Detection of an anomaly in the variation of an estimate of a charge level of an electrical energy storage module

Abstract

The present disclosure is based on the implementation of a coulombic counting method from time-stamped current measurement data for a calculation of equivalent SOC, then of a variation of equivalent SOC over a time interval, which corresponds to the image of the current delivered during this time interval. The latter is then compared to one or more thresholds to qualify this variation as normal, or abnormally slow (“SOC freeze”) or fast (“SOC jump”). Abstract figure: Figure 2

Description

Title of the invention: Detection of an anomaly in the variation of an estimate of a charge level of an electrical energy storage module Technical field

[0001] The present disclosure relates to the field of electrochemical energy storage devices, and more particularly to the diagnosis of batteries or accumulators. Prior art

[0002] The control of electrical energy storage systems using batteries, for example Lithium Ion batteries, is partly based on the estimation of state variables, such as the state of charge of the battery for example. This state of charge, also called SOC for "State Of Charge", is a unitless parameter, generally expressed as a percentage. The SOC is defined as the ratio of the actual available charge Qt of the tested element to the maximum charge Qmax>t of the same element in its current state of health. The equation below represents this parameter:

[0003] [Math.l] Q SOC = x 100

[0004] A lack of reliability of this key indicator is a risk factor for the effective control of electrical energy storage systems, also called BESS (for the English “Battery Energy Storage System”). However, phenomena of SOC jump (i.e., a sudden and unpredictable variation of the SOC indicator) and frozen SOC (i.e., maintaining the same SOC value while the battery is nevertheless exchanging current) have been observed on different real systems. The appearance of these phenomena results in a loss of usable capacity of the battery, difficulties in operating the system, and finally loss of income for the operator.

[0005] It is therefore important for BESS operators to better understand the causes of these phenomena, to be able to identify and quantify them.

[0006] To do this, the work of academics and industrialists focuses mainly on the development of methods for estimating the state of charge, SOC, and / or methods for improving their accuracy. Thus, the documents “End of Discharge SOC Jump Elimination”, Application Report, April 2020 by Texas Instruments and Orion BMS “Diagnosing State of Charge Calculation Jumps”, 2018 by Ewert Energy Systems both address the issue of the reliability of the SOC indicator. However, they are mainly interested in the causes of the problem of unreliability, and propose solutions to avoid it.

[0007] To date, little work has focused on characterizing the reliability or accuracy of this battery state of charge indicator from a user's perspective. In the case of stationary storage, such a user may be a battery integrator, who receives the state of charge estimate from a battery supplier's control system. Such a user may also be the storage system operator, who receives the state of charge estimate from the lower control level of the BESS, who may have calculated it himself, or simply relayed it from another source.

[0008] However, documents WO2021136419A1 and CN108931739A disclose methods for determining the accuracy of an SOC estimate, or for determining the error in estimating the SOC of a battery system, in particular in an electric vehicle. Summary

[0009] The present disclosure improves the situation, in particular in the detection of SOC estimation anomalies of the frozen SOC type, or in English "SOC Freeze" and of the SOC jump type, or in English "SOC jump".

[0010] A method is proposed for detecting an anomaly in variation of an estimate, by a battery management system, BMS, of a charge level, SOC, of ​​at least one electrical energy storage module. Such a method comprises: a. A collection of time-stamped data provided by the battery management system, comprising charge level estimation data and measurement data of a charge / discharge current of said at least one storage module; b. A calculation, from the time-stamped charge level estimation data, of a variation in the estimated charge level over a set of determined time intervals of duration dt; c. A calculation, from the time-stamped current measurement data and a capacity of the electrical energy storage module, of an equivalent variation in charge level over the set of time intervals of duration dt determined; d. A grouping of the time intervals into a plurality of groups of time intervals associated with an estimated load level variation equal to a determined reference value, and, for each of the groups: i. a calculation of a sum of the equivalent load level variations over the intervals of said group; ii. a comparison at a threshold of a difference of the determined reference value and the sum and, depending on a result of the com parison, a detection of an anomaly of variation of the estimate.

[0011] According to another aspect, there is provided a battery management system, BMS, of at least at least one electrical energy storage module, which comprises a memory and a processor configured to execute the aforementioned method of detecting an anomaly in the variation of an estimate of the charge level, SOC, of ​​this or these electrical energy storage module(s).

[0012] According to another aspect, there is proposed a method for managing an electrical energy storage system comprising a plurality of electrical energy storage modules, which implements, for each of said modules, the aforementioned method for detecting an anomaly in variation of an estimate of charge level, SOC, and which comprises a determination of a performance indicator of the storage system, as a function of a number of anomalies in variation of the estimate detected for each of the modules.

[0013] According to another aspect, there is provided a computer program comprising instructions for implementing all or part of a method as defined herein when this program is executed by a processor. According to another aspect, there is provided a non-transitory, computer-readable recording medium on which such a program is recorded.

[0014] The features set out in the following paragraphs may, optionally, be implemented, independently of one another or in combination with one another:

[0015] Such a method for detecting an anomaly in variation of an SOC estimate also comprises a restitution to a user of an indicator of detection of an anomaly in variation of the estimate and of the sum of equivalent variations in the calculated load level.

[0016] Such a method for detecting an anomaly in variation of an SOC estimate also comprises a determination of a state of health, SOH, of said at least one electrical energy storage module.

[0017] The capacity of said at least one electrical energy storage module taken into account for the calculation of the equivalent variation in charge level is a residual capacity of said at least one electrical energy storage module calculated according to the following formula:

[0018] [Math.2] Residual G — OH XC nom in which Cresidueile denotes the residual capacity, Cn„m denotes a nominal electrical capacity of said at least one module and SOH is a unitless parameter expressed as a percentage which denotes the state of health of said at least one module.

[0019] Calculation of the equivalent variation of charge level over a time interval of duration dt is calculated according to the following formula:

[0020] [Math.3] dSOCéQ,(iva}ent = -^^ xl00~7^- x 100 in which I denotes the charge / discharge current of said at least one electrical energy storage module and CrexidueIle denotes the residual capacity of the electrical energy storage module.

[0021] The threshold can be defined by the user, according to his needs, for example in order to verify the announced precision performances of the SOC. In one embodiment, this threshold is an error of estimation of SOC of the battery management system, for example the maximum error announced by the manufacturer of the battery management system, or the sum of this maximum error and the resolution announced by the manufacturer, or the sum of this maximum error and a known error of measurement of the current by the sensors. Brief description of the drawings

[0022] Other characteristics, details and advantages will appear on reading the detailed description below, and on analyzing the attached drawings, in which: Fig.l

[0023] [Fig.l] presents a block diagram of an electrical storage system comprising one or more electrical energy storage modules and a corresponding battery management system. Fig. 2

[0024] [Fig.2] shows an exemplary embodiment of a method for detecting an anomaly in the estimation of the variation of SOC of an electrical energy storage module, for example one of the modules of [Fig.l]. Fig. 3A

[0025] [Fig.3A] shows a graphical representation of a SOC charge level as a function of time, with grouping of time intervals according to one embodiment. Fig. 3B

[0026] [Fig.3B] shows the graphic representation of the SOC charge level of [Fig.3A], on which an anomaly detection threshold S has been represented. Fig. 3C

[0027] [Fig.3C] shows the graphical representation of the SOC charge level of [Fig.3A] and [Fig.3B], on which the result of the anomaly detection according to one embodiment has been represented. Fig. 4A

[0028] [Fig.4A] shows a graphical representation of a SOC charge level as a function of time, with grouping of time intervals according to another embodiment. Fig. 4B

[0029] [Fig.4B] shows the graphical representation of the SOC load level of [Fig.4A], on which an anomaly detection threshold S and the result of the anomaly detection according to this embodiment are shown. Fig. 5A

[0030] [Fig.5A] shows a graphical representation of a SOC charge level as a function of time, with grouping of time intervals according to another embodiment and use of a zero reference value D. Fig. 5B

[0031] [Fig.5B] shows the graphical representation of the SOC charge level of [Fig.5A], with another grouping of time intervals and use of a non-zero reference value D. Description of the embodiments

[0032] The general principle of the method described is based on the implementation of a coulombic counting method from time-stamped current measurement data for a calculation of equivalent SOC, then of a variation of equivalent SOC over a time interval, which corresponds to the image of the current delivered during this time interval. The latter is then compared to one or more thresholds to qualify this variation as normal, or abnormally slow (“SOC freeze”) or fast (“SOC jump”).

[0033] Reference is now made to [Fig. 1], which presents in block diagram form an example of an electrical energy storage system 10, or BESS (for the English “Battery Energy Storage System”). Such a BESS 10 comprises a plurality of devices 1C, four of which have been shown by way of example in [Fig. 1]. Each of these devices 111 to 114 comprises an electrical energy storage module, for example in the form of a pack of electrochemical battery cells, and a battery management system, also called BMS (for the English “Battery Management System”).

[0034] The BMS has multiple functions, including the evaluation of the state of health SOH of the electrical energy storage module, the estimation of its charge level SOC, the control of the balance between the different cells of the pack, or even the estimation of the remaining autonomy time for the module.

[0035] A bidirectional power flow P is established between the devices 111 to 114 and a conversion electronics module 12, which allows the power supply of a electrical network to which the BESS 10 is connected.

[0036] The BESS 10 also comprises a control module 13, which sends control signals to the conversion electronics module 12, and receives from the latter measurement data of voltage and intensity of the flow P. The control module 13 can also interrogate the BMS to receive from the latter estimated or calculated values ​​of SOC, SOH, temperature, or even charge / discharge current of the storage modules of the devices 111 to 114.

[0037] The control module 13 is in communication with an upper control stage 14, which controls the acquisition of data from the BESS 10, monitors its operation and controls it. This upper control stage 14 is for example controlled by the operator of the BESS 10, or an integrator of this battery storage system.

[0038] We will now describe, in relation to [Fig. 2], an example of implementation of a method for detecting an anomaly in estimating a variation in SOC of an electrical energy storage module. This module is for example a battery or a battery cell pack of one of the devices 11; of [Fig. 1]. Such a method is based on an analysis of data from operational systems, such as that illustrated in [Fig. 1]. It is therefore simple to implement, and does not require any hardware modification of existing BESSs.

[0039] In one example, this method is implemented in the control module 13 of the BESS 10. In another example, this method is implemented in the upper control stage 14 of the BESS 10. Indeed, a user of such a method could for example be, in the case of stationary storage, a battery integrator who receives the estimate of the state of charge by the control system of the battery supplier, or even a storage system operator who receives the estimate of the state of charge by the lower control level (whether it is calculated at the lower level or simply relayed).

[0040] In a first operation 21, measurement or estimation data from the BMS of one of the devices 11 are collected. These data are time-stamped: thus, they can then be processed according to the date on which they were measured or calculated. These time-stamped data include at least measurement data of a charge / discharge current I(t) of the electrical energy storage module, and data for estimating the charge level SOC by the BMS. Alternatively, other data can also be collected, such as the voltage U(t) at the terminals of the storage module, its state of health SOH, or the temperature T(t).

[0041] The collected data are acquired during a period of operational and uninterrupted operation of the device 11;. Thus, each type of data takes the form of a time series, here of the SOC and the current of the device in operation. operational. Operational operation is here understood to be distinguished from a maintenance or test phase for which the operating conditions can be chosen to correspond to test conditions. In operational operation, on the contrary, the device follows usual operating conditions, not imposed specifically for the measurements.

[0042] It is preferable that each collected time series be usable. Also, the collection operation 21 may comprise, or be preceded by, a selection, or filtering, eliminating from the collection the time series including an unusable time step, for example in the event of absence of value, error message, poor alignment of measurements, etc. At the end of the selection, the collected time series are all continuous and usable.

[0043] Thus, only time-stamped records of the charge / discharge current and the SOC estimation are required for the deployment of this approach, making it simple to implement and accessible for most existing and already deployed systems.

[0044] During an operation referenced 22, the time-stamped charge level estimation data, SOC(t), are used to calculate a variation in estimated charge level, over a set of determined time intervals of duration dt. For example, a time step dt= 1 second is chosen, and the variation in charge level estimated by the BMS is calculated over contiguous time intervals of 1s, according to the following formula:

[0045] [Math.4] dSOCeslimé = SOC (t) - SOCU - dt)

[0046] This gives the increment of the SOC estimated by the battery management system (BMS).

[0047] In one embodiment, this time step dt is that of the data sampling frequency (often identical between SOC and current). The sampling period should preferably be an order of magnitude lower than the period of change of the SOC value, so as to properly capture each of the value increments. For example, for an IC discharge with 0.5% SOC increments, the SOC change period will be of the order of 18s and sampling every 1-2s would make it possible to comfortably capture all the value increments.

[0048] During an operation referenced 23, which can take place before, after, or in parallel with the operation referenced 22, the time-stamped current measurement data collected during the operation 21 are used to calculate an equivalent variation in charge level over the same set of time intervals of duration dt. This equivalent variation is calculated by an integral calculation according to the following formula:

[0049] [Math.5] .4 dSOCèquiva!en{ = ..... x 100 in which I denotes the charge / discharge current measured by the BMS and collected during operation 21, and C denotes the capacity of the electrical energy storage module.

[0050] We thus calculate the increment of the equivalent SOC corresponding to the battery current between t-dt and t.

[0051] The capacity C can also be provided by the BMS. In one example, the nominal electrical capacity Cnom of the storage module is used. In another example, for a more precise calculation, the state of health SOH of the storage module is also taken into account, and the nominal capacity C„om is replaced in the equation by the residual capacity Cresiduelle of the storage module, calculated according to the following formula:

[0052] [Math.2] ^residual X Cn(m,

[0053] It is recalled that SOH is a unitless parameter expressed as a percentage which designates the state of health of the electrical energy storage module, and which can be part of the data collected during operation 21: the state of health indicator provided by the BMS is then used. Alternatively, it is also possible to use a third-party state of health indicator. In an advantageous embodiment, the method can comprise an operation for diagnosing the SOH state of health, based on reference tests, also called capacity tests, according to which a complete charge and discharge cycle is carried out on site, for which the discharged capacity in Ah is calculated (Cresidual) - In this case, one can either directly use CresidueUe^ or go through the product SOH X Ciwm.

[0054] Considering time intervals of sufficiently short duration dt, and taking into account the state of health SOH, the integral calculation of the equivalent variation of charge level can be approximated according to the following formula:

[0055] [Math.3] dSOCé^üvaleid = -^ xl00~7^- x 100

[0056] In an operation 24, the successive time intervals of duration dt are grouped into a plurality of groups of time intervals each associated with a variation in estimated charge level equal to a determined reference value dSOC estimated = D. This reference value D can be set by a user of the electrical energy storage module.

[0057] In a first embodiment, D=0, and the time intervals of duration dt successive are then grouped into a plurality of groups of time intervals each associated with a variation in the estimated load level dSOC calculated during the operation referenced 22 is zero over the successive time intervals [t0; t0+dt], [t0+dt; t0+2dt] and [t0+2dt; t0+3dt], these three time intervals are grouped into a first group of time intervals, denoted Gl. Several groups Gj are thus constructed, each of which groups together a set of consecutive time intervals over which the variation in the estimated load level dSOC calculated during the operation referenced 22 is zero.

[0058] Figure 3A illustrates such a grouping of time intervals, within the framework of a graphical representation of the charge level SOC as a function of time t. In this figure 3A, three groups of time intervals referenced G1, G2 and G3 are represented by way of example. For each of these groups G1 to G3, the variation in estimated charge level dSOCestimated from the BMS (step 21) is zero, as symbolized by the horizontal solid line. The variation in equivalent charge level dSOC^qujvaiaü on the other hand is non-zero, and illustrated, for each of the groups G1 to G3, by an ascending dotted line.

[0059] In an operation 25, we calculate, for each of the groups Gj formed during the operation referenced 24, a sum of the equivalent variations of charge level dSOCequivalent over 'cs intervals of the group. Thus, for the group Gl for example, we calculate this sum according to the following formula:

[0060] [Math.6] dSOCçquivalen(\G\.) dSOCtyuivaJenvd)

[0061] In an operation 26, the sum thus calculated for each group Gj is compared to an anomaly detection threshold, noted S in FIG. 3B. In one embodiment, for example, the error E of charge level estimation, as specified in the construction parameters of the BMS, generally expressed in the form ±E% of SOC, is used as anomaly detection threshold S. More generally, this threshold S is defined by the user. It is of course appropriate to distinguish this error threshold from the resolution.

[0062] For example, if for group Gl, the sum of the equivalent variations in charge level dSOCGl) is not zero but is less than the SOC estimation error specified by the BMS manufacturer, it is considered that there is a priori no SOC estimation anomaly, and that the difference between dSOCéquivaient( Gl) and dSOCesümé( Gl) = 0 is within the operating error margin of the BMS. It should be noted that the minimum anomaly detection threshold value that makes sense is the resolution of the SOC estimation of the BMS. It would indeed not be relevant to define an anomaly detection threshold of 0.2% for a re solution of the SOC variable of 0.5% because, by construction, this threshold would be exceeded most of the time.

[0063] If, on the other hand, for the group Gl, the sum of the equivalent variations in charge level dS OC equivalent {Gl) is not zero but, moreover, is greater than a threshold of interest defined by the user (e.g., the SOC estimation error specified by the BMS manufacturer), an anomaly in the estimation of the SOC is detected. Such an anomaly is of the “SOC freeze” type: in fact, according to the estimation provided by the BMS, the variation in SOC is zero, and the SOC remains constant. On the other hand, according to the time-stamped current measurement data collected during operation 21, the charge level has changed over the time interval [t0; t0+3dt], according to an equivalent increment sufficiently large not to fall within the threshold of interest defined by the user (for example, within the margin of error of the BMS).This increment corresponds to an exchange of current in Ah over the time interval [t0; t0+3dt] which should have triggered a SOC increment: if this is not the case, an anomaly is detected.

[0064] In the embodiment illustrated in [Fig.3C], no anomaly is detected. Indeed, for each of the groups of time intervals G1 to G3, we have:

[0065] [Math.7] | dSOCeAimé ( Gz ) - dSOC^a^t ( Gi ) | < S

[0066] Thus, the difference between the estimated SOC variation and the equivalent SOC variation is always lower than the threshold S, in absolute value, for each of the groups Gl to G3: no SOC estimation anomaly is therefore detected, as symbolized by the mention “OK” for each interval Gl to G3.

[0067] In another embodiment, the reference value D is non-zero, in order to allow the operation of the BESS 10 to be analyzed over larger sets of time steps.

[0068] As illustrated by the curves of Figures 4A to 4C, the use of a non-zero reference value D makes it possible, for example, to identify the cumulative error which would not be identified with a grouping of time intervals associated with a variation in the estimated zero charge level, taken individually.

[0069] In Figure 4A, in the context of a graphical representation of the charge level SOC as a function of time t, a group of time intervals referenced G' 1 is represented as an example, formed of the three groups of time intervals G1, G2 and G3 illustrated in Figures 3A to 3C. In Figure 4A, the estimated charge level variation dSOCextimé from the BMS (step 21) is symbolized by a solid line, and the equivalent charge level variation dSOCécluivaimt is illustrated by an ascending dotted line.

[0070] On the group of time intervals G' 1, the sum of the level variations of estimated load dSOC^timed( G'1) is equal to D.

[0071] In an operation 25, a sum of the equivalent variations in load level dSOCequivalent(G'1) over the intervals of the group is calculated for the group G' 1. During operation 26, the difference, in absolute value, between dSOCequivalent {G' \ ) and dSOC^né(G'1 ) is compared to the anomaly detection threshold, noted S in figure 4B, as defined by the user in the example of figure 4B, an anomaly is detected (symbol "NOK") because |^5GCestimated(G'l) - dSOCequivalent^ G'1 ) | > 5.

[0072] Thus, by comparing figures 3C and 4B, it is understood that the grouping of the time intervals by dSOCes&mé of non-zero reference value D (group G' 1) makes it possible to identify a cumulative error which is not identified on each group Gl, G2, G3 taken individually for which dSOC — 0.

[0073] Figures 5A and 5B illustrate another exemplary embodiment, in which the effect on the detection of SOC estimation anomalies of using a zero or non-zero reference value D for grouping the time intervals is compared.

[0074] In Figure 5A, we consider four groups of time intervals referenced Gl to G4, which were constructed using a reference value D=0. Thus, on each group Gl to G4, we have dSOC ■ (Gi) = 0, as represented by the horizontal solid line. We consider a threshold S defined by the user. The variation dSOCéquh.aim^G^ for each of the groups Gl to G3 is negative (the SOC decreases over each of these groups of time intervals), as represented by the dotted lines, but remains below the threshold S for each of these groups. For group G4, the variation dSOC^qUi,.aiml(G4) is positive, but, again, remains below the threshold S. However, there is a significant jump in SOC between the groups of time intervals G3 and G4, which is therefore not detected as an anomaly in this case of Figure 5A where the time intervals associated with a variation in the estimated load level of zero (r / 5OCesdLmé = 0) are grouped.

[0075] In Figure 5B, we consider a non-zero reference value D, which leads to constructing a group of time intervals G' 1 including the groups Gl to G4 of Figure 5A. On the group G' 1, we have t / SOC^jG'l) = D-

[0076] In an operation 25, a sum of the equivalent variations in load level dSOCequivalent(G'1) over the intervals of the group is calculated for the group G' 1. During operation 26, the difference, in absolute value, between dSOCéqUiValent{G' V) and dSOC G'1 ) is compared to the anomaly detection threshold, noted S in figure 5B: an anomaly is then detected (symbol "NOK") because |d5OCestimated(G'l) - dSOC equiyaUnt^G'l) | >5.

[0077] It is thus possible to identify excessively rapid SOC variation rates (also called “SOC jumps”).

[0078] Thus, the use of a non-zero reference value D for the grouping of time intervals makes it possible to give a lower weight, in the anomaly detection, to the resolution of the SOC variable, as announced for example by the manufacturer of the BESS 10.

[0079] Returning to the flowchart of [Fig. 2], in an optional operation 27, the anomaly detection results of the operation referenced 26 can be used to develop a performance indicator of the BESS 10 system. Such an indicator can take the form of a frequency of occurrence of frozen SOC phenomena, or a distribution of the amplitude of these frozen SOCs. In another example, upon installation of the BESS 10, the performance indicator is initialized to a value of 10, and decremented by one point each time an anomaly is detected during operation 26. In particular, it is possible to monitor the number or frequency of occurrences over time of the detection of anomalies, and observe whether this number or frequency of occurrences is the same for all the elements of the battery. This performance indicator can be relayed, via the upper control stage 14, to the BESS operator 10.

[0080] This anomaly detection indicator can be returned to a user on a human-machine interface of the BESS 10. In one embodiment, this indicator is returned in association with the value of the sum of the equivalent variations in load level calculated, which allows the user to refine the anomaly detection diagnosis, and provides him with more complete and richer information on the identified problem.

[0081] The methods and variants described above may be implemented by computer means, in particular a computer on which a program is recorded for implementing such a method when this program is executed by a processor. Such a program may also be stored on a non-transitory recording medium readable by a computer.

[0082] The present disclosure for the detection of anomalies in the estimation of SOC of an electrical energy storage system in operation differs from known techniques in particular by: - the use of data traditionally available within the BMS, which can be accessed without hardware modification of existing systems. The process can thus be directly implemented on all electrical energy storage systems already deployed and operational; - the ability to operate in real time, on a running system. This allows BESS operators to detect SOC estimation anomalies as soon as their appearance, and therefore to put in place the necessary corrective measures quickly for optimized operation of the system; - grouping the equivalent increments of SoC variation over a period over which the BMS's estimate of the SoC remains unchanged or equal to a determined reference value and comparing the total increment obtained to a threshold defined by a user, which may be, for example, the BMS' SoC estimation error. We thus propose a simple calculation method, which consumes little resources, and is directly applicable to the detection of anomalies of the "SOC freeze" and "SOC jump" type; - taking into account the state of health of the battery to define the comparison thresholds. This makes it possible to increase the accuracy of the detection of SOC estimation anomalies, by relying on the particular state of the battery at the time the calculations are carried out; - the possibility of carrying out the steps of the anomaly detection method described in the present disclosure by using sliding time windows for grouping the equivalent increments of SOC variation.

[0083] Furthermore, the possibility for the BESS operator to have reliable information in real time on the accuracy of the SOC estimation allows it to make documented market-oriented arbitration choices to respond to a given service.

[0084] Such a method also improves the operational safety of the BESS. Indeed, in view of the alerts generated by the detection of SOC estimation anomalies, it is possible for the operator of the BESS to detect that maintenance is necessary on the electrical energy storage system, for example to replace one or more electrical energy storage modules for which anomalies are frequently detected. Examples

[0085] For example, a battery with a capacity of 200Ah is considered. According to the current measurement data collected during the operation referenced 21, this battery was charged at a current of 100A for a duration of 15min. On the other hand, according to the result obtained during the operation referenced 22, the SOC indicator remained unchanged during this period: dSOCeÿümé( 15min) = 0- The equivalent variation in charge level over this period of a quarter of an hour, as calculated during operations 23 and 25, gives a charge increment of 25Ah, or an equivalent SOC increment of 12.5%: dSOC^^ x 100 = -¾^ x 100= 12.5% ​​'-residual

[0086] Depending on the anomaly detection threshold chosen for operation 26, this increment, not considered by the BMS SOC gauge, could trigger the identification of a frozen SOC, or “SOC freeze”. Industrial application

[0087] The present technical solutions make it possible to identify abnormal behaviors of the SOC variable proposed by a BMS based on limited operational data of the BESS battery storage system studied.

[0088] The estimation of the indicators is done without modifying the behavior of the electrical energy storage system. The service value of the system is therefore not impacted by such processes. No additional tools or equipment are necessary (probes, sensors). Only basic physical quantities such as the SOC estimate and the current are used. However, these parameters are usually measured and available in existing systems.

[0089] Operators or integrators of battery storage systems can use the proposed method to challenge the reliability of the state of charge indicator provided by a third party (typically by the lower control level). This method can be used as a performance indicator for storage systems in operation to identify problems related to the estimation of the state of charge. It then uses "online" data, and can be used at any level of battery architecture for which the monitoring system calculates a state of charge (cell, module or rack, etc.).

[0090] This method can also be used in offline mode on historical data. Thus, Battery Management System providers can use this approach to characterize a posteriori the behavior of their state of charge estimator.

[0091] There are in fact two ways of implementing the process: - For real-time applications (stationary storage or embedded in the storage system). The algorithmic method can be integrated into the battery management system (or BMS for "Battery Management System"). The sufficient raw data then correspond to the instants (to, t0 + dt, ..., t0 + n*dt). Operations 21 to 26 can be executed continuously over a sliding window, or periodically. Anomaly detection can be done at each execution (iteration) of operations 21 to 26. The method then uses the anomalies detected locally, or can export them. - For a posteriori analysis applications (stationary or on-board energy storage). The algorithmic method can be deported from the BESS storage system itself, for example integrated into a more general data processing tool in which part or all of the collected data is processed.

[0092] The present disclosure is not limited to the examples of methods, systems, programs and recording media described above, only to as an example, but it encompasses all the variants that a person skilled in the art may consider within the framework of the protection sought. List of reference signs

[0093] - 10: Electrical energy storage system - lli: electrical energy storage device; - 12: conversion electronics module; - 13: control module; - 14: upper control floor; - 21 to 27: operation. List of cited documents Patent documents

[0094] For all useful purposes, the following patent documents are cited: - patcitl: WO2021136419A1 (publication number); - patcit2: CN108931739A (publication number). Non-patent literature

[0095] For all useful purposes, the following non-patent element(s) is (are) cited: - nplcitl: “End of Discharge SOC Jump Elimination”, Application Report, April 2020 by Texas Instruments; - nplcit2: Orion BMS “Diagnosing State of Charge Calculation Jumps”, 2018 by Ewert Energy Systems..

Claims

1.

2. Claims Method for detecting an anomaly in the variation of an estimate, by a battery management system, BMS, of a charge level, SOC, of ​​at least one electrical energy storage module, characterized in that it comprises: a. A collection (21) of time-stamped data provided by said battery management system, comprising data for estimating said charge level and data for measuring a charge / discharge current of said at least one storage module; b. A calculation (22), from said time-stamped data estimating said charge level, of a variation in the estimated charge level over a set of determined time intervals of duration dt; c. A calculation (23), from said time-stamped current measurement data and a capacity of said electrical energy storage module, of an equivalent variation in charge level over said set of determined time intervals of duration dt; d. A grouping (24) of said time intervals into a plurality of groups of time intervals associated with an estimated charge level variation equal to a determined reference value, and, for each of the groups: i. a calculation (25) of a sum of the equivalent charge level variations over said intervals of said group; ii. a comparison (26) with a threshold of a difference between said determined reference value and said sum and, depending on a result of said comparison, a detection of an anomaly in variation of said estimate. Method for detecting an anomaly according to claim 1, characterized in that it also comprises a restitution to a user of an indicator of detection of an anomaly of variation of said estimate and of said sum of equivalent variations of level of calculated load.

3. Method for detecting an anomaly according to any one of claims 1 and 2, characterized in that it also comprises a determination of a state of health, SOH, of said at least one electrical energy storage module.

4. Method for detecting an anomaly according to claim 3, characterized in that the capacity of said at least one electrical energy storage module taken into account for the calculation of said equivalent variation in charge level is a residual capacity of said at least one electrical energy storage module calculated according to the following formula: [Math.2] G residual — SOH X Clmm in which Cresidliel[e designates said residual capacity, Cnom designates a nominal electrical capacity of said at least one module and SOH is a unitless parameter expressed as a percentage which designates the state of health of said at least one module.

5. Method for detecting an anomaly according to claim 4, characterized in that the calculation (23) of said equivalent variation in charge level over a time interval of duration dt is calculated according to the following formula: [Math.3] dSOC&,,M,,, = S^ xlOO- / ^ x 100 rus whieile res i duel le in which I designates the charge / discharge current of said at least one electrical energy storage module and Crésiduelle designates the residual capacity of said electrical energy storage module.

6. Method for detecting an anomaly according to any one of claims 1 to 5, characterized in that said threshold is defined by a user of said battery management system.

7. Computer program comprising instructions for implementing the method according to one of claims 1 to 6 when this program is executed by a processor.

8. Non-transitory recording medium readable by a computer on which is recorded a program for implementing the method according to one of claims 1 to 6 when this program is executed by a processor.

9. Method for managing an electrical energy storage system (10) comprising a plurality of electrical energy storage modules, characterized in that it implements, for each of said modules, the method for detecting an anomaly according to any one of claims 1 to 6, and in that it comprises a determination (27) of a performance indicator of said storage system (10), as a function of a number of anomalies of variation of said estimate detected for each of said modules.

10. Battery management system, BMS, of at least one electrical energy storage module, characterized in that it comprises a memory (M) and a processor (PROC) configured to execute: a. A collection (21) of time-stamped data, comprising data for estimating said charge level and data for measuring a charge / discharge current of said at least one storage module; b. A calculation (22), from said time-stamped data estimating said charge level, of a variation in the estimated charge level over a set of determined time intervals of duration dt; c. A calculation (23), from said time-stamped current measurement data and a capacity of said electrical energy storage module, of an equivalent variation in charge level over said set of determined time intervals of duration dt; d. A grouping (24) of said time intervals into a plurality of groups of time intervals associated with an estimated charge level variation equal to a determined reference value, and, for each of the groups: i. a calculation (25) of a sum of the equivalent variations in load level over said intervals of said group; ii. a comparison (26) with a threshold of a difference between said determined reference value and said sum and, depending on a result of said comparison, a detection of an anomaly in variation of said estimate.

Images