Method for self-configuring the charging conditions of a rechargeable battery

Abstract

A method for self-parameterizing the charging conditions of an electrochemical battery, in particular a lithium battery, said battery being connected to a suitable power supply, comprising: detecting at least one inflection point p in the battery's charging voltage curve Ubat as a function of the electrical charge rate expressed as a percentage of the battery's capacity C; for each detected inflection point p, disconnecting the battery from its power supply for a predetermined relaxation period; at the end of said relaxation period, recording the open-circuit voltage UxR of the battery for each inflection point, said open-circuit voltage UxR constituting a marker of an intermediate charge level usable as a charging cut-off threshold during battery charging. (See abstract figure: Fig. 2)

Description

Title of the invention: Method for self-setting the charging conditions of a battery of accumulators

[0001] The invention relates to a method for self-setting the charging conditions of an electrochemical battery, particularly a lithium battery, especially useful in the specific context of battery charging with the aim of extending their lifespan. It is well known that this type of battery ages rapidly when charged to a full or near-full charge, and it is therefore strongly advised not to fully recharge them but to stop charging when the battery has reached an intermediate charging threshold, that is, a state of charge below the battery's theoretical capacity.

[0002] Methods for stopping the charging of a battery at a target state of charge significantly below 100% of its capacity are already known. Document EP3220471, in particular, describes a method based on the existence of one or more characteristic inflection points in the battery's charging voltage curve. These specific points on the curve define charge levels that are, in principle, more easily identifiable and can be exploited during charging to maximize battery life by avoiding overcharging, especially, but not exclusively, when the ambient temperature is high. Choosing these known and identifiable inflection points as reference thresholds for charging has the advantage of limiting the error in the achieved charge level, particularly for LFP (or LiFePO4) technology batteries, which have a very flat charging curve.

[0003] When the charging voltage reaches the target threshold, corresponding to a predetermined battery charge level, the battery is disconnected from its power supply to begin a relaxation phase of predefined duration (generally a few minutes) during which charging is stopped. The battery is considered charged not when its charging voltage is greater than or equal to the target threshold, but when its open-circuit voltage after relaxation is greater than said target threshold. The target threshold is then considered a stopping threshold, or a stopping value, during the battery charging operation.

[0004] The disadvantage of this method is that the position and value of the inflection points, i.e. points on the charging curve known and usable as reference thresholds for stopping charging, depend on the electrochemical characteristics of the battery used and can vary quite significantly depending on the manufacturer. Therefore, if the battery supplier changes, the entire characterization must be repeated and the new values ​​for the charging cut-off thresholds must be redefined. Similarly, as part of a maintenance procedure, replacing a battery with an equivalent model requires resetting the charging scheme.

[0005] In practice, the method disclosed by EP3220471 therefore includes a preliminary configuration step in which correlation data between the battery's state of charge and its open-circuit voltage are determined. This configuration step can be carried out in the laboratory or in the factory, on the battery itself or on an identical reference battery. It is this reconfiguration, in particular, aimed at identifying the relevant characteristics of the charging curve—especially for determining the charging cut-off thresholds—that must be repeated when replacing a battery with another model and / or changing suppliers.

[0006] The objective of the present invention is to eliminate this configuration step for determining the correlation data, a step that can be considered cumbersome because it requires testing and preparation prior to the use of new batteries. The invention addresses the shortcomings of previous methods by proposing a self-configuration process that is executed automatically and systematically each time a battery is replaced. It therefore significantly improves the ease of use of existing batteries.

[0007] To this end, the method for self-configuring the charging conditions of an electrochemical accumulator battery, in particular a lithium battery conventionally connected to a power source capable of charging it, is such, according to the invention, that it comprises:

[0008] - the detection of at least one inflection point in the load voltage curve of the battery Ubat function of the electrical charge rate expressed as a percentage of the battery's capacity C;

[0009] - for each inflection point p detected, a disconnection of the battery from its source feeding during a predetermined relaxation period

[0010] - at the end of said relaxation period, the memorization of the open-circuit voltage UxR of the battery for each of the inflection points, said open-circuit voltage UxR constituting a marker of an intermediate charge level usable as a charge stop threshold during the charging of a battery.

[0011] These phases are repeated for each of the inflection points or thresholds to be determined: there are generally two or three thresholds corresponding to as many intermediate charge levels, and therefore several possibilities for stopping the charge during the recharging of a battery to extend its lifespan. The objective of the invention is thus achieved. namely to propose the means to best adapt to the charging profile of each battery in order to manage it optimally, that is to say in the way most suited to each battery treated, with the ultimate aim of optimizing its long-term operation.

[0012] According to a method specific to the invention, the detection of an inflection point is carried out by locating a maximum of a function giving the slope of points distributed on the battery charging curve, said function being P = AUbat / SIbat, for which:

[0013] - AUbat represents the difference in charging voltage between two distinct points of the battery charging curve at two times t and t + At, and

[0014] - SIbat represents the sum of the current values ​​Ibat entering the battery, measured according to a periodic sampling of periodicity Tedans the time interval At corresponding to the charging time elapsed between said points.

[0015] The principle underlying the operating method of the invention lies in the fact that, for the detection of inflection points to be reliable, it is considered necessary to rely on measuring the derivative of the function expressing the battery voltage as a function of the stored charge, and not as a function of time. It is known that the derivative of a function provides a measure of the slope of the function's curve, here the battery charging curve, at a given point. It allows us to understand the local appearance of the curve, in particular to determine when inflections occur and what their nature is. This approach also allows the method to operate regardless of fluctuations in the battery current, which can even temporarily reverse when the system consumes more energy than it receives.

[0016] To arrive at the function P = AUbat / SIbat used in the method of the invention, we start with the slope of the curve at a given point, defined, therefore, by the derivative calculated at that point, and which is approximated by the formula: dUbat / dQbat ~ AUbat / AQbat, where AUbat is the change in battery voltage (in volts) between two points on the charging curve and AQbat is the change in battery charge (in coulombs or ampere-hours) between these two points. This change in charge AQbat between two points on the charging curve is in fact equal to the integral of the current with respect to time, calculated over the charging time separating these two points in practice.

[0017] In a digital system based on periodic sampling of voltage and current, this integral can be approximated by a discrete sum:

[0018] AQbat « SIbat. Te

[0019] where Te is the sampling period (in seconds) of the digital system and Ibat is the battery charging current (in amperes) at the nth sampling instant. It is understood that the smaller Te is, the more precise the approximation provided by the sum above, and the fewer terms need to be added for the same duration. The integration rate is obviously high. The slope of the curve at a given point AUbat / AQ bat, can therefore be expressed in the form AUbat / SIbat. Te.

[0020] Since the term Te of the relation is constant (for example 1 minute) and we seek to detect where the points of maximum slope of the load voltage curve are located without needing to know the exact value of the slope, we can simply locate the maxima of the function P:

[0021] P = AUbat / SIbat

[0022] This is the relationship used by the process of the invention.

[0023] According to this method, a slope Pn is iteratively calculated between pairs of points on the battery charging curve whose spacing on the curve corresponds to a charged capacity cc in the battery substantially equal to a predefined percentage p of the battery capacity C. Advantageously, in the invention, the percentage p of charged capacity cc in the battery can be between 3% and 5% of the battery capacity C, preferably equal to 4%. This percentage p is chosen with a value sufficiently high so that the voltage variation corresponds to a significant and measurable increase, which makes it possible to limit the inaccuracy related to tolerances in the voltage and current measurements.The disadvantage resulting from choosing a high value is that the time between two slope measurements is relatively long, leading to a total number of measurements along the load curve that is not very high, and whose significance and effectiveness in finding the inflection points of the curve can therefore be questioned.

[0024] In the method of the invention, to address this problem, several slope calculations are performed simultaneously, offset by a charge value substantially lower than that used in the slope calculation. The offset charged capacity value is preferably a submultiple of the charged capacity value used for the slope calculation. More precisely, the slope Pn is actually calculated periodically, corresponding to an increase in the charged capacity of the battery equal to a fraction f of the charged capacity cc used for the slope calculation. Each time the battery charge increases by this fraction, a new slope value is calculated and available with the previously stated accuracy, considered appropriate. Advantageously, the fraction f is between 1 / 5 and 1 / 3 of the charged capacity cc, preferably equal to 1 / 4 thereof.

[0025] If, for example, we consider that the slope Pn is calculated between two points on the battery charging curve corresponding to a charged capacity equal to 4% of the battery capacity C, the system actually performs sequential slope calculations on ranges of the same length equivalent to 4% charge, but offset from each other by a distance corresponding to a load value, for example, of 1% of the loaded capacity. In this case, specifically, 4 interlaced slope calculations spaced 1% apart will be performed instead of a single calculation, and the number of slope measurements is considerably higher, certainly more significant for a successful search for the inflection points of the load curve.

[0026] According to the method of the invention, the calculated slope Pn is then smoothed by an infinite impulse response digital filter that calculates a filtered slope PFn. This filter conventionally uses a recurrence equation to smooth the result, i.e., the slope value, as a function of its past output and the value of Pn. In one possibility, the infinite impulse response digital filter used in this case is of the type:

[0027] PFn = (k. PFn l + Pn) / (k + 1)

[0028] k being a filtering coefficient between 5 and 10. Preferably, the selected filtering coefficient k is equal to 7.

[0029] The plot of the PFn function curve described above is developed along the battery charging curve, associating each point on the curve with a point on the PFn curve that gives its smoothed slope. It is then easy to observe that this PFn curve reaches a local maximum in several places. In the operation that will be described in more detail below, as soon as the PFn function curve begins to decline again, a maximum is considered to have been detected, each maximum corresponding to a desired inflection point on the battery charging curve. According to the invention, the method comprises:

[0030] - the detection of the beginning of each decrease in the PFn curve;

[0031] - the assimilation of this beginning of each decrease to an inflection point p.

[0032] Then, in accordance with what has been indicated previously, for each inflection point p detected, the method includes disconnecting the battery for a predetermined relaxation period and, at the end of said relaxation period, memorizing the open-circuit voltage UxR of the battery for each of the inflection points, said open-circuit voltage UxR constituting a marker of an intermediate charge level usable as a charging stop threshold during a battery charging process.

[0033] Additionally, the invention may include storing, for each detected inflection point p, the value UxC of the battery charging voltage. In this case, more generally, the self-parameterization method of the invention measures and stores the voltage reached UiC at the first inflection point, cuts off the charge for a given relaxation time (for example, 5 minutes), then remeasures the battery voltage UiR and also stores it. The charge is reactivated, and the process continues to determine the thresholds U2c / U2r, U3C / U3r, etc., corresponding to the other points of inflection, in other words to the other charge levels, easily identifiable by this method. Alternatively, still for each inflection point p detected, the value of the load voltage UxC is obtained by extrapolation from the stored open-circuit voltage value UxR, by adding to UxR a predetermined value related to the duration of the relaxation period.

[0034] The method also includes a step of validating the voltage values ​​found for the inflection points, such as:

[0035] - the load voltage values ​​UxC and the open-circuit voltage values ​​UxR are compared, for each inflection point p, at predetermined lower or upper threshold values;

[0036] - if the comparison tests are positive, the load voltage values ​​UxCet of open-circuit voltage UxR are validated and used as markers of load stop thresholds, respectively under load and after relaxation.

[0037] Once all thresholds are determined, the method concludes the self-detection procedure for the battery's charging cut-off thresholds. With these thresholds detected and stored, the battery's charging conditions are considered to be configured. It should be noted that the method described above, although using a current-over-time integration method, differs from the coulometric counting charging method in that the mathematical integration of the current is applied only over a limited period, and not over the entire charging process, thus limiting long-term drift. Furthermore, it should be noted that this method is not used to calculate the battery's current charge level but to automatically determine the charging cut-off threshold(s).

[0038] The invention will be better understood with the aid of the following detailed description of the method for self-setting the charging conditions of a battery, with reference to the accompanying drawings, for which:

[0039] [Fig-1] represents a plurality of curves with an x-axis that measures the battery charge rate as a percentage (%), including mainly the battery charge voltage curve Ubat and the filtered slope curve PFn, as well as the interlaced step curves of the points on the charge curve whose slope is measured;

[0040] [Fig.2] shows a flowchart detailing the processing of self-detection of inflection points; and

[0041] [Fig.3] represents a flowchart showing the details of the calculation of the filtered slope for each point p considered.

[0042] Figure 1 essentially shows the charging curve, that is, the curve of the charging voltage Ubat (in mV) (on the left-hand axis of the figure) as a function of the charge rate (in percentage), for an LFP (Lithium Iron Phosphate) battery. It notably shows, in a relatively flat middle portion, two thresholds UiC and U2c which are precisely the inflection points sought, characteristic of the UxC charge thresholds of such a battery, which we want to identify in particular to determine markers allowing to carry out recharges which do not compromise the life of the battery.

[0043] A second curve (on the right of the figure) shows on its ordinate the amplified and then filtered slope PFn of the points of the previous curve, also as a function of the battery charge level, and it therefore has the same abscissa axis. This second curve has two maxima, and the charge cut-off thresholds identified as Uic and U2c correspond, in accordance with the method of the invention, to the beginning of each decrease in the PFn curve.

[0044] Alongside the Ubat curve, just below its trace, four interlaced step curves Uechi, Uech2, Uech3, and Uech4 show the successive pairs of points for calculating the slopes of the main Ubat curve, located at a distance corresponding to 4% of the battery's charged capacity. Each of these interlaced curves Uechi, Uecb2, Uecb3, and Uech4 develops for points separated by a distance corresponding to 1% of the battery's charged capacity. As mentioned previously, four interlaced slope calculations spaced 1% apart are therefore performed instead of a single calculation every 4%. While the latter would certainly guarantee a significant voltage variation for the slope calculation, it would result in a reduced number of obtained values.On the contrary, the interlacing of curves at each 1% variation of the load rate makes it possible to obtain a much higher number of slope measurements, and consequently much more significant from the perspective of finding inflection points.

[0045] With reference to [Fig. 2], an example of an implementation of the method of the invention is illustrated by the flowchart shown. For each of the thresholds to be determined, the variables used to total the current are initialized, and the main quantities calculated at each loop iteration are stored. The software loop begins with a wait of a fixed duration Te, which is 1 minute here, then the voltage Ubat across the battery terminals and the current Ibat injected into or drawn from the battery are measured. The latter is added to two totals SIbat and SIi%, which are assigned to different functions. SIbat is used in the slope calculation, and SIi% is compared to a threshold corresponding to 1% of the battery capacity multiplied by the time Te between two successive measurements. If this threshold is exceeded, the total SIi% is reset to 0, and the slope calculation is performed according to the method illustrated in [Fig. 3].Otherwise, we loop back to the waiting stage to continue the accumulation.

[0046] The slope calculation is shown in [Fig. 3]. It is only performed if a previous measurement of Ubat has already been stored in a table UbatPrec, which contains 4 cyclic index values ​​from 1 to 4, UbatPrec[1] to UbatPrec[4]. It is therefore easy to understand that the iterative calculation of the slope Pn can only begin after the 5th measurement. In this case, as already mentioned, calculating the slope Pn amounts to dividing the difference between the battery voltage Ubat and its value 4 measurement iterations prior, UbatPrec[1], by the difference between the current totalizer SIbat and its value 4 measurement iterations prior, SIbatPrec[1]. Immediately after this calculation of slope Pn, the calculated slope is injected into a first-order digital filter called an infinite impulse response (IIR) filter, whose coefficient is preferably chosen to be equal to 7, and which obeys the following relationship:

[0047] PFn = (k. PFn l + Pn) / (k + 1)

[0048] Next, the battery voltage value Ubat and the current totalizer value SIbat are stored in SIbatPrec[i] before the value of i is incremented. If it exceeds 4, i is reset to 1, with a test systematically performed to ensure this. Thus, four slope calculations are interleaved, each corresponding to a 4% increase in battery charge, but the filtered slope value PFn is updated with each 1% increase in battery charge.

[0049] Once the filtered slope PFn of rank n has been calculated, the rest of the process is shown in [Fig. 2] and consists of a series of tests. First, a local maximum of the slope is located by comparing the current value PFn to a maximum value PFmax obtained so far and stored, allowing the detection of a slight decrease—on the order of a few percent of the maximum value reached—by comparing the current value PFn with a value PFmax—e, e being predetermined and corresponding to these few percent. It is the beginning of the decrease thus found that serves, in the process of the invention, to identify the significant load levels, the load thresholds to be stored, as can be clearly seen in [Fig. 1].

[0050] It is then verified that the battery voltage Ubat is within a plausible range, depending on the threshold being sought. On the curve in [Fig. 1], threshold no. 1 (between 30 and 35% charge) is, for example, only validated if Ubat < 3350mV, while threshold no. 2 (between 75 and 80% charge) is only validated if Ubat > 3370mV. In the first case, since the charge level is relatively low, the comparison is made against an upper limit, whereas in the second case, the charge level being relatively high for such a battery, the comparison is made against a lower limit. If the comparison is validated, the Ubat value is stored in a variable UxC, where x is the number of the inflection point or threshold being sought. Then, charging is stopped for a predefined time (5 minutes in the example of the [Fig. 2]) and the battery voltage is remeasured after relaxation to store it in UxR. Then, charging can be reactivated to potentially detect another threshold. It should be noted that the UxR thresholds thus found can be used within the framework of the charge termination process described in the previously mentioned patent EP3220471.

[0051] According to an additional possibility, the method of the invention may also include a storage of the value of the current totalizer associated with each threshold, and a control aimed at determining that the difference in totalized current between 2 thresholds is sufficient to exclude possible detections of non-significant secondary thresholds.

[0052] An installation for implementing the method described above, which is not the subject of the present invention, is described in patent EP3220471, the current sensor being essential for the present method. It may be a very low value series resistor (< 1Q) interposed between the battery and the power source, with or without an amplification circuit, or a Hall effect or magnetoresistive current sensor.

[0053] In the preceding description, according to one possible application, the battery may be intended to power the drive motor of a moving element for closing, blocking, sun protection, or screening, for example, a motorized shutter. The method of the invention generally relates to a battery of electrochemical accumulators, in particular a lithium battery, intended to be charged by connection to an external power source and, for example, to supply electrical energy to the electric drive motor of such a moving element. The present invention is not limited to the type of battery explicitly mentioned above in the description or shown in [Fig. 1], but can be applied to any type of battery, in particular lithium batteries such as lithium-ion batteries, but also lead-acid batteries, NiMH batteries, etc.

Claims

Demands

1. A method for self-parameterizing the charging conditions of an electrochemical accumulator battery, in particular a lithium battery, said battery being connected to a power source suitable for charging it, comprising the detection of at least one inflection point p in the charging voltage curve of the battery Ubat as a function of the electrical charge rate expressed as a percentage of the capacity C of the battery, characterized by: - ​​for each inflection point p detected, a disconnection of the battery from its power source for a predetermined relaxation period; - at the end of said relaxation period, the storage of the open-circuit voltage UxR of the battery for each of the inflection points, said open-circuit voltage UxR constituting a marker of an intermediate charge level usable as a charging stop threshold during the charging of a battery.

2. A method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein the detection of an inflection point is carried out by locating a maximum of a function giving the slope of points distributed on the charging curve of the battery, said function being P = A11^ / 21^, for which: - AUbat represents the difference in the charging voltage between two distinct points of the charging curve of the battery at two instants t and t + At, and - SIbat represents the sum of the values ​​of current Ibat entering the battery, measured according to a periodic sampling of periodicity Tedans the time interval At corresponding to the charging time elapsed between said points.

3. A method for self-parameterizing the charging conditions of a battery according to the preceding claim, wherein a slope Pn is iteratively calculated between pairs of points on the battery charging curve whose spacing on the curve corresponds to a charged capacity cc in the battery substantially equal to a predefined percentage p of the battery capacity C.

4. A method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein the percentage p of charged capacity cc in the battery is between 3% and 5% of the capacity C of the battery, preferably equal to 4%.

5. A method for self-parameterizing the charging conditions of a battery of accumulators according to any one of claims 3 and 4, wherein the slope Pn is calculated according to a periodicity corresponding to an increase in the capacity charged in the battery equal to a fraction f of the capacity charged cc.

6. A method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein the fraction f is between 1 / 5 and 1 / 3 of the charged capacity cc, preferably equal to 1 / 4 thereof.

7. A method for self-parameterizing the charging conditions of a battery of accumulators according to any one of claims 5 and 6, wherein the calculated slope Pn is smoothed by an infinite impulse response digital filter which calculates a filtered slope PFn.

8. A method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein the infinite impulse response digital filter is of the type: PFn = (k.PFn.1 + Pn) / (k+l) k being a filtering coefficient between 5 and 10.

9. Method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein the filtering coefficient k is equal to 7.

10. A method for self-parameterizing the charging conditions of a battery of accumulators according to any one of claims 7 to 9, comprising: - the detection of the beginning of each decrease in the PFn curve; - the assimilation of this beginning of each decrease to an inflection point p.

11. A method for self-setting the charging conditions of a battery according to the preceding claim, comprising the memorization, for each detected inflection point p, of the value UxC of the battery charging voltage.

12. A method for self-parameterizing the charging conditions of a battery of accumulators according to claim 10, comprising, for each inflection point p detected, the value of the charging voltage UxC is obtained by extrapolation from the stored open-circuit voltage value UxR, by adding to UxR a predetermined value related to the duration of the relaxation period.

13. A method for self-parameterizing the charging conditions of a battery of accumulators according to the preceding claim, wherein: - the values ​​of charging voltage UxC and open-circuit voltage UxR are compared, for each inflection point p, to predetermined lower or upper threshold values; - if the comparison tests are positive, the values ​​of charging voltage UxC and open-circuit voltage UxR are validated and used as markers of charging stop thresholds, respectively during charging and after relaxation.

Images