METHOD FOR MANAGING THE STATE OF CHARGE OF A HYBRID VEHICLE

Abstract

The state of charge of a traction battery of a hybrid vehicle power train is managed by, during a phase of running of the vehicle to a current destination, predicting a temperature that a battery will reach, after the power train is switched off, at a time of departure to a future destination; estimating, as a function of the battery temperature previously predicted, a minimum state of charge of the battery making it possible to provide, during a phase of running to the future destination, a predefined minimum power level; and maintaining the state of charge of the battery close to the minimum state of charge.

Claims

1-12. (canceled)

13. A method for managing a state of charge of a traction battery of a hybrid vehicle power train, the method comprising, during a phase of running of the vehicle to a current destination: predicting a temperature that the battery will reach, after the power train is switched off, at a time of departure to a future destination; estimating, as a function of the battery temperature previously predicted, a minimum state of charge of the battery making it possible to provide, during a phase of running to the future destination, a predefined minimum power level; and maintaining the state of charge of the battery close to the minimum state of charge.

14. The method as claimed in claim 13, wherein the predicting the temperature that the battery will reach includes among its parameters: an elapsed time between the end of the current run and the start of the future run, and/or; a model of variation of the ambient temperature between the end of the current run and the start of the future run, and/or; a model of thermal inertia of the battery giving the trend of the temperature of the battery as a function of the trend of the ambient temperature.

15. The method as claimed in claim 14, wherein the elapsed time between the end of the current run and the start of the future run is: obtained directly from a run scheduling system, or; deduced statistically from previous runs obtained from a run memorizing system, said previous runs exhibiting similarities with the current run, or; equal to a constant.

16. The method as claimed in claim 14, wherein, the location of the vehicle being known from a geolocation system and a measured minimum temperature at said location being known from a meteorological information broadcasting system, the model of variation of the ambient temperature is defined such that: the variation is nil when the measured ambient temperature is below the minimum temperature; the variation is equal to a negative constant when the measured ambient temperature is above a predefined value greater than the minimum temperature; and the variation trends linearly between the minimum temperature and the predefined value.

17. The method as claimed in claim 14, wherein the model of thermal inertia of the battery is defined such that the temperature of the battery varies identically with the ambient temperature.

18. The method as claimed in claim 13, wherein the estimating the minimum state of charge includes selecting the maximum value from among a plurality of state-of-charge values, said plurality including at least: a minimum state-of-charge value to reach the current destination, estimated as a function of the measured current temperature of the battery and; a minimum state-of-charge value to reach the future destination, estimated as a function of the temperature previously predicted.

19. The method as claimed in claim 18, wherein the plurality of state-of-charge values also includes a minimum state of charge to ensure a predefined life of the battery.

20. The method as claimed in claim 18, wherein the minimum state-of-charge values to reach the current destination and to reach the future destination are calculated in real time by a method of linear modelling of the trend of the available power in the battery as a function of the state of charge of the battery.

21. The method as claimed in claim 20, wherein the linear modelling method is a recursive least squares method.

22. The method as claimed in claim 21, wherein, on each new estimation of a state-of-charge value obtained recursively from the preceding estimation, said preceding estimation is multiplied by an omission factor <1.

23. A non-transitory computer readable medium storing a program that, when executed by a computer, causes the computer to execute the method as claimed in claim 13.

24. A hybrid vehicle comprising the computer as claimed in claim 23.

Description

[0050] FIG. 1 illustrates by a graph an example of ambient temperature (X) drop values that can be used for the predictive calculation of the future temperature of the battery. In order to avoid power degradations, the prediction of the ambient temperature can be gauged rather pessimistically, for example by taking a drop of Tamb of 15 C. as constant for the positive Tambs. When Tamb reaches its minimum, for example approximately 20 C. in Paris, it is considered that Tamb will not be able to go below, and the Tamb drop is therefore taken to be equal to 0. A linear interpolation between Tamb=0 C. and Tamb=20 C. approx. is an example of variation of Tamb for the predictive calculation according to the invention in Paris.

[0051] Thus, if considering an elapsed time between the current run and the next run that is long enough for the temperature of the battery to have the time to converge toward predicted Tamb, the graph of FIG. 2 illustrates, from the drop of Tamb of FIG. 1, the prediction of future Tbat as a function of current Tamb.

Estimation of the Current Target SOC

[0052] The invention exploits the link between the SOC of the battery and the maximum available power in the battery: when the SOC increases, the maximum available power increases also and vice versa. This link can, for example, be characterized in tables giving the maximum available power of the battery as a function of its temperature and of its state of charge, or by another algorithm implemented in the battery computer.

[0053] One principle of the present exemplary embodiment is to identify this link in real time by a linear modeling, of the type, for example, of the recursive least squares method. Indeed, the linear modeling is coherent if limited to a restricted zone of SOC level. When the battery has a fairly high capacity (which is the case for most hybrid vehicles), the SOC variations are fairly slow to allow a sufficient number of samples making it possible to clearly identify the linear link over the restricted zone. In the present exemplary embodiment, the recursive least squares method can be written as follows, in which a and b denote the real linearization coefficients and t denotes time:

Pbat(t)=a(t)*SOC(t)+b(t)

[0054] Another principle of the present exemplary embodiment is to reduce the weight of the old measurements in favor of the more recent ones, according to an exponential law. For that, on each step of the recurrence, the weight of the old measurements is multiplied by an omission factor <1, thus, at the (n+1)th step: [0055] the first estimation is weighted by .sup.n [0056] the second estimation is weighted by .sup.n1 [0057] the nth estimation is weighted by [0058] the new estimation is weighted by 1

[0059] Take:

[00001]$\left(t\right)=\left(t-1\right)+L\left(t\right)\left[y\left(t\right)-{}^T\left(t\right)\left(t-1\right)\right]$$L\left(t\right)=\frac{P\left(t-1\right)\left(t\right)}{\left(t\right)+{}^T\left(t\right)P\left(t-1\right)\left(t\right)}$$P\left(t\right)=\frac{1}{\left(t\right)}\left[P\left(t-1\right)-\frac{P\left(t-1\right).Math.\left(t\right){}^T\left(t\right)P\left(t-1\right)}{\left(t\right)+{}^T\left(t\right)P\left(t-1\right)\left(t\right)}\right]$

in which: [0060] P=covariance matrix (22) [0061] =vector of the parameters (21) [0062] =control vector (21)
and:

[00002]$\left(t\right)=\left[\begin{array}{c}\mathrm{SOC}\left(t\right)\\ 1\end{array}\right]$$\left(t\right)=\left[\begin{array}{c}a\left(t\right)\\ b\left(t\right)\end{array}\right]$$y\left(t\right)=\mathrm{Pbat}\left(t\right)$

[0063] The advantages of the use of such a method are multiple, but, among others, one that can be cited is the fact that identifying the real time link makes it possible to be robust to temperature variations or even to the aging of the battery. It is also possible to cite the fact that the omission factor makes it possible to adjust the filtering of the learning to render it more or less dynamic.

[0064] In order to increase the robustness of the method, it may possibly be advantageous to add optional enhancements to it, such as discarding the samples in which the variations of SOC and/or of Pbat are too great. They may be measurement noises or undesirable disturbances, or they may even concern charge/discharge peaks during which the battery is too excited and its voltage (which is the image of the maximum available Pbat) increases on a spot basis.

[0065] It is also possible to envisage discarding the samples in which the variations of SOC and of Pbat are of opposite signs, for example measurement noises or transient errors.

[0066] It is also possible to envisage discarding the samples in which the variations of SOC and of Pbat are too small, in order to avoid saturating linearization with duplicated samples, in the case where the vehicle remains stopped with a low consumption of the auxiliaries for example.

[0067] It is also possible to envisage adding a delay in the initialization of the linearization method before taking it into account, in order to ensure a sufficient number of samples (and, during this time, it is possible to use a default value for the target SOC for example).

[0068] It is finally possible to envisage saturating the coefficients of the identified straight line, in order to ensure a minimum slope for varying the SOC, to cover the case in which, from startup, if the vehicle remains stopped for a long time with low consumption of the auxiliaries, the initialization delay would not be sufficient.

[0069] FIG. 3 illustrates the convergence of the calculation of SOC_target, by targeting a minimum available battery power of 45 kW: [0070] at an instant t1: [0071] SOC=19% [0072] Tbat=10 C. [0073] Pbat=12 kW [0074] SOC_current_target=50% [0075] at an instant t2: [0076] SOC=26% [0077] Tbat=0 C. [0078] Pbat=35 kW [0079] SOC_current_target=31% [0080] at an instant t3: [0081] SOC=31% [0082] Tbat=0 C. [0083] Pbat=42 kW [0084] SOC_current_target=41%

[0085] It can be seen in the example of FIG. 3 that, by virtue of the local linearization and the monotonic link between Pbat and SOC, the target SOCs calculated cause the current SOC to rise to converge toward the SOC corresponding to target Pbat. It can also be seen that the calculation adapts also to battery temperature changes (i.e. initial Tbat=Tamb=10 C.; final Tbat=0 C.).

Estimation of the Future Target SOC

[0086] The principle is the same as that described previously for the calculation of SOC_current_target. The BMS sends the projection of the maximum available power level in the battery corresponding to the current SOC and to the future Tbat. Since the level of demand for performance on departure for the future run can be different relative to the demands of the current run, the targets of Pbat discharge max can therefore differ.

[0087] FIG. 4 illustrates an example in which current Tamb is 15 C. (predicted Tbat=20 C.) and target Pbat for the future run is 15 kW: [0088] at an instant t1: [0089] SOC=19% [0090] Tbat on next run=20 C. [0091] Pbat=5 kW [0092] SOC_future_target=37% [0093] at an instant t2: [0094] SOC=26% [0095] Tbat on next run=20 C. [0096] Pbat=1135 kW [0097] SOC_future_target=33% [0098] at an instant t3: [0099] SOC=31% [0100] Tbat on next run=20 C. [0101] Pbat=15 kW [0102] SOC_future_target=31%

[0103] As in the preceding case, it can be seen in the example of FIG. 4 that, by virtue of the local linearization and of the monotonic link between Pbat and the SOC, the target SOCs calculated cause the current SOC to rise to converge toward the SOC corresponding to target Pbat.

Estimation of the Final Target SOC

[0104] The final target SOC is the maximum between: [0105] The target SOC calculated for the current run; [0106] The target SOC calculated for the future run; [0107] The minimum target SOC necessary to ensure the regulation of the SOC in charge sustaining mode.

[0108] The invention described previously therefore clearly has the main advantage of adjusting the battery charge level as a function of the internal and the ambient temperatures, in order to guarantee a minimum required performance level (whether it be for vehicle takeoff or for any other service provided, for both the current run and the next run.