Method and controller module for operating a DC-DC converter

Abstract

In the method, an output current ripple suppresses the output current of the switchable DC-DC converter (DCDC) using a compensation model, the parameters (A1, A2,...) of which are adapted or learned during the runtime of the DC-DC converter (DCDC).

Description

[0001] 202500327

[0002] 1

[0003] Description

[0004] Method and control module for operating a DC-DC converter

[0005] The disclosure relates to a method and a control module for operating a DC-DC converter. Furthermore, the disclosure relates to a DC-DC converter arrangement, a computer program, and a computer-readable medium.

[0006] The AC / DC converter in an electric vehicle (on-board charger) typically consists of two subsystems: a power factor correction (PFC) stage and a high-voltage DC-DC converter (HV-DC-DC). The two subsystems are connected via a DC link capacitor. The primary function of the PFC is to draw alternating current (AC) from the charger's input side (e.g., the mains connection). The primary function of the HV-DC-DC converter is to supply direct current (DC) to the charger's output side (e.g., the high-voltage battery). On average, the AC and DC power outputs must be equal (efficiency is ignored here), but the instantaneous power outputs cannot be equal. This means that whenever the instantaneous AC power is greater than the instantaneous DC power output, the excess is stored in the DC link capacitors. In this case, the DC link voltage increases.Whenever the instantaneous AC power is less than the instantaneous DC power, the remaining power or charge is drawn from DC link capacitors. In this case, the DC link voltage drops. This means that the intermediate circuit voltage is never a pure DC voltage, but always has an AC component above its average value.

[0007] Many HV-DC-DC converters are sensitive to their output-to-input voltage ratio. One example of such a converter is a resonant converter, particularly an LLC DC-DC converter. This type of converter is typically controlled by frequency or phase-shift modulation of the H-bridge PWM pattern. When the voltage ratio changes (which occurs when the DC link voltage oscillates), the controller must adjust the control signal to maintain a constant DC output current. Simply relying on the closed-loop controller is often insufficient for several reasons: The controller must "see" a deviation before it can react, so suppressing the entire current ripple is inherently impossible with this approach. The controller must meet stability criteria to ensure robust control in the event of unexpected disturbances. This typically leads to...

[0008] 2

[0009] A rather "slow" controller is unable to adequately suppress the current ripple. Speeding it up can destabilize the system under different operating conditions. Another way to suppress ripple is to implement a feedforward function that provides a control signal offset depending on the input and output voltages (or their ratio). This approach has two problems: Most HV-DC-DC converters have nonlinear and operating-point-dependent system behavior, so a predefined function must have a high degree of granularity, leading to significant testing and calibration effort and increased computational load. The second problem is that a predefined function can only reflect the nominal system behavior. Tolerances, aging, thermal effects, etc., are generally not modelable.Therefore, this function will not completely suppress current ripple in almost all samples.

[0010] One task to be solved is therefore to provide a method and a control module for operating a DC-DC converter that enables a reliable provision of a constant output current from the DC-DC converter.

[0011] The problem is solved by the features of independent patent claims.

[0012] Advantageous embodiments are characterized in the dependent claims.

[0013] According to a first aspect, the problem is solved by a method for operating a switchable DC-DC converter using a controller module. The DC-DC converter is, in particular, a DC-DC converter of an onboard charging device of a vehicle. The DC-DC converter has a modulation unit, or a modulation unit is associated with the DC-DC converter. The controller module comprises a current control module, a feedforward control module, and a computation module. The controller module is preferably implemented in software. Alternatively, the controller module can be implemented in hardware or in both software and hardware.

[0014] The process includes the following steps:

[0015] • Reading, by means of the feedforward control, a provided first voltage measurement signal or provided first voltage measurement data that are representative of an input voltage at the input of the DC-DC converter, and / or reading a provided second 202500327

[0016] 3

[0017] voltage measurement signal or from provided second voltage measurement data that are representative of an output voltage at the output of the DC-DC converter;

[0018] • Determine, using a feedforward control, a set of characteristics of the input voltage or the output voltage;

[0019] • Reading in, by means of current control, a provided initial current measurement signal or provided initial current measurement data that are representative of a current to be controlled of the DC-DC converter, in particular the output current of the DC-DC converter;

[0020] • Determine, using current control, a first control deviation depending on a specified target current and the first current measurement signal or the first current measurement data;

[0021] • Determine, using feedforward control, at least one first parameter value (A1) depending on the first control deviation and the set of characteristics;

[0022] • Determining, using feedforward control, an offset control variable depending on the set of parameters and the set of characteristics;

[0023] • Determine, using the calculation module, a resulting manipulated variable depending on a controller manipulated variable that provides the current control and the offset manipulated variable, and provide the resulting manipulated variable for the modulation unit of the DC-DC converter to control the modulation unit of the DC-DC converter.

[0024] The input voltage at the input of the DC-DC converter is, for example, a DC link voltage. This DC link voltage can be present, in particular, at a DC link capacitor at the output of an upstream rectifier or power factor correction stage. The DC link voltage is decomposed into several periodic basis functions. A periodic basis function can also be described as a characteristic of the input voltage. The set of characteristics thus comprises the determined periodic basis functions.

[0025] Preferably, the basic functions are each multiplied by a respective model parameter to generate a modeled uncompensated current ripple value. The current ripple value can also be referred to as the AC model value. The model parameters are also referred to as parameter values ​​in the application. 202500327

[0026] 4

[0027] The modeled current ripple values ​​are converted into a control signal offset, for example, using a sensitivity function, also known as a scaling function. The model parameters are adjusted during operation of the DC-DC converter, in particular continuously. Stochastic gradient descent is used for this purpose, based on a cost function that represents a current error (control deviation).

[0028] The set of parameters includes one or more parameters, and the set of characteristics includes one or more characteristics.

[0029] The main benefits of revelation are:

[0030] • Hardware tolerances, aging and thermal effects are largely compensated;

[0031] • Higher component tolerances are acceptable = cost savings;

[0032] • The method makes it possible to minimize the ripple current for all operating points;

[0033] • Complex feedforward models are avoided = Shorter development, calibration and testing time = Cost savings;

[0034] • The learned parameters reflect physical properties of the system (e.g., current-voltage sensitivity). This can be used for diagnostics;

[0035] • The method makes it possible to almost completely suppress or even completely suppress the ripple of the output current;

[0036] • The method enables continuous learning = special states are not required and it avoids the risk of abrupt changes to the system;

[0037] • The computational effort required for parameter updates is very low = A "cheap" PC is sufficient = Cost savings;

[0038] • Parameter updates are easily adjustable via a small set of parameters = learning rate, regularization rate, parameter limits, etc.;

[0039] • The algorithm is applicable to a wide variety of activation schemes (e.g.

[0040] Frequency or phase shift modulation / half-bridge or full-bridge control).

[0041] The method can be used in particular for all DC power stages where the causal relationships between the oscillation patterns of the input voltage and / or the output voltage and the corresponding effect on the ripple of the output current are known (or suspected).

[0042] 5

[0043] In at least one advantageous embodiment according to the first aspect, the set of characteristics comprises at least a first alternating component and a second alternating component of the input voltage or the output voltage.

[0044] Preferably, the set of characteristics comprises a first alternating component comprising a sine oscillation with a fundamental frequency and a second alternating component comprising a corresponding cosine oscillation with the fundamental frequency.

[0045] In an optional embodiment, the set of characteristics includes a third alternating component comprising a sine wave with a second harmonic frequency relative to the fundamental frequency, and a fourth alternating component comprising a corresponding cosine wave with the second harmonic frequency.

[0046] Alternatively or additionally, a fifth alternating component, comprising a sine wave with a third harmonic frequency relative to the fundamental frequency, and a sixth alternating component, comprising a cosine wave with the third harmonic frequency, can be taken into account.

[0047] In at least one advantageous embodiment according to the first aspect, determining the set of characteristics includes:

[0048] • Determining an AC component signal of the input voltage or the output voltage and providing the AC component signal as the first AC component;

[0049] • Delaying the AC component signal by a predetermined delay duration and providing the delayed AC component signal as the second AC component.

[0050] The AC component signal is determined based on the currently measured input or output voltage of the DC-DC converter and on an average input or output voltage. Specifically, the AC component signal is equal to the currently measured input or output voltage minus the average input or output voltage.

[0051] The delay duration depends in particular on the mains frequency of an AC input voltage of the onboard charging device. The delay duration is preferably one quarter of the period relative to the mains frequency.

[0052] 6

[0053] In at least one advantageous embodiment according to the first aspect, determining the set of characteristics includes:

[0054] • Determining a phase angle of the input voltage or the output voltage using a phase-locked loop module and providing the first AC component and a second AC component using the phase-locked loop module depending on the determined phase angle.

[0055] The PLL module uses, in particular, an oscillator that has a frequency equal to the mains frequency of an alternating voltage at the input of the onboard charging device.

[0056] In at least one advantageous embodiment according to the first aspect, determining the set of characteristics includes:

[0057] • Determining a Fourier amplitude spectrum and a Fourier phase spectrum using a Fourier transform of the input voltage and the output voltage, respectively;

[0058] • Determining at least a first alternating component and a second alternating component depending on amplitude values ​​and phase values ​​of the determined spectra for selected frequencies;

[0059] • Providing at least one first exchange share and one second exchange share.

[0060] In at least one advantageous embodiment according to the first aspect, the parameter(s) of the set of parameters are determined depending on a cost function to be minimized, which, for example, represents a mean squared error, where each error is equal to the first control deviation determined in each case. Preferably, one error is determined per sampling cycle. This error is summed for n sampling steps.

[0061] The method enables the reduction of output current ripple. The goal is to learn the system characteristics during system operation (during AC charging). To achieve this, a machine learning approach called Stochastic Gradient Descent (SGD) is used. SGD aims to minimize an objective function, often a loss function, by iteratively updating model parameters. In the presented application, the goal is to minimize the mean squared measurement signal error, which represents the residual error of the controlled system. Or, in simpler terms: the goal is to minimize the

[0062] 7

[0063] to learn about operating point-specific and sample-specific system sensitivity and to use this knowledge to suppress current ripple.

[0064] In at least one advantageous embodiment according to the first aspect, the parameter(s) of the parameter set are each determined based on a stochastic gradient descent algorithm, depending on a learning rate. In particular, the stochastic gradient descent algorithm considers only one sample at a time, i.e., n = 1.

[0065] In at least one advantageous embodiment according to the first aspect, the parameter(s) of the parameter set are each determined depending on a regularization rate, wherein the regularization rate incorporates an existing solution into the solution determination. In particular, the regularization rate enables the inclusion of an existing solution and its accuracy in the solution determination.

[0066] In at least one advantageous embodiment according to the first aspect, the set of parameters includes at least a first parameter and a second parameter, and determining an offset control variable for current control includes:

[0067] • Determining a ripple current that depends on a sum comprising at least a first product and a second product, where the first product is equal to the first parameter multiplied by the first AC component and the second product is equal to the second parameter multiplied by the second AC component;

[0068] • Determining the offset control variable depending on the determined ripple current and a scaling factor.

[0069] The updated parameters, which can also be called model parameters, are combined with the extracted features, the determined characteristics, to perform the model's interference (model evaluation). The output is a modeled value of the uncompensated current ripple, which is added to the main control signal coming from the controller.

[0070] The modeled value of the uncompensated current ripple is translated into the offset control variable via a predefined sensitivity function, where the modulated value, in its simplest form, is just a gain value. In a more advanced version, this gain can be a function of the operating point (average input voltage, average output voltage, average current).

[0071] 8

[0072] Finally, the resulting control variable is used to actuate the hardware. The resulting signals are measured and fed back into the algorithm.

[0073] According to a second aspect, the problem is solved by a control module for operating a switchable DC-DC converter. The DC-DC converter is, in particular, a DC-DC converter of an onboard charging device of a vehicle. The DC-DC converter has a modulation unit, or a modulation unit is associated with the DC-DC converter. The control module comprises a current control module, a feedforward control module, and a computation module. The control module is configured to execute the method according to the first aspect or advantageous embodiments of the method according to the first aspect. The control module is preferably implemented in software. Alternatively, the control module can be implemented in hardware or in both software and hardware.

[0074] In particular, the controller module can be part of a control device associated with the DC-DC converter. The control device can, in particular, include a processor unit with a processor and at least one program memory, wherein the processor executes the process steps according to the first aspect.

[0075] According to a third aspect, the task is solved by a converter arrangement comprising a controller module according to the second aspect and a switchable DC-DC converter, wherein the DC-DC converter has a modulation unit or a modulation unit is associated with the DC-DC converter.

[0076] According to a fourth aspect, the above-mentioned task is solved by a computer program comprising instructions which, when the program is executed by a control computer, cause the control computer to execute the method for operating a switchable DC-DC converter or an optional embodiment of the method.

[0077] The control computer has a processor and program memory. Alternatively, the program memory can be assigned to the control computer. The processor can have a central processing unit (CPU). The processor can be a general-purpose processor, a microprocessor, a microcontroller, or a 202500327

[0078] 9

[0079] It could be a digital signal processor (DSP) or an application-specific integrated circuit (ASIC).

[0080] According to a fifth aspect, the above-mentioned task is solved by a computer-readable medium containing instructions which, when executed by a control computer, cause the control computer to execute the method for operating a switchable DC-DC converter or an optional embodiment of the method.

[0081] The computer program can be stored on a computer-readable storage medium, in particular non-volatile memory. The storage medium can be memory built into the processor, memory located outside the processor on a circuit board, or portable storage. The memory is configured to store associated program instructions and related data.

[0082] Furthermore, the computer program can be made available on a network such as the Internet, from which it can be downloaded by a user when needed.

[0083] According to a sixth aspect, the problem is solved by a method for operating a switchable DC-DC converter using a controller module that has current control and feedforward control, wherein

[0084] • The feedforward control is based on a compensation model that has as input variables an input voltage and / or an output voltage and a first control deviation of the current control, determines a ripple current that represents an AC component of the output current, and determines an offset control variable of the feedforward control depending on the ripple current; • and parameters of the compensation model are adapted or learned during the operating time of the DC-DC converter depending on the input voltage or the output voltage and depending on the first control deviation.

[0085] The method makes it possible to suppress output current ripple of the DC-DC converter using a compensation model whose parameters are adjusted or learned during the system's operation.

[0086] 10

[0087] Further advantageous embodiments are disclosed in the attached claims and in the following description of exemplary embodiments with reference to the attached figures.

[0088] The description of the items listed here is not limited to the individual specific embodiments. Features of different embodiments can be combined – insofar as technically feasible – to form further embodiments. For example, variations or modifications described with regard to one embodiment may also be applicable to other embodiments, unless otherwise stated.

[0089] They show:

[0090] Figure 1 shows an exemplary block diagram of a transducer arrangement.

[0091] Figure 2 shows an exemplary detailed block diagram of a feedforward control system.

[0092] Figures 3a and 3b show a minimization of a cost function using a gradient descent method.

[0093] Figure 4 shows a derivation of a formula for determining parameters of a compensation model.

[0094] Figure 5 Equations for calculating a ripple current and

[0095] Figure 6 shows different signal patterns.

[0096] In the figures, the same reference symbols are used for elements with essentially the same function; however, these elements do not have to be identical in every detail.

[0097] Figure 1 shows an exemplary block diagram of a transducer arrangement.

[0098] The converter arrangement comprises a controller module 10 and a switchable DC-DC converter. The DC-DC converter includes a 202500327

[0099] 11

[0100] Modulation unit 40. Alternatively, modulation unit 40 can be assigned to the DC-DC converter.

[0101] The control module 10 includes a current control 20, a feedforward control 30 and a calculation module 50.

[0102] The current control 20, for example, has a comparator 22 and a controller 24.

[0103] The comparator 22 is configured to determine a first control deviation lerr, which can also be referred to as current error, depending on a predefined target current Iset and a current measurement signal or current measurement data that represents a current lout of the DC-DC converter. The comparator 22 preferably determines a difference. The current lout can, in particular, be an output current of the DC-DC converter.

[0104] The current error is used for two purposes. Firstly, it serves as an input for the closed-loop controller 24; secondly, it is used in the algorithm for updating the model parameters.

[0105] Controller 24 is configured to determine a control variable SC based on the control deviation. Controller 24 includes, for example, a proportional-integral element (Pl element). The control variable SC can also be referred to as a control signal.

[0106] The feedforward control unit 30 is configured, based on a compensation model that has as input variables an input voltage llin and / or an output voltage as well as the first control deviation lerr of the current control 20, to determine a ripple current l^ppie, which represents an AC component of the output current, and to determine an offset control variable Soffset of the feedforward control unit 30 depending on the ripple current. Furthermore, the feedforward control unit 30 is configured to adapt or learn parameters A1, A2, ... of the compensation model during the operating time of the DC-DC converter, depending on the input voltage or the output voltage and depending on the first control deviation.

[0107] Figure 2 shows another exemplary block diagram of the current control 20 and the feedforward control 30.202500327

[0108] 12

[0109] The feedforward control 30 includes, for example, an extraction module 32. The extraction module 32 is configured, in particular, to determine a set X of characteristics of the input voltage 11, depending on a provided first voltage measurement signal or on provided first voltage measurement data that are representative of an input voltage 11 at the input of the DC-DC converter. The set X of characteristics includes, for example, a first characteristic and a second characteristic. The first characteristic is, for example, a first AC component Xi of the input voltage 11, and the second characteristic is, for example, a second AC component X2 of the input voltage U1.

[0110] The first voltage measurement signal or data is provided, for example, by a voltage sensor of the DC-DC converter. The input voltage is measured, for example, at a DC-link capacitor of an upstream power factor correction device.

[0111] The extraction module 32 is specifically designed to determine an alternating component of the input voltage and to provide the determined alternating component as the first alternating component Xi.

[0112] The AC component is, for example, equal to the difference between the measured input voltage Uin of the DC-DC converter and the averaged input voltage Ü^.

[0113] The AC signal can therefore be determined according to the following equation:

[0114]

[0115] From the AC component signal, at least a first AC component and a second AC component are determined. In an optional configuration, further AC components can be determined. The first AC component represents, for example, a fundamental sine wave and the second AC component represents a corresponding fundamental cosine wave.

[0116] For example, the AC component of the input voltage is determined and provided as the first AC component. The first AC component is delayed by a predetermined delay period, in particular by a quarter.

[0117] 13

[0118] One period of the mains frequency of the charging device's input voltage. The delayed AC component is provided as the second AC component.

[0119] The following applies as a standard to the first exchange share:

[0120]

[0121] and for the second exchange share:

[0122] X2 = cos(F Inp )

[0123] Standardization is optional.

[0124] Preferably, only the fundamental frequency is considered, so that the set X of characteristics includes only a first characteristic X1 and a second characteristic X2, and the set A of parameters includes only a first parameter A1 and a second parameter A2.

[0125] The feedforward control 30 includes in particular an adaptation module 34 (Figure 2). The adaptation module 34 is configured to determine the parameters A1, A2 of the set A of parameters, depending on the first control deviation lerr and the set M of characteristics.

[0126] In particular, the adaptation module 34 is designed to determine a first parameter A1 and a second parameter A2 depending on the first control deviation lerr and the first alternating component X1 and the second alternating component X2.

[0127] The first parameter A1 can be determined or updated according to the following formula, for example:

[0128] Ai = ßA + a(jM eas — IDC) * X

[0129] The second parameter A2 can be determined or updated according to the following formula, for example:

[0130] 2— ßA2+ a(J Meas [ DC ) * X2202500327

[0131] 14

[0132] Here, a is a learning rate and β is a regularization rate. iMess is the measured current lout of the DC-DC converter, specifically the measured output current of the DC-DC converter and l D c is the specified target current Iset.

[0133] The parameters are determined using a stochastic gradient descent method. This method comprises the following steps:

[0134] 1. First, a cost function that is to be minimized is defined.

[0135] 2. The cost function is a function of the set A of parameters (Ai , A2, ...). 3. A starting value is defined for each parameter Ai , A2.

[0136] 4. The gradient of the cost function along the set of parameters is determined -> partial derivative with respect to each element of the set of parameters.

[0137] 5. A gradient descent is performed in the direction of the derivative.

[0138] Steps 4 and 5 are repeated until the minimum is found.

[0139] Figure 3a illustrates the finding of the minimum using a gradient descent method with one parameter, here the first parameter A1.

[0140] Figure 3b illustrates the finding of the minimum using a gradient descent method with two parameters, here the first parameter A1 and the second parameter A2.

[0141] Figure 4 illustrates the derivation of equations (1) and (2).

[0142] First, the cost function J is defined. For example, the mean squared error (MSA) is chosen as the cost function J. The error l A c is equal to the non-uncompensated AC component of the output current lAc.Nocompensation of the DC-DC converter minus the modeled AC component of the output current IAC.MCII of the DC-DC converter.

[0143] The modeled AC component of the output current IAC.MCII can also be referred to as the ripple current lRi. PP i.e. become. Assuming that there is a causal relationship between the oscillation patterns of the input voltage and / or the output voltage and the corresponding effect on the ripple of the output current, the AC component of the output current, here ripple current202500327

[0144] 15

[0145] The values ​​shown in Figure 5 are calculated. This causal relationship is described using a compensation model.

[0146] The cost function is differentiated with respect to each parameter (Figure 4). Figure 4 shows this differentiation as an example for the first parameter Ai.

[0147] The error l can be used to calculate the parameters. AC On the other hand, it can be determined by measurement. The error l AC results from the difference between the measured current iMeas, lout and the specified target current Iset, l D c.

[0148] For the determination or updating of the respective parameter, here as an example the first parameter A1, only one sample is used based on the stochastic gradient descent method, i.e. the current error that is current for the evaluation period.

[0149] For the calculation, the gradient of the cost function is weighted with a learning factor a. Furthermore, the previously determined value for the first parameter Ai is weighted with a regularization rate β.

[0150] The feedforward control 30 further comprises an evaluation module 36. The evaluation module 36 is configured, depending on the updated parameters Ai, A2 and the set of characteristics, in particular the first AC component Xi and the second AC component X2, and the ripple current lRi. PP to determine with

[0151] ^Ripple = A1* X1+ A2* X2

[0152] The feedforward control 30 further comprises a conversion module 38. The conversion module 38 is configured in particular according to the ripple current lRi determined with the updated parameters Ai, A2. PPie, to determine an offset control variable Soff. For example, the offset control variable Soff can be determined based on a scaling function. In particular, the offset control variables Soff can be stored as assigned to predefined ripple current values, for example in a conversion table.

[0153] The offset control variable Soff, like the controlled control variable SC, is fed to the calculation module 50. The calculation module 50 determines the resulting control variable SR for the modulation unit 40, which preferably comprises a pulse width modulator.

[0154] 16

[0155] Figure 6 (below) shows a simulation of the input voltage of the DC-DC converter. In the example shown, the input signal comprises a superposition of a sinusoidal voltage and a second harmonic of the sinusoidal voltage.

[0156] Figure 6 (center) shows a simulation of the output current of the DC-DC converter. It shows that after a certain time, when the parameters A1, A2, ... of the set of parameters have been determined with sufficient accuracy, the current exhibits virtually no AC components.

[0157] Figure 6 (above) shows a comparison of the measured AC component of the current of the DCDC converter and the modeled ripple current.

Claims

202500327 17 Patent claims Patent claims 1. Method for operating a switchable DC-DC converter using a controller module (10), wherein the DC-DC converter has a modulation unit (40) or a modulation unit (40) is associated with the DC-DC converter, the controller module (10) has a current control (20), a feedforward control (30) and a computation module (50) and the method comprises the following steps: Reading, by means of the feedforward control (30), a provided first voltage measurement signal or provided first voltage measurement data that are representative of an input voltage (llin) at the input of the DC-DC converter, and / or reading a provided second voltage measurement signal or provided second voltage measurement data that are representative of an output voltage at the output of the DC-DC converter, Determine, by means of the feedforward control (30), a set (X) of characteristics of the input voltage (llin) or the output voltage, read in, by means of the current control, a provided first current measurement signal or provided first current measurement data that are representative of a current (lout) to be controlled of the DC-DC converter (DCDC), Determine, using current control (20), a first control deviation (lerr) depending on a specified setpoint current (Iset) and the first current measurement signal or the first current measurement data; determine and / or update, using feedforward control (30), a set (A) of parameters depending on the first control deviation (lerr) and the set (X) of characteristics; Determine, using the feedforward control (30), an offset control variable (Soff) depending on the set of parameters (A) and the set (X) of characteristics; determine, using the calculation module (50), a resulting control variable (SR) depending on a controller control variable (SC) provided by the current control (20) and the offset control variable (Soff); and provide the resulting control variable (SR) to the modulation unit (40) of the DC-DC converter for control of the DC-DC converter by the modulation unit (40). 18 2. The method according to claim 1, wherein the set (X) of characteristics comprises at least a first alternating component (Xi) and a second alternating component (X2) of the input voltage (llin) or the output voltage.

3. The method of claim 2, wherein determining the quantity (X) of characteristics comprises: - Determining an AC component signal of the input voltage (llin) or the output voltage and providing the AC component signal as the first AC component (Xi), - Delaying the AC component signal by a predetermined delay time and providing the delayed AC component signal as the second AC component (X2).

4. The method of claim 2, wherein determining the quantity (X) of characteristics comprises: - Determining a phase angle of the input voltage or the output voltage using a phase-looked loop module and providing the first AC component (Xi) and a second AC component (X2) using the phase-looked loop module depending on the determined phase angle.

5. The method of claim 2, wherein determining the quantity (X) of characteristics comprises: - Determining a Fourier amplitude spectrum and a Fourier phase spectrum using a Fourier transform of the input voltage (Uin) or the output voltage, - Determining at least the first alternating component (Xi) and the second alternating component (X2) depending on amplitude values ​​and phase values ​​of the determined spectra for selected frequencies.

6. Method according to one of the preceding claims, wherein the parameter(s) (A1, A2, ...) of the set (A) of parameters is determined depending on a cost function (J) to be minimized, which in particular represents a mean square error, wherein each error is equal to the first control deviation (lerr) determined in each case.

7. A method according to any of the preceding claims, wherein the parameters (A1, A2, ...) of the set (A) of parameters are each determined based on a stochastic gradient descent algorithm depending on a learning rate (a). 19 8. Method according to claim 6 or 7, wherein the parameter(s) are determined depending on a regularization rate (β), wherein the regularization rate (β) incorporates an existing solution into the solution determination.

9. Method according to one of the preceding claims, wherein the set (A) of parameters comprises at least a first parameter (Ai) and a second parameter (A2) and includes determining an offset control variable (Soff) for current control (20): - Determining a ripple current (lRi) PPie) depending on a sum of at least one first product and a second product, where the first product is equal to the first parameter (Ai) multiplied by the first AC component (Xi) and the second product is equal to the second parameter (A2) multiplied by the second AC component (X2), - Determining the offset control variable (Soff) depending on the determined ripple current (lRi) PPie ) and a scaling factor 10. Method for operating a switchable DC-DC converter using a control module (10) comprising current control (20) and feedforward control (30), wherein the method comprises the following steps: - Determine, by means of the feedforward control (30), a ripple current representing an AC component of the output current, based on a compensation model which has as an input an input voltage and / or an output voltage and a first control deviation of the current control (20), - Determining an offset control variable of the feedforward control (30) depending on the ripple current and - Adapting or learning parameters (Ai , A2, ...) of the compensation model during the runtime of the DC-DC converter depending on the input voltage or the output voltage and depending on the first control deviation.

11. Control module (10) for operating a switchable DC-DC converter, wherein the DC-DC converter has a modulation unit (40) or a modulation unit (40) is associated with the DC-DC converter, the control module (10) has a current control (20), a feedforward control (30) and a computation module (50) and is configured to carry out the method according to any one of claims 1 to 10. 20 12. Transducer arrangement (1) comprising - a control module (10) according to claim 11 and - a switchable DC-DC converter, wherein the DC-DC converter has a modulation unit (40) or a modulation unit (40) is associated with the DC-DC converter, 13. Computer program comprising instructions which, when the program is executed by a control computer, cause the latter to execute the method according to any one of claims 1 to 10.

14. Computer-readable medium comprising instructions which, when executed by a control computer, cause the control computer to execute the method according to any one of claims 1 to 10.

Images