DE102022202373A1 - Model-based predictive control of a battery charging process - Google Patents

Abstract

The invention relates to a model-based predictive control of a charging process of a battery (7) for a drive train (5) of a motor vehicle (1) using a domestic charging unit (3). A method in this regard includes in particular the steps of - determining secondary conditions by executing a secondary conditions algorithm, whereby the secondary conditions are determined depending on the state of charge information of the battery (7) and time information, and - determining charging currents by means of which the battery (7) is carried out the domestic charging device (3) is to be charged at N discrete times within an upcoming prediction horizon, by executing an MPC algorithm, the charging currents being determined depending on the additional conditions determined by executing the constraints algorithm and while minimizing a cost function of the MPC algorithm' takes place.

Description

The invention relates to a model-based predictive control of a charging process of a battery, in particular a battery for a drive train of a motor vehicle, wherein the battery is charged using a domestic charging unit. What is required in this context is, in particular, a method and a processor unit.

In modern electric vehicles, the positive and negative effects of certain driving maneuvers on the efficiency of the vehicle are increasingly being considered and corresponding operating and driving strategies are derived. A rapidly growing approach here is trajectory planning with model predictive control. In the case of driving maneuver planning, a large number of trajectories are evaluated over a virtual horizon with regard to the fulfillment of a target function (e.g. "minimize the energy requirement") and compliance with the applicable restrictions or "constraints" (e.g. speed limits, minimum speed), the optimal trajectory is selected and ultimately implemented.

An object of the present invention can be seen in providing an improved control system for charging a battery of an electric motor vehicle. The task is solved by the subject matter of the independent patent claims. Advantageous embodiments are the subject of the subclaims, the following description and the figures.

According to the present invention, it is proposed, assuming that an executive control device or a domestic charging unit with a corresponding CPU as well as at least one domestic energy supplier is available, to apply an MPC approach to domestic charging of the battery. For this purpose, a new target function as well as constraints or secondary conditions are formulated. The model-based predictive control (MPC) method is based on a system model that describes the behavior of the system. Furthermore, the MPC method is based on an objective function or a cost function that describes an optimization problem and determines which state variables should be minimized. The appropriate optimization problem and the corresponding applicable solution space are identified and a target function including constraints or side conditions is derived, so that it is then possible to use an MPC solver to solve the optimization problem.

In particular, a combination of a human-machine interface (English-language technical term: human-machine interface; abbreviated: HMI), a prediction model, a constraint handler, and an MPC solver is proposed. The subject of the invention is in particular the derivation of the applicable constraints from the HMI inputs as well as the identification of an optimal charging curve based on a prediction of the expected domestic energy requirement using an MPC solver. In particular, this assumes that an existing energy supplier cannot feed into the grid at the applicable withdrawal conditions, since in this case it would not be a multi-dimensional optimization problem and the use of the MPC approach would be unnecessary. In the case of the presence of an energy supplier as described above, a typical domestic scenario results in which the interaction between the electric vehicle, local energy supplier and household electricity demand can be optimized predictively.

• - Ermitteln von Nebenbedingungen durch Ausführen eines Nebendingungs-Algorithmus', wobei die Nebenbedingungen in Abhängigkeit von Ladezustandsinformationen der Batterie und Zeitinformationen ermittelt werden, und
• - Ermitteln von Ladeströmen, mittels welcher die Batterie durch die häusliche Ladeeinrichtung innerhalb eines vorausliegenden Prädiktionshorizonts zu N diskreten Zeitpunkten geladen werden soll, durch Ausführen eines MPC-Algorithmus', wobei das Ermitteln der Ladeströme in Abhängigkeit von den durch Ausführen des Nebendingungs-Algorithmus` ermittelten Nebenbedingungen und unter Minimierung einer Kostenfunktion des MPC-Algorithmus' erfolgt.

In this sense, according to a first aspect of the invention, a method for model-based predictive control of a charging process of a battery for a drive train of a motor vehicle by means of a domestic charging unit is proposed. The procedure includes in particular the following steps:

• - Determining secondary conditions by executing a secondary condition algorithm, whereby the secondary conditions are determined depending on the battery charge status information and time information, and
• - Determining charging currents by means of which the battery should be charged by the domestic charging device at N discrete times within a prediction horizon, by executing an MPC algorithm, the charging currents being determined depending on those determined by executing the secondary condition algorithm Additional conditions and minimizing a cost function of the MPC algorithm are carried out.

• - Nebenbedingungen durch Ausführen eines Nebendingungs-Algorithmus' zu ermitteln, wobei die Nebenbedingungen in Abhängigkeit von Ladezustandsinformationen der Batterie und Zeitinformationen ermittelt werden, und
• - durch Ausführen eines MPC-Algorithmus' Ladeströme zu ermitteln, mittels welcher die Batterie durch die häusliche Ladeeinrichtung innerhalb eines vorausliegenden Prädiktionshorizonts zu N diskreten Zeitpunkten geladen werden soll, wobei das Ermitteln der Ladeströme in Abhängigkeit von den durch Ausführen des Nebendingungs-Algorithmus` ermittelten Nebenbedingungen und unter Minimierung einer Kostenfunktion des MPC-Algorithmus' erfolgt. Die Prozessoreinheit kann dabei insbesondere in eine Ladeeinheit eines Hauses oder in ein Kraftfahrzeug integriert sein, insbesondere in ein Steuergerät des Kraftfahrzeugs. Die folgenden Ausführungen gelten gleichermaßen für das Verfahren gemäß dem ersten Aspekt der Erfindung und für die Prozessoreinheit gemäß dem zweiten Aspekt der Erfindung.

According to a second aspect of the invention, a processor unit for model-based predictive control of a charging process of a battery for a drive train of a motor vehicle is provided by means of a domestic charging unit. The processor unit is designed in particular to

• - Determine secondary conditions by executing a secondary condition algorithm, whereby the secondary conditions are determined depending on the charge status information of the battery and time information, and
• - to determine charging currents by executing an MPC algorithm, by means of which the battery is to be charged by the domestic charging device at N discrete times within an upcoming prediction horizon, the charging currents being determined depending on the secondary conditions determined by executing the constraints algorithm and while minimizing a cost function of the MPC algorithm. The processor unit can be integrated in particular in a charging unit of a house or in a motor vehicle, in particular in a control unit of the motor vehicle. The following statements apply equally to the method according to the first aspect of the invention and to the processor unit according to the second aspect of the invention.

In addition to the cost function, the MPC algorithm can in particular contain a charging model that describes the charging of the battery. If the MPC algorithm is executed, in particular by means of the processor unit, then a charging process of the battery is determined or predicted based on the charging model. A possible solution space is limited by the additional conditions and the remaining possible solutions are evaluated using the cost function.

Physically, charging currents in particular are calculated or specified by executing the MPC algorithm. Due to the charging voltage available at home (usually 230V or 400V), the product of the charging current and charging voltage results in a charging power. The planned charging curve corresponds to the charging power at the N different discretization points above the charging horizon. The charging horizon is defined in particular as the period of time that extends from a current point in time to an end point in time at which the battery is charged with a desired energy content. In this sense, the prediction horizon is defined in one embodiment as a temporal loading horizon, which represents the period between a loading end time and a current time.

The discretization of the charging horizon takes place in particular in N-1 equally distributed time periods. If N remains the same, these sections become smaller as the loading time increases and the optimization result becomes correspondingly more precise. In this sense, according to one embodiment, the charging horizon is or is divided into N-1 equally distributed time periods.

In order to be able to optimize the charging curve, the MPC algorithm contains an optimization goal in the form of the cost function, which can in particular be a continuously differentiable minimization function. The basic goal here is to minimize costs. Since the costs in this case relate to the electricity costs incurred by the household and the charging process, it can be specified for the cost function to be minimized in particular that the cost function to be minimized contains a first positive term, a second positive term and a third negative term, where the first term represents the charging current by which the battery is to be charged by the domestic charging device within one of the time periods. The second term represents an average domestic electricity demand that occurs in a house within the time period to which the domestic charging unit belongs. The domestic electricity demand describes in particular the average electricity demand of the house that occurs at time i for the duration of the following discretization section. The third term represents an average output current that is delivered to the house by an electricity supplier. The delivery current describes in particular an averaged current that can be delivered by an electricity supplier to the household or the house and the associated domestic charging unit over the same period of time. The electricity supplier may be an electricity provider separate from the house that provides electricity to the house via a power line. Alternatively, the electricity supplier can also be part of the house, e.g. in the form of a photovoltaic system.

Both the average domestic electricity demand and the average output current are the result of a prediction model that extends over the charging horizon. As a current intensity secondary condition for the three current intensities mentioned above, it can also be specified that the sum of the charging current and the average domestic electricity requirement minus the average output current does not exceed a current limit value which represents the maximum technically adjustable current intensity of a power connection in the house. With the help of the cost function, the MPC algorithm can test any possible charging curves with regard to their cost optimality. The optimization space is mathematically trimmed by the secondary conditions, which ensures that the optimal charging curve found is both physically correct and logically feasible.

As far as determining the state of charge secondary condition is concerned, the necessary state of charge information can, on the one hand, describe a target state of charge up to which the battery should be charged. Furthermore, the state of charge information can also describe a current state of charge, which represents a current state of charge of the battery. Using this information, a first state of charge constraint can be determined by executing the constraint algorithm, which represents the difference between the target state of charge and the current state of charge of the battery. The first charge state secondary condition describes an energy content that must be supplied to the battery as a minimum and corresponds to a net energy value. In addition, a second state of charge secondary condition can be specified, according to which the current state of charge must not exceed a maximum possible state of charge of the battery.

With regard to the determination of the charging time secondary condition, the time information required for this can, on the one hand, describe a charging end time at which the battery should be charged up to the target state of charge. Furthermore, the time information can also describe a current time. Using this information, a loading time constraint can be determined by executing the constraint algorithm, which represents the period between the loading end time and the current time. The charging time secondary condition therefore describes the maximum period of time that may be used to charge the battery to the specified level.

The target charging state and the charging end time can be specified or entered in particular using a human-machine interface. A human-machine interface (HMI) can generally be understood as a function or component of a specific device or software application that allows humans to operate and interact with machines. Some examples of common HMI devices are touchscreens, keyboards or voice controls.

According to a further embodiment, the current charge status and the current time are determined using a detection unit of the motor vehicle. This is particularly efficient since the detection unit of the motor vehicle typically detects the charging status and/or the current time anyway.

• 1 eine Seitenansicht eines Kraftfahrzeugs mit einer Batterie, eines Hauses mit einer Ladeeinheit sowie eines System zur modelbasierten prädiktiven Regelung,
• 2 einen Ablauf eines Verfahrens zur modelbasierten prädiktiven Regelung eines Ladevorgangs der Batterie mittels der Ladeeinheit nach 1 und
• 3 einen Prädiktionshorizont der modelbasierten prädiktiven Regelung.

Exemplary embodiments of the invention are explained in more detail below with reference to the schematic drawing, with the same or similar elements being provided with the same reference numerals. This shows

• 1 a side view of a motor vehicle with a battery, a house with a charging unit and a system for model-based predictive control,
• 2 a sequence of a method for model-based predictive control of a charging process of the battery using the charging unit 1 and
• 3 a prediction horizon of model-based predictive control.

1 shows a motor vehicle 1, for example a passenger car, a house 2 with a domestic charging unit 3 and a system 4 for model-based predictive control (hereinafter referred to as “MPC System 4”).

The motor vehicle 1 includes a drive train 5, which has an electric machine 6, which can be operated as a motor and as a generator, and a battery 7. During engine operation, the electric machine 6 can drive wheels of the motor vehicle 1 via a gearbox 8 of the drive train 5, the gearbox 8 being able to have a constant ratio, for example. The battery 7 can provide the electrical energy required for this. The battery 7 can be charged by the electrical machine 6 when the electrical machine 6 is operated in generator mode (recuperation). The battery 7 can also be charged on the home charging unit 3. The drive train of the motor vehicle 1 can also optionally have an internal combustion engine 9, which can drive the motor vehicle 1 in addition to the electric machine 6. The internal combustion engine 9 can also drive the electric machine 6 to charge the battery 7. The motor vehicle 1 further comprises a detection unit 13 for recording status data relating to the motor vehicle 1. The detection unit 13 can measure current state variables of the motor vehicle 1, record corresponding data and make it available to the MPC system 4. In the context of the present invention, these are in particular current charge states of the battery 7 and current times. The processor unit 3 can access this information, for example via a communication interface.

In the exemplary embodiment shown, the MPC system 4 comprises a processor unit 10, a memory unit 11 and a communication interface 12. A computer program product 14 can be stored on the memory unit 11. The computer program product 14 can be executed on the processor unit 10, for which purpose the processor unit 10 and the memory unit 11 are connected to one another by means of the communication interface 12. When the computer program product 14 is executed on the processor unit 10, it instructs the processor unit 10 to fulfill the functions described in connection with the drawing or to carry out method steps. The MPC System 4 is in 1 shown as a unit separate from the motor vehicle 1 and from the domestic charging unit 3. However, this is purely an example. The MPC system 4 can also be integrated into the motor vehicle 1 or into the domestic charging unit 3.

The computer program product 14 contains an MPC algorithm 15 for model-based predictive control of a charging process of the battery 7 using the domestic charging unit 3. The computer program product 14 further contains a secondary condition algorithm 27 for determining secondary conditions for the MPC algorithm 15, by executing the secondary condition -Algorithm 27 additional conditions for the MPC algorithm 15 are determined. The calculation of the additional conditions takes place depending on the charge status information of the battery and depending on time information, as will be described in more detail below.

The MPC algorithm 13 contains a charging model 16 for the battery 7 and a cost function 17 to be minimized. The processor unit 10 executes the MPC algorithm 15 and thereby predicts a charging process of the battery 7 based on the charging model 16, with the cost function 17 being minimized . The output of the optimization by the MPC algorithm 15 results in optimized charging currents or charging powers recorded by the battery 7 for N calculated points within a respective forecast horizon of the model-based predictive control. The processor unit 10 can control the domestic charging unit 3 based on the calculated charging currents or charging powers. However, this can also be done by a control unit 18 of the domestic charging unit 3.

The domestic charging unit 3 includes an electrical charging cable 19 on the output side, which can be electrically connected to an electrical charging connection 20 of the motor vehicle 1. The charging connection 20 is electrically connected to the battery 7, so that the battery 7 can be charged with electricity from the domestic charging unit 3 via the charging cable 19. The domestic charging unit 3 further comprises a power input connection 21 on the input side, via which the domestic charging unit 3 is supplied with current or voltage, which is provided by a power line 22. In the exemplary embodiment shown, a current flows through the power line 22, which is provided by an electricity provider that generates electricity centrally and passes it through the power line 22. Alternatively, the electricity can also be generated by a power generation system 23 set up for this purpose (e.g. a photovoltaic system) in the house 2 and passed through the power line 22. For example, a voltage can be applied to the power input connection 21 via the power line 22, which is 230V or 400V, which corresponds to a usual voltage available at home.

A user 25 can enter specifications regarding the charging of the battery 7 via a human-machine interface 24. The human-machine interface 24 is communicatively connected to the MPC system 4. The human-machine interface 24 can do this as follows 1 shown is a computer (eg desktop PC, laptop, tablet or smartphone) that is separate from the MPC system 4, the motor vehicle 1 and the domestic charging unit 3 and has an input unit, for example a keyboard. The human-machine interface 24 can alternatively be integrated into the MPC system 4 or into the motor vehicle 1 or into the domestic charging unit 3, for example in the form of a keyboard, a touchscreen or a voice input unit.

In a first method step 100, the user 25 enters a target charging state SOC _target and a charging end time t _target via the human-machine interface 24. With the target state of charge SOC _target , the user 25 indicates how high the state of charge should be after the charging process has been completed. The state of charge (in English: State of Charge or SoC for short) can generally be the current energy content of the electric battery 7 in relation to its maximum energy content. The target state of charge SOC _target therefore describes an energy value with which the battery should be charged after the charging process has been completed. In an example chosen for illustration, the user 25 enters that the target state of charge SOC _target should be 95% of the maximum energy content of the battery 7. Assuming that the maximum energy content of the battery is 50 kWh, then the target state of charge SOC _target of 95% entered by the user 25 would correspond to an energy content of 47.5 kWh (= 0.95 * 50 kWh).

The charging end time t _target describes when this target state of charge SOC _target should be reached, in that the charging end time t _target determines a latest time at which the battery 7 should be charged up to the target state of charge SOC _target . The loading end time t _target can, for example, be specified in the form of a date and time. In an example chosen for illustration, the user 25 enters that the target state of charge SOC _target of the battery 7 should be reached the following morning, for example on January 19, 2022 at 6:00 a.m.

In a second method step 200, a current charge state SOC _current and a current time t _current are determined. For example, the current charge level SOC _current can result in a value of 40%. Assuming that the maximum energy content of the battery was again 50 kWh, then the determined current charge level of 40% would correspond to an energy content of 20 kWh (= 0.40*50 kWh). The current time t _current can again be specified, for example, in the form of a date and time. As an example, it can be determined that the current time is 8:00 p.m. on January 18, 2022. Determining the current state of charge SOC _current and the current time t _current can be done, for example, by the detection unit 13 of the motor vehicle 1. The second method step 200 can take place after the first method step 100 or at the beginning, in particular at the same time or in parallel with the first method step 100 (as shown by 2 shown as an example). The target state of charge SOC _target and the current state of charge SOC _current represent state of charge information of the battery 7. The end of charge time t _target and the current time represent time information.

In a third method step 300, the processor unit 10 executes the constraints algorithm 27. The target state of charge SOC _target and the current state of charge SOC _current serve as input values in order to determine a first state of charge secondary condition C _SOC1 as the output value. A difference is formed between the target state of charge SOC _target and the current state of charge SOC _current of the battery 7. This difference corresponds to an energy value which must at least be supplied to the battery 7 via the domestic charging unit 3 in order to increase the energy content of the battery 7 from its current value SOC _current to the required value SOC _target . In the example described above, the calculation is C _SOC1 = SOC _target - SOC _current = 47.5 kWh - 20 kWh = 27.5 kWh.

Furthermore, in a fourth method step 400, a second charge state secondary condition C _SOC2 is specified, according to which the current _charge state SOC _current may not exceed the maximum energy content of the battery SOC _max (=100%; in the selected exemplary embodiment this corresponds to 50 kWh). This second charge state secondary condition C _SOC2 can in particular be specified once and continuously recalculated or checked each time the MPC algorithm 15 is executed. Expressed in a formula, the second state of charge condition C means _SOC2 : SOC _current ≤ SOC _max = 100% = 50 kWh. In the example chosen, this additional condition is fulfilled (SOC _current = 20 kwH < 50 kWh or 40% < 100%).

Analogously to the loading time, the processor unit 10 also executes the secondary condition algorithm 27 in a fifth method step 500. The loading end time t _target and the current time t _current serve as input values in order to determine a loading time secondary condition C _duration as the output value. A difference is formed between the loading end time t _target and the current time t _current . This difference corresponds to a duration or a maximum tolerated period within which the battery 7 should be charged via the domestic charging unit 3 in such a way that the energy content of the battery 7 is increased from its current value SOC _current to the required value SOC _target . In the example described above, the calculation is C _duration = t _target - t _current = (January 19, 2022, 6:00 a.m.) - (January 18, 2022, 8:00 p.m.) = 10 h. In other words, the first charge state secondary condition C _SOC1 describes the energy content that must at least be supplied to the battery 7. This energy content or energy content corresponds to a net energy value E _charge,min . The loading time constraint C _duration describes the maximum time that may be used for this. The fifth method step 400 can in particular take place at the same time or in parallel with the third method step 300. Furthermore, by executing the constraint algorithm 27 once, both the first charge state constraint C _SOC1 and the first charging time constraint C _duration can be determined.

• C_SOC3: ${\displaystyle {\sum }_{i=1}^N{I}_{Load,i}}\ast U_{Load}\ast \frac{t_{target}-t_{current}}{N}\le E_{Charge,min}\ast \frac{100\text{\%-}SO{C}_{current}}{SO{C}_{target}-SO{C}_{current}};$
  
• C_SOC4: ${\displaystyle {\sum }_{i=1}^N{I}_{Load,i}}\ast U_{Load}\ast \frac{t_{target}-t_{current}}{N}\le E_{Charge,min}.$
  

From the additional conditions C _SOC1 , C _SOC2 and C _duration described above, it can be deduced that for a sum of N-1 temporal, equally long loading sections 28 of a prediction horizon ahead in time between N discretization points 29 (illustrated in 3 ) the following charge state additional conditions must still apply:

• C _SOC3 : ${\displaystyle {\sum }_{i=1}^N{I}_{LOad,i}}\ast U_{LOad}\ast \frac{t_{tarGet}-t_{current}}{N}\le E_{CHarGe,min}\ast \frac{100\text{\%-}SO{C}_{current}}{SO{C}_{tarGet}-SO{C}_{current}};$
  
• C _SOC4 : ${\displaystyle {\sum }_{i=1}^N{I}_{LOad,i}}\ast U_{LOad}\ast \frac{t_{tarGet}-t_{current}}{N}\le E_{CHarGe,min}.$
  

The third state of charge secondary condition C _SOC3 prevents (like the second state of charge secondary condition C _SOC2 ) a charge level of over 100% from being reached. The third state of charge secondary condition C _SOC3 can be specified in addition to or as an alternative to the second state of charge secondary condition C _SOC2 . The third state-of-charge condition C _SOC3 restricts the MPC algorithm 15 from “overcharging” the battery 7. The fourth state of charge secondary condition C _SOC4 describes that within the charging horizon 30 with the duration t _target -t _current at least the net energy value E _Charge,min described above is charged into the battery 7.

The user 25 of the vehicle 1 can specify a minimum SOC, in particular the target state of charge SOCt _arget , which he would like to find at the next departure (“desire regarding departure state”). This minimum expected SOC _target corresponds to the minimum energy to be supplied E _charge,min . The duration between the start of the charge and the next departure corresponds to t _target - t _current . The fourth state of charge secondary condition C _SOC4 therefore states that the MPC algorithm 15 must plan the charging process in such a way that at least the amount of energy E _charge,min is charged within t _target - t _current , so that the user's wish mentioned above is fulfilled.

The third state of charge constraint C _SOC3 and the fourth state of charge constraint C _SOC4 are taken into account each time the MPC algorithm 15 is executed. This also means that they each have to be recalculated by the constraint algorithm 27. It may be that the current SOC _current charge level was 50% in a previous plan, but is now 51% in a current plan. Accordingly, the formulas must be re-solved by the constraint algorithm 27 and the numerical results must be available as a constraint for the MPC algorithm 15.

After the charge state secondary conditions C _SOC1 , C _SOC2 , C _SOC3 and C _SOC4 as well as the charging time secondary condition C _duration have been determined, the processor unit 10 executes the MPC algorithm 15 in a sixth method step 600. The charge state constraints C _SOC1 , C _SOC2 , C _SOC3 and C _SOC4 and the charging time constraint C _duration are used as restrictions on the solution space of predicted charging processes. The MPC algorithm 15 follows a gradient-based optimization principle and plans a charging curve for the battery 7 in the present application. Physically, this means that the MPC algorithm 15 specifies a charging current I _Load for the battery 7. Due to the voltage U _Load available at home (usually 230V or 400V, see above), the product of the charging current I _Load and the charging voltage U _Load results in the charging power P _Load . The charging voltage U _Load is applied to the charging cable 19, so that the charging current I _Load flows through the charging cable 19.

By the processor unit 10 executing the MPC algorithm 15, the charging currents I _Load are determined or predicted. Using the predicted charging currents I _Load , the battery 7 is to be charged at N discrete times within a prediction horizon 30 in advance. The prediction horizon 30 is made up of the individual loading sections 28. Each time the MPC algorithm 15 is called, the prediction horizon 30 extends in time between the loading end time t _target and the current time t _current . As time progresses, the prediction horizon 30 decreases steadily, which means that the frequency of the discretization points 29 also increases and the time length of the loading sections 28 decreases.

The charging currents I _Load are determined or predicted depending on the secondary conditions C _SOC1 , C _SOC3 and C _SOC4 and C _duration determined by executing the constraint algorithm 14, the predetermined second state of charge condition C _SOC2 , and while minimizing the cost function 17 of the MPC algorithm 15. The planned charging curve corresponds to the charging power at the N different discretization points 29 above the prediction horizon 30, which in the present example is a charging horizon. The charging horizon 30 is defined in the selected example as the period of charging, ie it corresponds to the charging time secondary condition C _duration = t _target - t _current . The discretization of the charging horizon 30 takes place in N-1 equally distributed time periods 28. If N remains the same, these time periods 28 become smaller as the charging duration continues and the optimization result becomes correspondingly more precise.

In order to be able to optimize the charging curve, the MPC algorithm 15 contains an optimization goal in the form of the cost function 17, which is a continuously differentiable minimization function. The basic goal here is to minimize costs. Since in this case the costs are incurred the electricity costs of the household and the charging process, the following applies to the cost function 17 or f _min to be minimized: $$f_{min}={\displaystyle {\sum }_{i=1}^N{I}_{LOad,i}}\ast U_{LOad}\ast \frac{t_{tarGet}-t_{current}}{N}+I_{HOuse,demand,i}\ast U_{LOad}-I_{supply,i}\ast U_{LOad}.$$

The size I _{house,demand,i} describes the average electricity requirement of house 2, which occurs at time i for the duration of a subsequent discretization section. The quantity I _supply,i describes the average current that can be supplied by an above-mentioned electricity supplier in the same period of time. Both of these variables are the result of a prediction model that extends over the charging horizon t _target - t _current .

Ultimately, the following constraint or additional condition applies to the three current levels: I _Load,i + I _{house,demand,i} - I _supply ≤ x. The value x describes a current limit value, which represents the maximum technically adjustable current strength of a power connection in the house 2. This current intensity secondary condition can be taken into account as a further secondary condition when determining the charging curve.

With the help of the cost function 17 described above, the MPC solver can now test any possible charging curves with regard to their cost optimality. The optimization space is mathematically limited by the constraints or additional conditions described, which ensures that the optimal charging curve found is both physically correct and logically implementable. The method steps described above (without entering the data using the human-machine interface 24) are repeated continuously and cyclically.

Reference symbols

CSOC1

first state of charge secondary condition

CSOC2

second state of charge secondary condition

Cduration

Loading time constraint

ILoad

charging current

SOCcurrent

current charging status

SOCtarget

Target state of charge

SOC max

maximum charge level

tcurrent

current time

ttarget

Loading end time

ULoad

charging voltage

1

vehicle

2

House

3

domestic charging unit

4

System for model-based predictive control

5

Drivetrain

6

electric machine

7

battery

8th

transmission

9

Internal combustion engine

10

Processor unit

11

Storage unit

12

Communication interface

13

Acquisition unit

14

Computer program product

15

MPC algorithm

16

charging model

17

Cost function

18

Control unit

19

electric charging cable

20

electrical charging port

21

Power input connection

22

power line

23

Electricity generating plant

24

Human-machine interface

25

User

27

Constraint Algorithm

28

loading sections

29

Discretization point

30

Prediction horizon

100

first step of the process

200

second procedural step

300

third step of the process

400

fourth step of the process

500

fifth procedural step

600

sixth procedural step

Claims (13)

Method for model-based predictive control of a charging process of a battery (7) for a drive train (5) of a motor vehicle (1) using a domestic charging unit (3), the method comprising the steps: - Determining secondary conditions (C _SOC1 , C _duration ) by executing a constraints algorithm (27), whereby the constraints (C _SOC1 , C _duration ) are determined depending on the state of charge information (SOC _target , SOC _current ) of the battery (7) and time information (t _target , t _current ), and - determining of charging currents (I _Load,i ), by means of which the battery (7) is to be charged by the domestic charging device (3) within an upcoming prediction horizon (30) at N discrete times (29), by executing an MPC algorithm (15 ), whereby the determination of the charging currents (I _Load,i ) depending on the secondary conditions (C _SOC1 , C _duration ) determined by executing the constraints algorithm (27) and while minimizing a cost function (17) of the MPC algorithm ( 15) takes place. Procedure according to Claim 1 , where the prediction horizon (30) is defined as a temporal charging horizon, which represents the period between a charging end time (t _target ) and a current time (t _current ). Procedure according to Claim 2 , whereby the charging horizon (30) is divided into N-1 equally distributed time periods (28). Procedure according to Claim 3 , whereby - the cost function (17) to be minimized contains a first positive term, a second positive term and a third negative term, whereby - the first term represents the charging current (I _Load,i ), by means of which the battery (7) is powered by the domestic charging device (3) is to be charged within one of the time periods (28), - the second term represents an average domestic electricity requirement within the time period (28) occurs in a house (2) to which the domestic charging unit (3) belongs, and - the third term represents an average output current that is delivered to the house (3) by an electricity supplier. Method according to one of the preceding claims, wherein the state of charge information - describes a target state of charge (SOCt _arget ), up to which the battery (7) is to be charged, and - describes a current state of charge (SOC _current ), which describes a current state of charge of the battery (7 ) represents. Procedure according to Claim 5 , wherein by executing the constraint algorithm (27), a first state of charge constraint (C _SOC1 ) is determined, which represents the difference between the target state of charge (SOCt _arget ) and the current state of charge (SOC _current ) of the battery (7). Procedure according to Claim 6 , whereby a second state of charge secondary condition (C _SCO2 ) is additionally specified, according to which the current _state of charge (SOC _current ) must not exceed a maximum possible state of charge (SOC _max ) of the battery (7). Method according to one of the preceding claims, wherein the time information - describes a charging end time (t _target ) at which the battery (7) should be charged up to the target state of charge (SOCt _arget ), and - describes a current time (t _current ). Procedure according to Claim 8 , whereby by executing the constraint algorithm (27), a loading time constraint (C _duration ) is determined which represents the period (30) between the loading end time (t _target ) and the current time (t _current ). Procedure according to Claim 9 , whereby the target charging state (SOCtarget) and the charging end time (t _target ) are specified by means of a human-machine interface (24). Procedure according to Claim 10 , whereby the current state of charge (SOC _current ) and the current time (t _current ) are determined by means of a detection unit (13) of the motor vehicle (1). Processor unit (10) for model-based predictive control of a charging process of a battery for a drive train (5) of a motor vehicle (1) by means of a domestic charging unit (3), the processor unit (10) being set up to - additional conditions (C _SOC1 , C _duration ) by executing a constraint algorithm (27), the constraints (C _SOC1 , C _duration ) being determined depending on the state of charge information (SOC _target , SOC _current ) of the battery (7) and time information (t _target , t _current ). , and - by executing an MPC algorithm (15) to determine charging currents (I _Load,i ), by means of which the battery (7) is charged by the domestic charging device (3) within a preceding prediction horizon (30) at N discrete times (29 ) is to be charged, with the determination of the charging currents (I _Load,i ) depending on the secondary conditions (C _SOC1 , C _duration ) determined by executing the constraints algorithm (27) and while minimizing a cost function (17) of the MPC Algorithm' (15) takes place. Processor unit (10). Claim 12 , wherein the processor unit (10) is integrated into a charging unit (3) of a house (2) or into a motor vehicle (1).

Images