EP1570281A1

EP1570281A1 - Method for predicting the voltage of a battery

Info

Publication number: EP1570281A1
Application number: EP03757996A
Authority: EP
Inventors: Richard Bopp; Armin KLÖPFER
Original assignee: DaimlerChrysler AG
Current assignee: Daimler AG
Priority date: 2002-12-09
Filing date: 2003-10-16
Publication date: 2005-09-07
Also published as: WO2004053509A1; DE10257588B3; US20060055373A1; JP2006509205A; US7332892B2

Abstract

The invention relates to a method for predicting a voltage of a battery, particularly a vehicle battery. The inventive method makes it possible to predict a drop in voltage before it actually occurs due to a load being applied. According to said method, a filtered battery voltage and a filtered battery current are first determined from battery-related data, such as the battery voltage, battery current, battery temperature, and dynamic internal resistance. An ohmic drop is determined from a differential current between the filtered battery current and a given load current via the dynamic internal resistance. Furthermore, a polarization voltage is calculated as a function of the filtered battery current and is then filtered. A predicted battery voltage is calculated from the filtered battery voltage minus the ohmic drop and the filtered polarization voltage. A decision concerning additional measures can be made based on said predicted battery voltage.

Description

Method for predicting a voltage of a battery

The invention relates to a method for predicting a voltage of a battery, in particular a vehicle battery.

Conventionally, there is the problem that, for example, in a vehicle electrical system, the voltage drops when the battery is bad or discharged under certain loads to such an extent that important systems, such as the braking system, no longer have the full range of functions, and the driver may only have a large vehicle Can operate restrictions.

From DE 39 36 638 Cl a method is known in which the consumers in a vehicle electrical system are switched off or down when the battery of the vehicle falls below a certain charge level in order to prevent the battery from being over-discharged. Which or which consumers are or will be switched off depends on which group of consumers they belong to. Such a group is composed, for example, of "conditionally switchable consumers" (BSV) and / or "switchable consumers" (SV). The group is always completely switched off or reduced in consumption. Each group has a priority regarding the vehicle safety time or its importance. The switching off or switching back of the individual groups begins with the group with the lowest priority. If this does not lead to an improvement in the state of charge of the battery, further groups are subsequently switched off or switched back until the battery charge level reaches a certain level.

Furthermore, from DE 199 60 079 AI a method for Einzw. Disconnection of various classes of consumers by means of switching elements in the context of an energy management carried out by a control unit. The switching elements are activated in such a way that the selected priorities for the activation of the switching elements can be changed dynamically during operation. This makes it possible to adapt the switching priorities depending on the operating state during operation. The consumers are switched off by changing the switching priority so that the perceptibility of the operating states is suppressed as far as possible.

With these conventional methods, a consumer or a group of consumers is only switched off or switched back on when a poor state of charge has already been recognized. In order to prevent a safety-relevant system, such as the braking system, from being fully functional due to a downshift, a calculation-intensive method for calculating the state of charge of the battery is currently being used, which significantly increases the costs of the associated control unit.

It is therefore an object of the present invention to provide a simple and inexpensive method for predicting a voltage of a battery, with which a condition of a bad or discharged battery, in which a voltage dip can occur under certain loads, can be predicted and enables this , take appropriate countermeasures before this condition occurs so that certain safety-relevant consumers remain fully functional. According to the invention, this object is achieved by a method for predicting a voltage of a battery with the features according to claim 1. Further advantageous developments of the invention are specified in the subclaims.

The method according to the invention for predicting a voltage of a battery now enables timely detection of critical battery states, in particular critical on-board electrical system states in the vehicle, and initiation of countermeasures, such as a consumer shutdown or an increase in the engine speed.

These and other objects, features and advantages of the present invention will become apparent from the following description of a preferred embodiment in conjunction with the drawing.

It shows:

Fig. 1 is a flowchart of a method according to the invention for predicting a voltage of a battery U _j pred

FIG. 2 shows a flowchart of a subroutine "calculation of the polarization voltage U_pol" from FIG. 1,

Fig. 3 is a flowchart of a subroutine "filtering the polarization voltage U_pol" from Fig. 1 and

4 shows a representation of exemplary current-dependent curves of the polarization voltage.

The method according to the invention for predicting a voltage of a battery, in particular a vehicle battery, is now described in more detail below with reference to the flow diagrams according to FIGS. 1 to 3.

To ensure that certain safety-critical consumers, such as Sensotronic Brake Control or SBC (electrohydraulic brake) remain fully functional, a vehicle battery voltage must not fall below a certain minimum voltage since a voltage drop occurs when a load is applied. With the aid of the method according to the invention, it can now be predicted which battery voltage U_pred will be set when discharging with a predetermined current I_pred, ie a defined expected load.

In a first step S1, current battery data, such as, for example, the battery voltage U_batt, the battery current I_batt, the battery temperature T_batt and the dynamic internal resistance Rdi of the battery are recorded or queried by external detection and calculation devices. The battery voltage U_batt, the battery current I_batt and the battery temperature T_batt are detected by sensors and transmitted to a control device or queried by the control device which carries out the method according to the invention for predicting the voltage of a battery. The dynamic internal resistance Rdi is calculated by a known routine and the calculation result is likewise transmitted to the control device or queried by the control device. Such a method for calculating the dynamic internal resistance Rdi is known, for example, from DE 102 08 020 AI, in which the value obtained for the dynamic internal resistance is already filtered. The values for the battery voltage U_batt, the battery current I_batt, the battery temperature T_batt and the dynamic internal resistance Rdi are transmitted to the control device at predetermined intervals t, for example every 50 ms, or are queried by the control device. With the recorded values of the battery current I_batt, negative values mean a discharge and positive values mean a charge of the battery.

It is then checked in a step S2 whether this functional sequence is a first sequence. For this purpose, the status of a bit is checked, which at a the first functional sequence is set and is reset with each restart. If the bit is set, that is to say a functional sequence (steps S1 to S12) has already taken place, the sequence proceeds to step S3. Otherwise, the process proceeds directly to step S5, so that a quick prediction of the voltage of the battery is possible immediately after the restart.

In a step S3 it is determined whether a time Tx, here 500 ms, has already expired, i.e. after 500 ms, the process proceeds to step S4, otherwise the process returns to step S1.

If it is determined in step S3 that the conditions are met, the battery voltage U_batt and the battery current I_batt are filtered in a step S4 by means of a low-pass filter. The filtering determines a filtered battery voltage value U_filt and filtered battery current value I_filt from the battery voltage U_batt and the battery current I_batt, from which the ripple is filtered out. The filtered battery voltage value U_filt and the filtered battery current value I_filt after the low-pass filtering result from the following equations:

U_filt (t _n ) = (U_batt - U_filt (tn-i)) * (1 - exp (-t / T)) + + U_filt (t _n- ι)

I_filt (t _n ) = (I_batt - I_filt (t _n -ι)) * (1 - exp (-t / T)) + + I_filt (t _n-1 )

Here T is a filter constant that is selected, for example, as 500 ms, while t is an interval in which a set of values is read out in each case and that is, for example, 50 ms. t _n is the current time, while t _{n-1 is} the time of the last calculation. If no previous calculation has been carried out, predetermined initialization values are used. On the basis of settling times of a low-pass filter used, values are defined for the initialization as follows, for example: U_filt = 11.8, I_filt = 0.0 and Rdi = 5.0.

The input variables are read into the low-pass filter as quickly as possible, provided the values are valid, i.e. the hardware for recording the battery voltage U_batt and the battery current I_batt must supply valid values. When the function for predicting a voltage of a battery is called for the first time, for example after a period Tx, i.e. in the example 500 ms a quick prediction is carried out. The filtering by the low pass is not included, i.e. steps S3 and S4 are skipped in the first function call. In the first 5 seconds after this function call, for example, all time constants are set to 1 second, since this enables the method to settle quickly.

The calculation of the predicted battery voltage U_pred, i.e. the functional sequence takes place after a period T, i.e. in the example after 500 ms.

The prediction of the predicted battery voltage U_pred is only carried out if the battery current I_batt is greater than the predetermined load current I_pred to which the prediction is based. A predetermined tolerance Toi, for example of 5A, is permitted. A calculation for a battery current I_batt less than I_pred is not necessary since the current voltage drop would then be greater than a voltage drop to be predicted. Therefore, the flow then returns to step S1.

It is thus checked in step S5 whether the following conditions required for performing the calculation of the predicted battery voltage U_pred are met: I_filt> (I_pred - Toi) and

I_batt> I_pred - Toi

The second-mentioned condition is also queried here, since larger currents are reached during a starting process, but a calculation would otherwise be permitted due to the filter. However, such errors should be excluded.

If it is recognized in step S5 that the conditions listed above are not met, the prediction of the predicted battery voltage Ujpred is not carried out and the process returns to step S1.

If it is recognized in step S5 that the above conditions are met, an ohmic voltage drop across the dynamic internal resistance Rdi is then calculated in step S6. For this purpose, the voltage drop U_ri resulting from the predetermined load current I_pred at the dynamic internal resistance Rdi is calculated from the filtered battery current and internal resistance values (I_filt and Rdi) in accordance with the following formula:

U_ri = (I_filt - I_pred) * Rdi

Since the predetermined load current I_pred is always a discharge current, it must also be used negatively. The range of values for the predetermined load current I_pred is, for example, between -80A and -150A.

A polarization voltage U_pol is then calculated in a step S7. The subroutine for calculating the polarization voltage in step S7 is shown in more detail in FIG. 2. The polarization voltage U_pol has several chemical causes, ie it consists of several partial voltages. These partial voltages include a breakdown voltage or activation voltage, a crystallization on voltage and a diffusion voltage. The breakdown voltage arises from the fact that the local distribution of the ions first has to build up when the current changes, but this does not happen as quickly as the current sets in, and the distribution of the charged particles on the surface is comparable to a capacitor. The crystallization voltage is the voltage required to detach molecules on the surface of the electrode from their assembly and make them accessible for a reaction. Finally, the diffusion voltage is the voltage required to remove the reaction products from the electrode surface. These partial voltages each have a dependency on the battery current, namely the current magnitude and the current direction, and the temperature, which has an e-function.

The polarization voltage U_pol as a whole can be described precisely enough by two simple reciprocal functions. The polarization voltage U_pol can be determined as follows, distinguishing in each case whether the battery is being charged, i.e. I_filt> 0, or the battery is discharged, i.e. I_filt <0.

It is therefore first decided in a step S7-1 whether the filtered battery current I_filt is greater than zero. Depending on the decision result, in step S7-2a or S7-2b the polarization voltage U_pol is calculated according to the following formula:

For I_filt> 0:

U_pol = (U_pol_0 + (ki_lad * I_filt / (ik_lad + l_filt))) * K.

For I_filt ≤ 0:

U_pol = (U_pol_0 + (ki_ela * I_filt / (ik_ela - l_filt))) * Ki

Kx in the above equations is a correction factor that is 1 when the predetermined load current I_pred is -100A while it is for a predetermined load current I_pred in the range between -80A to -150A results from (1- (I_pred + 100) / 100 * 0, 2). It is obvious to the person skilled in the art that if a load current range deviating from this load current range is desired, a corresponding, adjusted correction value can be determined.

The parameters U_pol_0, ki_lad, ik_lad, ki_ela and ik_ela are predetermined parameters. For example, U__pol__0 can be 0.7V at 0 ° C. The temperature dependency is - 9mV / ° C. This results in:

U_pol_0 = 0.7V - 0.009V / ° C * T_batt [T_batt in ° C]

ik_lad and ik_ela are empirical parameters that describe the curvature of the curve of the polarization voltage U_pol as a function of the filtered battery current I_filt. 4 shows such a curve profile for different battery temperatures T_batt. For example, the value for ik_lad can be 80A and the value for ik_ela can be 20A. ki_ela is dimensionless and should be defined so that when I_filt = I_pred the value for U_pol = 0V.

So the following applies:

ki_ela = U_pol_0 * (ik_ela - I_pred) / (-I_pred * [VA]) or ki_lad = U_pol_0 * (ik_lad - I_pred) / (-I_pred * [VA]) * K ₂ .

A correction factor K ₂ must be taken into account when loading, since very large overvoltages can occur during loading, which would be too large for a calculation. Through this correction or Compensation factor K ₂ , these voltages can also be calculated.

This description of the polarization voltage U_pol applies when the battery is in a quasi-static, ie steady state, ie when the battery current I batt is constant. Due to the chemical reactions hidden behind this phenomenon, the polarization voltage U_pol changes only slowly. The change follows two superimposed time constants. The parameter U_pol determined as described above thus consists of a fast and a slowly settling part U_pol_fast_roh and U_pol_slow_roh:

U_pol_fast_roh = 0.6 * U_pol and U_pol_slow_roh = 0.4 * Ujool,

i.e. 60% of U_pol swings in quickly and 40% swings in slowly.

Therefore, in a further step S8, the exact sequence of which is illustrated in more detail in FIG. 3, the polarization voltage U_pol is filtered, this filtering preferably being carried out by two low-pass filters, one for a rapidly settling part U_pol_fast_roh of U_pol and one for a slowly settling part U_pol_slow_roh share of U_pol.

First, in a step S8-1, the polarization voltage U_pol is divided into the still unfiltered raw values of the polarization voltage U_pol_fast_roh and U_pol_slow_roh. Then, in step S8-2, these two polarization voltage components U_pol_fast_roh and U_pol_slow_roh are filtered by means of two low-pass filters.

This results in:

U_pol_fast_filt (t _n ) = (U_pol_fast_roh - Upol_fast_filt (t _n -ι) *

* T + U_pol_fast_filt (t _n -ι)

U_pol_slow_filt (t _n ) = (U_pol_slow_roh - Upol_slow_filt (t _n _ι) *

* T + U_pol slow filt (t _n -ι) The time constants of the low-pass filter for U_pol_fast_roh and • U_pol_slow_roh are different, depending on whether charging, ie I_filt> 0, or discharging, ie I_filt ≤ 0. The time constants are, for example:

For I_filt> 0:

T for U_pol_fast_filt = 1 second

T for U_pol_slow_filt = 1 minute

For I_means ≤ 0:

T for U_pol_fast_filt = 1 second

T for U_pol_slow_filt = 30 seconds

In a further step S8-3, the filtered values of the two polarization voltage components U_pol_fast_filt and U_pol_slow_filt are then added in order to obtain a filtered polarization voltage U_pol_filt.

These values for parameters for determining the polarization voltage are also only examples and do not constitute a restriction.

In a step S9 carried out subsequently, the predicted battery voltage U_pred is then calculated from the voltage values for the filtered battery voltage U_filt, the ohmic voltage drop U_ri and the filtered polarization voltage U_pol_filt determined in steps S4, S6 and S7 and S8:

Ujpred = U_filt - U_ri - U_pol_filt

The predicted battery voltage U_pred determined in this way in step S9 is still limited upwards and downwards in step S10, for example by specifying the maximum value U_pred_max 12.5V and the minimum value U_pred_min 10V. An upper limit is not absolutely necessary, since there the battery charge is sufficient in any case; the- in the preferred exemplary embodiment, the maximum value U_pred_max is also set to a value close to a normal value of a full battery in the idle state. However, the downward limitation by a minimum value U_pred_min is necessary in any case, since the battery has aged, discharged from this voltage value in such a way that it is no longer possible to reliably predict the battery voltage from this threshold value due to an exponentially falling voltage. In the event that the predicted battery voltage U_pred is within the limit values U_pred_min and U_pred_max, the predicted battery voltage is then filtered in a further step S11, the time constant T of this filter being 3 minutes for both negative and positive current values. This further filtering in step S11 filters out jumps that occur as a result of a switch from loading to unloading.

This results in:

U_pred_filt (t _n ) = (U_pred_roh - U_pred_filt (t _n- ι)) * T +

+ U_pred_filt (t _n-1 )

where T is selected as 3 minutes, for example.

Finally, a check is carried out in step S12 as to whether the bit that indicates whether a first function call has already been set is set. If it is not set, this bit is set and then the process returns to step S1. Otherwise, the process returns directly to step S1.

In this way, a voltage of a battery, in particular a vehicle battery, can be reliably determined when the load current I_pred is predetermined. This prediction can be used for batteries of all types, in particular for vehicle batteries of all types, sizes and capacities. In summary, the present invention discloses a method for predicting a voltage of a battery, in particular a vehicle battery. The method according to the invention makes it possible to predict a voltage drop before it actually occurs due to a load. For this purpose, a filtered battery voltage and a filtered battery current are first determined from battery data, such as battery voltage, battery current, battery temperature and dynamic internal resistance. An ohmic voltage drop over the dynamic internal resistance is determined from a differential current between the filtered battery current and a predetermined load current. In addition, a polarization voltage is calculated as a function of the filtered battery current, which is then filtered. A predicted battery voltage is calculated from the filtered battery voltage minus the ohmic voltage drop and the filtered polarization voltage. Based on this predicted battery voltage, further measures can be decided.

Claims

claims

A method for predicting a voltage of a battery, comprising the steps:

(51) Acquisition or interrogation of battery data, of acquisition and calculation devices, the battery data being a battery voltage (U_batt), a battery current

(I_batt), a battery temperature (T_batt) and a dynamic internal resistance (Rdi),

(52) Checking whether the current functional flow is a first flow,

(53) if the result in step S2 is that an operation has already taken place, check whether a predetermined time (Tx) has passed, and if the predetermined time has not yet passed, return to step S1,

(54) when the predetermined time (Tx) has elapsed, filter the battery voltage (U_batt) and the battery current

(I_batt) using a low-pass filter and outputting a filtered battery voltage (U_filt) and a filtered battery current (I_filt),

(55) Check whether the filtered battery current (I_filt) is greater than a predetermined load current I_pred minus a tolerance (Toi) and the battery current (I_batt) is greater than a predetermined load current (I_pred) minus the tolerance (Toi), and if the conditions are not met, return to step S1,

(56) calculating an ohmic voltage drop (U_ri) at the dynamic internal resistance (Rdi), (57) calculating a polarization voltage (U_pol) as a function of the filtered battery current (I_batt_filt),

(58) filtering the polarization voltage (U_pol) by means of two low-pass filters separated according to a quickly settling component (U_pol_fast_roh) and a slowly settling component (U_pol_slow_roh) and outputting a filtered polarization voltage (U_pol_filt),

(59) calculating a predicted battery voltage by subtracting the ohmic voltage drop (U_ri) and the filtered polarization voltage (U_pol_filt) from the filtered battery voltage (U_batt_filt),

(510) limit the predicted battery voltage (U_pred) determined in step S9 up and down,

(511) filtering the predicted battery voltage (U_pred) and

(512) Check whether the bit indicating a first function call has been set and, if not, set the bit and return to step S1, or if yes, return to step S1.

2. A method for predicting a voltage of a battery according to claim 1, so that the dynamic internal resistance (Rdi) is determined by means of a buffer algorithm.

3. The method for predicting a voltage of a battery according to claim 1 or 2, so that the predetermined time (Tx) in step S3 is 500ms.

4. A method for predicting a voltage of a battery according to one of claims 1 to 3, characterized in that that the filtered battery voltage (U_filt) and the filtered battery current (I_filt) result from the following equations:

U_filt (t _n ) = (U_batt - U_filt (tn-i)) * (1 - exp (-t / T)) +

I_filt (t _n ) = (I_batt - l_filt (t _n -ι)) * (1 - exp (-t / T)) +

where T is a filter constant, t is an interval in which a set of values is read out and t _n is the current time, while t _n -ι is the time of the last calculation.

5. A method for predicting a voltage of a battery according to one of claims 1 to 4, so that steps S3 and S4 are skipped in a first function call immediately after a start

6. A method for predicting a voltage of a battery according to any one of claims 1 to 6, d a d u r c h g e k e n n z e i c h n e t that the tolerance (Toi) is selected as 5A.

7. A method for predicting a voltage of a battery according to any one of claims 1 to 6, d a d u r c h g e k e n n z e i c h n e t that the ohmic voltage drop based on the following

Equation is calculated:

U_ri = (I_filt - I_pred) * Rdi Method for predicting a voltage of a battery according to one of Claims 1 to 7, characterized in that the polarization voltage (U_pol) is calculated taking into account the specified conditions in accordance with the following equations:

For I_filt> 0:

U_pol = (U_pol_0 + (ki_lad * l_filt / (ik_lad + I_filt))) *

* Ki.

For I_filt ≤ 0:

U_pol = (U_pol_0 + (ki_ela * I_filt / (ik_ela - I_filt))) *

where K is a correction factor which is dependent on the predetermined load current (I_pred) and the parameters U_pol_0, ki_lad, ik_lad, ki_ela and ik_ela are predetermined parameters which have been determined empirically, and ki_ela is to be determined such that if the filtered battery current (I_filt ) and the predetermined load current (I_pred) the value of the polarization voltage (U_pol) is 0V.

Method for predicting a voltage of a battery according to one of Claims 1 to 8, characterized in that the polarization voltage (Ujpol) comprises a rapidly settling portion (U_pol_fast_roh) and a slowly settling portion (U_pol_slow_roh), the rapidly settling portion (U_pol_fast_roh) comprising 60% the polarization voltage (U_pol) and the slowly settling component (U_pol_slow_roh) make up 40% of the polarization voltage (U_pol) and each of these two components is filtered in step S8 by a low-pass filter , so that the following equations result:

U_pol_fast_filt (t _n ) = (U_pol_fast_roh -

- Upol_fast_filt (t _n -ι) *

* T + U_pol_fast_filt (t _n -ι)

U_pol_slow_filt (t _n ) = (U_pol_slow_roh -

- Upol_slow_filt (t _n-1 ) * * T + U_pol_slow_filt (t _n -ι)

and the total filtered polarization voltage (U_pol_filt) is obtained by adding the two filtered parts of the polarization voltage (U_pol_fast_filt, U_pol_slow_filt).

10. A method for predicting a voltage of a battery according to claim 8, characterized in that the correction factor K _x 1 when the predetermined load current (I_pred) is -100A, while it is for a predetermined load current (I_pred) between -80A and -150A from (1- (I_pred + 100) / 100 * 0.2) results.