detection method that can be applied to the majority of existing battery management systems. Description of the invention

[0009] The present invention aims to overcome all or part of the drawbacks of the prior art, particularly those described above, by providing a method for detecting failure in a Lithium-Ion battery. The proposed method offers good reliability and can be implemented by most existing battery management systems.

[0010] To this end, and according to a first aspect, a method is proposed for detecting a failure of a multi-cell Lithium-Ion battery. The method comprises, for at least one relaxation phase following a battery discharge phase: - for each cell: • a collection of several voltage measurements at the cell level during said relaxation phase, • the formation of a relaxation signal by calculating, for each voltage measurement collected during said relaxation phase, a logarithm of a normalized value of the voltage measurement, • a decomposition into empirical modes of the relaxation signal in order to obtain a representation in the form of a sum of a residual signal and one or more intrinsic components, - a determination of an incidence value representative of the battery's health status, for said relaxation phase, based on the intrinsic components thus obtained for the different battery cells, - an evaluation of a battery failure detection criterion based on the incidence value of said relaxation phase.

[0011] According to a first example, the proposed solution makes it possible to detect a battery failure at the moment the failure occurs. The failure may, for example, be due to abnormal battery aging or abnormal environmental conditions. When a failure is detected, measures can be taken to optimize the battery's lifespan or to protect it.

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020] In another example, the proposed solution allows for the assessment of a battery's health at the end of its first life, based on failures detected retrospectively. Failures experienced by the battery during its first life can be identified using incidence values ​​recorded throughout that period. This mapping of failures experienced by the battery during its first life can help determine whether it is feasible to reuse the battery for a second life, possibly with a different application. In particular modes of implementation, the method may further include one or more of the following characteristics, taken individually or in all technically possible combinations. In specific implementation modes, for a cell with index , a value of the relaxation signal at a given time z can be written in the form: Xk(t) = -ln(j \ \ ;vc ) / where is a voltage measured across the terminals of the cell whose index is a voltage measured across the battery terminals, Nc is the number of battery cells, and hi is the natural logarithm operator. In specific implementation methods, determining the incidence value for the relaxation phase under consideration involves: - for each intrinsic component of a cell, a calculation of the energy of said intrinsic component, - for each cell, a calculation of the cell's energy corresponding to a sum of the energies of the cell's intrinsic components, - a calculation of an average energy corresponding to an average of the energies of the battery cells, - a determination of the incidence value from the average energy calculated for the relaxation phase considered. In specific implementation modes, the evaluation of a detection criterion involves a comparison of the incidence value with a predetermined incidence threshold, for one or more consecutive relaxation phases. In particular implementation modes, the evaluation of a detection criterion involves a comparison of the incidence value with one or more incidence values ​​from previous relaxation phases. In particular implementation modes, the evaluation of a detection criterion includes a check whether the incidence values ​​of several consecutive relaxation phases are increasing for a predetermined number of consecutive relaxation phases.

[0021] In particular embodiments, the evaluation of a detection criterion involves calculating a slope of a linear interpolation performed for a set of several incidence values ​​corresponding to several consecutive relaxation phases, and comparing the slope with a predetermined slope threshold.

[0022] In particular embodiments, the method further comprises, for each relaxation phase, a statistical reliability analysis of elements enabling the determination of the incidence value, and a filtering of at least one of the following elements if it is deemed unreliable: - an intrinsic component of a cell, - a battery cell, - the incidence value.

[0023] In particular embodiments, the statistical reliability analysis includes, for each intrinsic component of the relaxation signal of a cell, a calculation of an entropy of the intrinsic component.

[0024] In particular embodiments, the statistical reliability analysis includes, for each cell, a calculation of an entropy from a sum of the intrinsic components of the cell's relaxation signal.

[0025] In specific implementation modes, the statistical reliability analysis includes: - a comparison of the energy of an intrinsic component of a cell with a threshold or with the energies of the other intrinsic components of the cell, and / or - a comparison of the energy of one cell with a threshold or with the energies of the other cells in the battery, and / or - a comparison of the incidence value determined for the relaxation phase considered with a threshold or with other incidence values ​​calculated for previous relaxation phases.

[0026] In particular embodiments, when a failure is detected, the method further includes a check whether the detected failure is related to the environment in which the battery has been operating.

[0027] In particular embodiments, the verification whether the detected failure is related to the environment includes a comparison, for a given period including said at least one relaxation phase, of incidence values ​​determined for the battery during said period with incidence values ​​determined for at least one other battery subjected to the same environment during said period.

[0028] In particular embodiments, the verification whether the detected failure is related to the environment includes a comparison, for a given period including said at least one relaxation phase, of environmental measurements carried out during said period with a predetermined threshold.

[0029] In particular embodiments, when a fault is detected, the method (100) includes storing the occurrence of the fault for a subsequent analysis of the battery's health status.

[0030] In particular embodiments, the discharge phase preceding the relaxation phase has a discharge depth of at least 60%, or preferably at least 90%.

[0031] According to a second aspect, a device is proposed for detecting a failure in a multi-cell Lithium-Ion battery. The device comprises: - a measurement system configured to provide voltage measurements taken at each cell during at least one relaxation phase following a battery discharge, - a computing unit connected to the measurement system.

[0032] The computing unit is configured to implement the method according to any one of the preceding implementation modes.

[0033] According to a third aspect, a battery management system, or BMS, is proposed, comprising a device as described above. Presentation of the figures

[0034] The invention will be better understood upon reading the following description, given by way of non-limiting example, and made with reference to the following figures:

[0035] [Fig-1] a graph representing the evolution over time of the voltage measured across the terminals of a Lithium-Ion battery, as well as the voltages measured across the terminals of each cell of the battery,

[0036] [Fig.2] a schematic representation of the main steps of an example of implementation of the method according to the invention for detecting a failure of a Lithium-Ion battery,

[0037] [Fig.3] a graph representing a relaxation signal, as well as the residual signal resulting from the empirical decomposition of this relaxation signal,

[0038] [Fig.4] a graph representing four empirical components obtained by the decomposition in empirical mode of the relaxation signal represented in [Fig.3],

[0039] [Fig.5] a particular method of implementing the determination of an incidence value representative of a battery health status, for a given relaxation phase,

[0040] [Fig.6] a graph representing incidence values, associated with the dates of relaxation phases for which they were determined, for different batteries subjected to different environmental conditions and different uses over a given period,

[0041] [Fig.7] a first graph representing the evolution over time of the value the impact of a battery and the evolution over time, during the same period, of the temperature experienced by the battery,

[0042] [Fig.8] a second graph representing the evolution over time of the the incidence value of a battery, as well as the evolution over time, during the same period, of the temperature experienced by the battery,

[0043] [Fig.9] a schematic representation of a device according to the invention allowing to detect a failure in a Lithium-Ion battery.

[0044] In these figures, identical reference numerals from one figure to another designate identical or analogous elements. For clarity, the elements shown are not necessarily to the same scale, unless otherwise stated. Detailed description of the invention

[0045] To detect a failure in a Lithium-Ion battery, the method according to the invention uses the voltage signals from the different cells of the battery. The voltage signals are measured during relaxation phases of the battery. A relaxation phase follows a more or less complete discharge phase of the battery.

[0046] Figure 1 is a graph showing the evolution over time of the voltage measured across the terminals of a lithium-ion battery, as well as the voltages measured across each cell of the battery. In the example considered and illustrated in Figure 1, the battery has twelve cells. Figure 1 represents four relaxation phases. Each relaxation phase follows a battery discharge phase and precedes a battery recharge phase. However, there is nothing preventing the inclusion of a relaxation phase between two discharge phases. During each relaxation phase, the voltage signals of the different cells can be observed to take on different values ​​(they follow a similar trend but with different values). During the discharge and recharge phases, the voltage signals of the different cells are essentially identical to the battery voltage signal.

[0047] From a physics perspective, the work published in the document "Anisotropic ionic transport properties in solid PEO based electrolytes," by R. Jeanne-Brou et al., makes it possible to link the evolution of voltage over time to the diffusion of electroactive species towards the charge transfer sites. The state of health (SOH) The characteristics of a battery can have an impact on this evolution. This is why the inventors decided to focus on the relaxation voltage of the cells of a Lithium battery, and more specifically on the logarithm of a normalized voltage value.

[0048] Fig. 2 schematically represents the main steps of an example of implementation of a method 100 according to the invention for detecting a failure of a Lithium-Ion battery.

[0049] As illustrated in [Fig.2], method 100 comprises, for at least one relaxation phase: - for each cell: • a collection of 110 several voltage measurements at the cell level during the relaxation phase, • a 120 formation of a relaxation signal by calculating, for each voltage measurement collected during the relaxation phase, a logarithm of a normalized value of the voltage measurement, • a decomposition 130 into empirical modes of the relaxation signal in order to obtain a representation in the form of a sum of a residual signal and one or more intrinsic components, - a determination of a representative incidence value indicating the battery's health status, for the relaxation phase considered, based on the intrinsic components thus obtained for the different battery cells, - an evaluation 160 of a battery failure detection criterion based on the incidence value of the relaxation phase.

[0050] These different steps can be implemented for each relaxation phase following a battery discharge phase during its lifetime. However, in a variant, nothing would prevent these steps from being implemented only for a subset of the relaxation phases. In particular, the inventors have observed that the longer the discharge phase preceding the relaxation phase, the more relevant the incidence value obtained for that relaxation phase. Therefore, it could be considered, for example, to consider only the relaxation phases that follow a discharge phase with a depth of discharge (DOD) of at least 60%. A depth of discharge of 60% means that the battery's state of charge (SOC) is 40% (SOC = 1 - DOD).Even more optimally, we can consider only the relaxation phases that follow a discharge phase with a discharge depth of at least 90%.

[0051] Voltage measurements during the relaxation phase are, for example, performed by a battery management system (BMS) connected to the battery cells. The measurements are, for example, taken with an acquisition frequency of between five and sixty seconds, for an acquisition duration of thirty to sixty minutes. However, there is nothing preventing the measurements from being taken with a different acquisition frequency and / or for a different acquisition duration. It is advantageous to use between forty and two hundred measurements per relaxation phase (using a larger number of measurements does not necessarily imply a significant improvement in the method, while using a smaller number of measurements may limit the method's performance).

[0052] The relaxation signal formation step 120 comprises, for each voltage measurement collected during the relaxation phase, a calculation of a logarithm of a normalized value of the voltage measurement. For example, the relaxation signal of a cell with index & can be written in the form:

[0055] / V / / ) . Xk(t) = - In (¾ /

[0054] where y^ is the voltage measured across the terminals of the cell with index k, is the voltage measured across the terminals of the battery, Nc is the number of cells in the battery, and In is the natural logarithm operator.

[0055] In variations, nothing would prevent the relaxation signal from being defined differently, in particular by changing the way the voltage measured across a cell is normalized, or by using a logarithm in a different base (for example, in a base equal to Nc). For normalization, one could, for example, consider using a polynomial characteristic of the average relaxation voltage of the battery (for example, of the form — Clt + b^) or simply using an average value (UB) of the battery voltage during the relaxation phase.

[0056] In step 130, the relaxation signal is decomposed according to an empirical mode decomposition (EMD for "Empirical Mode Decomposition" in English).

[0057] Empirical mode decomposition consists of decomposing a signal into a sum of functions, in a similar way to what Fourier series decomposition or wavelet decomposition does.

[0058] One of the particularities of decomposition into empirical modes is that the basis of functions into which the signal is decomposed is not known a priori, but is constructed adaptively according to the properties of the signal.

[0059] The decomposition into empirical modes corresponds to the first part of the Hilbert-Huang transform (HHT for "Hilbert-Huang Transform" in English). Empirical mode decomposition involves decomposing a signal into a sum of a residual signal and intrinsic mode functions (IMFs). In this application, these intrinsic mode functions are also referred to as "intrinsic components".

[0060] As previously stated, the intrinsic components are not defined analytically. Rather, they are determined adaptively according to the properties of the signal.

[0061] An intrinsic component (IMF) resulting from an empirical mode decomposition (EMD) must satisfy the following requirements: - the number of extrema (i.e., the sum of the number of local maxima and the number of local minima) and the number of zero crossings of the intrinsic component must be equal or differ by a maximum of one; - at every point of the intrinsic component, the average value of the envelope defined by the local maxima and of the envelope defined by the local minima is zero.

[0062] A signal decomposed by EMD can then be written in the form:

[0064] In this expression, corresponds to the residual signal, N is the number of intrinsic components of the EMD decomposition, and ci is the intrinsic component with index . Each successive intrinsic component contains oscillations of a frequency lower than that of the preceding one. The residual signal corresponds to a general trend of the signal

[0065] The decomposition into empirical modes comprises a succession of sifting processes. The first sifting process takes the signal s(t) directly as input. The sifting process consists of identifying all the local extrema of the input signal and linking the local maxima and minima, respectively, by interpolation using cubic splines, in order to obtain an upper envelope and a lower envelope, respectively. An average between the upper and lower envelopes can then be calculated and subtracted from the input signal. If the intermediate signal obtained (subtracting the input signal from the average of the upper and lower envelopes) is not an intrinsic component, the sifting process is repeated on the intermediate signal (which thus becomes the input signal for a new sifting process) until an intrinsic component is obtained.The sieving processes are repeated until the last intrinsic component is obtained, i.e. for example until the signal. The intermediate signal becomes monotonic or consists of only one local extremum. The remaining signal then corresponds to the residual signal.

[0066] A stopping criterion can be defined for the sieving process. For example, the stopping criterion is satisfied if the standard deviation between the results of two successive sieving processes is less than or equal to a predetermined stopping threshold. The stopping threshold can typically be between 0.2 and 0.3.

[0067] The document "The empirical mode decomposition and the Hilbert spectrum for non-linear and non-stationary time series analysis", Norden E. Huang et al., Proceedings of the Royal Society of London Series A (1998) 454, p. 903-995, describes in detail the decomposition into empirical modes, particularly in its sections 4 and 5.

[0068] Algorithms for decomposing into empirical modes are available in programming libraries, for example in MATLAB or Python.

[0069] The graphs in Figures 3 and 4 represent an example of the decomposition into empirical modes of a relaxation signal from one of the twelve cells of the battery. More specifically, [Fig. 3] represents the cell relaxation signal 31, and the residual signal 32 from the EMD decomposition of this signal 31. [Fig. 4] represents four intrinsic components (components Ci to C4, represented respectively by curves 41 to 44) obtained by the EMD decomposition of the signal 31.

[0070] It should be noted that, in other examples, a different number of intrinsic components could be obtained. In particular, the number of intrinsic components obtained during the EMD decomposition of the relaxation signal can vary from one cell to another. However, the number of intrinsic components generally remains below five. It is advantageous to set the stopping threshold at a relatively low level, on the order of 0.2, to extract a maximum of information from the relaxation signal. Using a lower stopping threshold imposes particularly long computation times.

[0071] The intrinsic components obtained in step 130 are then used in step 150 to determine an incidence value representative of the battery's health status for the relaxation phase considered.

[0072] Different methods can be considered to determine the incidence value associated with the relaxation phase considered.

[0073] For example, and as illustrated in [Fig. 5], the determination 150 of an incidence value may involve: - for each intrinsic component of a cell, a calculation of 151 of the energy of said intrinsic component, - for each cell, a calculation of 152 of a cell energy corresponding to a sum of the energies of the intrinsic components of the cell, - a calculation of 153 of an average energy corresponding to an average of the energies of the battery cells, - a determination 154 of the incidence value from the average energy calculated for the relaxation phase considered.

[0074] The energy Ej of an intrinsic component C{ corresponds, for example, to the integral of the square of the amplitude of the intrinsic component C, over the acquisition time of the relaxation phase considered: 100751 E.^c^dt

[0076] The energy Ek of a cell with index & for the relaxation phase considered can then be written in the form:

[0077]

[0078] where Mk is the number of intrinsic components obtained during the EMD decomposition of the relaxation signal associated with the cell of index k for the relaxation phase considered.

[0079] It should be noted that nothing would prevent, in a variant, calculating the energy Ek of a cell of index k by summing the energies of only a subset of the intrinsic components obtained by the EMD decomposition (for example by considering only a predefined maximum number of the first intrinsic components obtained by the EMD decomposition or, as will be seen later, by discarding certain intrinsic components deemed unreliable on a statistical point of view).

[0080] The average energy of the battery cells, for the relaxation phase considered, can then be written in the form:

[0081] Nc f^k

[0082] The incidence value of the relaxation phase considered can then correspond to the average energy Ë.

[0083] It should be noted that, in alternative embodiments, nothing would prevent calculating the incidence value in a different way, for example, by weighting the average energy according to the depth of discharge preceding the relaxation phase. Nor would anything prevent calculating the average energy as a weighted average of the energies of the different cells that make up the battery. The particular choice of a method for calculating the incidence value is merely a variant of the invention.

[0084] Generally, in the invention the incidence value is calculated as an average of the intrinsic energies of the battery cells, the intrinsic energy of a cell corresponding to a sum of energies of the intrinsic components of the cell. More broadly, it is possible to calculate the incidence value using any type of central tendency indicator of the intrinsic energies of the battery cells (for example, a median of the intrinsic energies, an average of the minimum and maximum values ​​of the intrinsic energies, a truncated average of the intrinsic energies, etc.).

[0085] The incidence value determined in step 150 can be used in real time to detect a battery failure at a time when the failure occurs (or at a time close to it).

[0086] According to another example, the incidence value determined for each relaxation phase can be stored in a computer file. Such a file can form a map (or digital passport) of the battery, providing information on any failures the battery may have experienced during its initial lifespan. In this computer file, each incidence value is associated with information about the time of occurrence of the relaxation phase to which it corresponds. This information could, for example, correspond to a date of occurrence, including the day and possibly the time at which the relaxation phase took place. This information could also simply correspond to a relaxation phase number (this number being incremented with each new relaxation phase observed by the cell during its initial lifespan).

[0087] The inventors have observed that the incidence values ​​tend to take on relatively high values, or they tend to increase over time, when the battery is subjected to unfavorable environmental conditions (for example, when it is subjected to particularly high or particularly low temperatures) or when the battery is subjected to an insufficient frequency of full recharges.

[0088] Generally, a battery management system rebalances the battery cells at the end of the charging cycle, when the state of charge (SOC) is close to 100%. When the battery is not fully recharged, rebalancing of the battery cells cannot occur. When battery cell rebalancing is not performed frequently enough, the battery may suffer from premature aging.

[0089] It is thus possible to define a criterion for detecting a battery failure from one or more incidence values ​​determined for one or more relaxation phases.

[0090] By way of example, [Fig.6] represents incidence values, associated with the dates of the relaxation phases for which they were determined, for different Lithium-Ion batteries subjected to different environmental conditions and different uses during a given period (between October 2021 and June 2023).

[0091] The battery denoted "la", whose incidence values ​​are represented on the graph of [Fig.6] by circles, was tested in a climatic chamber at 45°C with an initial use similar to that of an electric vehicle battery with a sequence of driving phases and daily full recharges.

[0092] The battery marked "1b", whose incidence values ​​are represented on the graph in [Fig.6] by crosses, was tested in a climatic chamber at 45°C with a second use which is similar to the regulation of electrical network frequency, with full recharges much less frequently (only one full recharge per week).

[0093] The battery marked "2a", whose incidence values ​​are represented on the graph of [Fig.6] by diamonds, was tested in a climatic chamber at 25°C with the first use.

[0094] The battery marked “2b”, whose incidence values ​​are represented on the graph in [Fig.6] by asterisks, was tested in a climatic chamber at 25°C with the second use.

[0095] The battery marked "3a", whose incidence values ​​are represented on the graph of [Fig.6] by squares, was tested with the first use in outdoor climatic conditions (high temperatures in summer and cold in winter).

[0096] The battery marked “3b”, whose incidence values ​​are represented on the graph in [Fig.6] by stars, was tested in outdoor climatic conditions with the second use.

[0097] It appears in [Fig. 6] that the batteries tested under the second usage (with a single full charge per week) generally exhibit fairly long periods during which the incidence values ​​are strictly increasing. Over time, they end up exhibiting relatively high incidence values. In contrast, the batteries tested under the first usage (with a full charge daily) generally exhibit lower or decreasing incidence values. Batteries subjected to adverse environmental conditions also generally exhibit higher incidence values ​​or longer periods of growth.

[0098] In step 160, different criteria can be used, individually or in combination, to detect a battery failure from the incidence value determined for one or more relaxation phases.

[0099] According to a first example, it is possible to compare the incidence value with a predetermined incidence threshold for one or more relaxation phases. If the incidence value is higher than the threshold for the relaxation phase in question, or if it remains higher than the threshold for a predetermined number of consecutive relaxation phases, then an alert can be triggered.

[0100] According to a second example, it is possible to compare the incidence value with one or more incidence values ​​from previous relaxation phases. For example, if the incidence values ​​of several consecutive relaxation phases are increasing for a predetermined number of consecutive relaxation phases, then an alert can be raised.

[0101] According to a third example, for each new relaxation phase considered, it is possible to perform linear interpolation on the incidence values ​​determined respectively for the new relaxation phase and for each of a number of previous relaxation phases, and to compare the slope of the linear interpolation with a predetermined slope threshold. If the slope is greater than the slope threshold, then an alert can be raised.

[0102] A failure can be detected when a certain number of alerts are triggered. When a failure is detected during the battery's lifetime, corrective measures can be implemented to optimize battery life. For example, the frequency at which a full charge (SOC = 100%) is performed can be increased. Such measures help to increase the frequency of battery cell rebalancing. As another example, and where possible, an adverse environmental factor can be eliminated.

[0103] The different parameters to be taken into account for these different detection criteria used for step 160 (incidence threshold, slope threshold, duration of a growth period, number of relaxation phases to be considered, number of alerts, etc.) can be determined empirically in the laboratory.

[0104] It should be noted that many other variants are conceivable for defining a criterion for detecting a battery failure based on the calculated incidence values. The particular choice of a criterion is only one variant of the invention.

[0105] If a fault is detected, it is possible to record the occurrence of the fault in memory (for example, with the associated date and time). This allows, at the end of a given period, a summary of the number of faults detected during that period to be compiled, and the condition of the battery to be assessed at the end of that period (for example, at the end of the battery's first life). This assessment of the battery's condition can, for example, influence a decision between recycling or reusing the battery for a second life.

[0106] As mentioned previously, another option may be to record the incidence values ​​in memory and to detect battery failures after a given period (a posteriori), based on the stored incidence values. In this case, evaluation step 160 of the battery failure detection criterion is not performed at the end of a phase of relaxation, but it is carried out at a later date from the stored incidence value corresponding to said relaxation phase.

[0107] It may be advantageous to know if a detected failure is related to the environment in which the battery has been operating.

[0108] For this purpose, and as illustrated in [Fig.2], method 100 may include an optional verification step 170 if a detected failure is related to the environment in which the battery has been operating.

[0109] This verification 170 may include, in particular, a comparison, for a given period, of the incidence values ​​stored for the battery during that period with incidence values ​​determined and stored for at least one other battery subjected to the same environment during that period. If the incidence values ​​observed for one or more other batteries subjected to the same environment show a similar abnormal trend over a given period, then it is highly probable that this abnormal behavior is related to the environment (for example, due to an exceptional increase in temperature experienced by the different batteries during that period).

[0110] Alternatively, or in addition, verification 170 may include a comparison, for a given period, of environmental measurements taken and stored during that period with a predetermined threshold. These may include, for example, measurements of the temperature experienced by the battery.

[0111] By way of example, figures 7 and 8 illustrate the evolution over time of the incidence value of a battery (curve 33) as well as the evolution over time, during the same period, of the temperature undergone by the battery (curve 34).

[0112] In the example illustrated in [Fig. 7], the failure highlighted by the peak in the incidence value is most likely related to the environment because a similar peak is observed at the same time for the temperature experienced by the cell. Conversely, in [Fig. 8], the continuous increase in the incidence value does not appear to be related to the environment.

[0113] If a failure is related to the environment, it may be possible to take corrective action at the battery's environmental level, and the failure will potentially have a limited impact over time. A failure that is not related to the environment may be related to the battery's usage (in particular, the frequency of full battery charges). Here again, a corrective measure is possible (increasing the frequency of full battery charges). However, a failure may also be related to a manufacturing defect in the battery, or to premature degradation of at least one battery cell. Such a failure may require the battery to be made safe to prevent a risk of accident (thermal runaway, risk of fire or explosion, etc.).

[0114] As illustrated in [Fig. 2], the method 100 according to the invention may also include an optional step 140 of statistically analyzing the reliability of the elements used to determine the incidence value. This statistical reliability analysis 140 may, in particular, make it possible to estimate whether the energy of an intrinsic component of a cell has an aberrant value, or whether the intrinsic energy of a cell has an aberrant value. If so, the intrinsic component and / or the intrinsic energy deemed unreliable may be filtered out (i.e., not taken into account) in the determination 150 of the incidence value. Alternatively, it is possible to consider that the incidence value itself is not sufficiently reliable. In this case, the relaxation phase is discarded and is not considered for monitoring the battery's health status.

[0115] According to a first example, the statistical reliability analysis 140 may include, for each intrinsic component of the relaxation signal of a cell, a calculation of an entropy of that intrinsic component. For example, an intrinsic component whose entropy is too low (below a predetermined entropy threshold) is filtered out.

[0116] According to another example, the statistical reliability analysis 140 may include, for each cell, a calculation of an entropy of a sum of the intrinsic components of the cell's relaxation signal (for example, for the sum of all the intrinsic components obtained by the EMD decomposition, or for the sum of a predefined maximum number of the first intrinsic components obtained by the EMD decomposition). For example, a cell for which the sum of the intrinsic components has too low an entropy is filtered out.

[0117] Different methods for calculating entropy can be considered, such as a Shannon entropy calculation or a Kolmogorov entropy calculation. A Shannon entropy threshold between 0.25 and 0.5 can in particular be considered.

[0118] According to yet another example, the statistical reliability analysis 140 may include: - a comparison of the energy of an intrinsic component of a cell with a threshold or with the energies of the other intrinsic components of the cell, and / or - a comparison of the energy of one cell with a threshold or with the energies of the other cells in the battery, and / or - a comparison of the incidence value determined for the relaxation phase considered with a threshold or with other incidence values ​​calculated for previous relaxation phases.

[0119] Various statistical tests can be considered to compare these energies with each other, despite a relatively small amount of data, such as: Peirce test, Chauvenet test, Grubbs test, etc.

[0120] The statistical reliability analysis 140 may also include a combination of the examples shown above (the different criteria for filtering an intrinsic component, a cell, or even the incidence value may be used together).

[0121] Figure 9 schematically represents an example of an embodiment of a device 10 for detecting a failure of a Lithium-Ion battery 20. The device 10 includes, in particular, a memory 11, a measurement system 13, and a processing unit 12 connected to the memory 11 and the measurement system 13. The memory 11 can be used, in particular, to store calculated incidence values ​​and / or detected failure occurrences (for example, with the associated date and time) for a subsequent assessment of the battery's health status. If the only objective is real-time failure detection, this memory 11 is not essential.

[0122] The measurement system 13 is configured to provide voltage measurements taken at each cell 21 of the battery 20 during at least one relaxation phase following a battery discharge.

[0123] The calculation unit 12 is configured to implement method 100 according to any one of the implementation modes described above.

[0124] The incidence values ​​calculated for relaxation phases observed by battery 20 during its life can be stored in memory 11.

[0125] The device 10 may further include a sensor configured to measure at least one physical parameter of the environment in which the battery 20 operates (for example, a temperature sensor).

[0126] The calculation unit 12 can also be configured to obtain incidence values ​​from one or more other batteries subjected to the same environment as battery 20.

[0127] Communication between the measuring system 13 and the processing unit 12 (particularly for transmitting voltage measurements) can be implemented by wired or wireless means. The measuring system 13 and the processing unit 12 can form a single physical entity (i.e., they can be integrated into the same housing, for example, in a battery management system (BMS)). However, there is nothing to prevent the measuring system 13 and the processing unit 12 from each being part of a separate physical entity. It is even conceivable that the processing unit 12 could be integrated into a remote server configured to collect data from several batteries.

[0128] The foregoing description clearly illustrates that, through its various features and their advantages, the present invention achieves the stated objectives. In particular, monitoring the average intrinsic energy of the battery cells makes it possible to Detecting battery failures (the intrinsic energy of a cell corresponds to the sum of the energies of the intrinsic components of the cell's relaxation signal). Failure detection can be performed in real time, to take corrective measures if necessary to optimize battery life, or retrospectively, at the end of the battery's first life, to assess the feasibility of reusing the battery in a second life.

Claims

Demands

1. A method (100) for detecting a failure of a Lithium-Ion battery (20), said battery (20) comprising several cells (21), the method (100) comprising, for at least one relaxation phase following a discharge phase of the battery: - for each cell (21): • a collection (110) of several voltage measurements at the cell (21) during said relaxation phase, • a formation (120) of a relaxation signal by calculating, for each voltage measurement collected during said relaxation phase, a logarithm of a normalized value of the voltage measurement, • a decomposition (130) into empirical modes of the relaxation signal in order to obtain a representation in the form of a sum of a residual signal and one or more intrinsic components, - a determination (150) of an incidence value representative of a state of health of the battery (20), for said relaxation phase,from the intrinsic components thus obtained for the different cells (21) of the battery (20), - an evaluation (160) of a criterion for detecting a failure of the battery (20) as a function of the incidence value of said relaxation phase.

2. Method (100) according to claim 1 wherein, for a cell (21) of index k, a value of the relaxation signal at an instant f can be written in the form: \ \ Ne / / where is a voltage measured across the terminals of the cell (21) of index k, is a voltage measured across the terminals of the battery, Nc is the number of cells (21) of the battery (20), and ln is the natural logarithm operator.

3. Method (100) according to any one of claims 1 to 2 wherein the determination (150) of the incidence value for the relaxation phase considered comprises: - for each intrinsic component of a cell (21), a calculation (151) of an energy of said intrinsic component, - for each cell (21), a calculation (152) of a cell energy corresponding to a sum of the energies of the intrinsic components of the cell (21), - a calculation (153) of an average energy corresponding to an average of the energies of the cells (21) of the battery (20), - a determination (154) of the incidence value from the average energy calculated for the relaxation phase considered.

4. Method (100) according to any one of claims 1 to 3 wherein the evaluation (160) of a detection criterion involves a comparison of the incidence value with a predetermined incidence threshold, for one or more consecutive relaxation phases.

5. Method (100) according to any one of claims 1 to 4 wherein the evaluation (160) of a detection criterion includes a comparison of the incidence value with one or more incidence values ​​from previous relaxation phases.

6. Method (100) according to any one of claims 1 to 5 wherein the evaluation (160) of a detection criterion includes a check whether the incidence values ​​of several consecutive relaxation phases are increasing for a predetermined number of consecutive relaxation phases.

7. Method (100) according to any one of claims 1 to 6 wherein the evaluation (160) of a detection criterion comprises a calculation of a slope of a linear interpolation performed for a set of several incidence values ​​corresponding to several consecutive relaxation phases, and a comparison of the slope with a predetermined slope threshold.

8. A method (100) according to any one of claims 1 to 7 further comprising, for each relaxation phase, a statistical reliability analysis (140) of elements enabling the determination (150) of the incidence value, and filtering of at least one of the following if it is deemed unreliable: - an intrinsic component of a cell (21), - a cell (21) of the battery (20), - the incidence value.

9. Method (100) according to claim 8 wherein the statistical reliability analysis (140) comprises, for each intrinsic component of the relaxation signal of a cell (21), a calculation of an entropy of the intrinsic component.

10. Method (100) according to any one of claims 8 to 9 wherein the statistical reliability analysis (140) comprises, for each cell (21), a calculation of an entropy from a sum of the intrinsic components of the relaxation signal of the cell (21).

11. Method (100) according to any one of claims 8 to 10 wherein the statistical reliability analysis (140) comprises: - a comparison of an energy of an intrinsic component of a cell (21) with a threshold or with the energies of the other intrinsic components of the cell (21), and / or - a comparison of an energy of a cell (21) with a threshold or with the energies of the other cells (21) of the battery (20), and / or - a comparison of the incidence value determined for the relaxation phase considered with a threshold or with other incidence values ​​calculated for previous relaxation phases.

12. Method (100) according to any one of claims 1 to 11 wherein, when a failure is detected, the method (100) further includes a check (170) whether the detected failure is related to the environment in which the battery (20) has been operating.

13. Method (100) according to claim 12 wherein the verification (170) whether the detected failure is related to the environment comprises a comparison, for a given period including said at least one relaxation phase, of incidence values ​​determined for the battery (20) during said period with values of incidence determined for at least one other battery (22) subjected to the same environment during said period.

14. Method (100) according to any one of claims 12 to 13 wherein the verification (170) whether the detected failure is related to the environment comprises a comparison, for a given period including said at least one relaxation phase, of environmental measurements carried out during said period with a predetermined threshold.

15. Method (100) according to any one of claims 1 to 14 wherein, when a failure is detected, the method (100) includes a memorization of the occurrence of the failure, for a post-hoc analysis of the health status of the battery.

16. Method (100) according to any one of claims 1 to 15 wherein the discharge phase preceding the relaxation phase has a discharge depth of at least 60%, or preferably at least 90%.

17. Device (10) for detecting a failure of a Lithium-Ion battery (20), said battery (20) comprising several cells (21), said device (10) comprising: - a measuring system (13) configured to provide voltage measurements taken at each cell (21) during at least one relaxation phase following a discharge of the battery, - a computing unit (12) connected to the measuring system (13), said computing unit (12) being configured to implement the method (100) according to any one of claims 1 to 16.

18. Battery management system, or BMS, comprising a device (10) according to claim 17.

Images