Control of a separately excited electric machine at a reduced calculation speed in the model-based flux calculation

Abstract

A method for controlling the phase voltage (Uphase) of a multi-phase electric machine provides for receiving a target torque (TqReq). Said target torque is mapped to a reference current (Id,q Ref) by means of a predefined mapping (TA). The phase voltage (Uphase) is controlled according to a closed control loop (RK), the reference variable of which is the reference current (Id,q Ref). A pilot controller (VS) generates, on the basis of the reference current (Id,q_Ref), a pilot control voltage (Ud,q Pre) which is fed into the control loop (RK). The pilot controller (VS) has a flux calculation (FB) which maps the reference current (Id,q Ref) to an associated magnetic reference flux (ΨRef) by means of a machine model. The pilot controller (VS) has a voltage calculation (SB) which maps the reference flux (ΨRef) to the pilot control voltage (Ud,q Pre). The flux calculation (FB) is carried out at a first repetition rate which is lower than the repetition rate of the voltage calculation (SB).

Description

[0001] Description

[0002] Control of an externally excited electric machine with reduced computation speed in model-based flux calculation

[0003] Electric drives, such as those used in electric vehicles, consist of an electric machine with a stator and a rotor. Besides permanent magnet machines with a permanent magnet rotor, which require rare earth elements due to their high installation density, there are also separately excited electric machines whose rotor has a winding and serves as a rotating (electro)magnet.

[0004] Furthermore, field-oriented control systems for electrical machines are known. These utilize a mapping of the phase current onto the stator using a (magnetic) model of the electrical machine.

[0005] In the control of a separately excited electric machine (synchronous machine), the stator and rotor currents are controlled independently using corresponding control parameters. The magnetic fluxes that generate the currents in both the stator and rotor are significantly coupled. In particular, the fluxes contain a very large common component that flows through the machine's air gap, resulting in strong coupling. This strong coupling means that the currents in the rotor and stator strongly influence each other. In the event of a torque change or disturbances, this strong coupling can increase the settling time or cause overshoots. The control dynamics are therefore degraded. Consequently, it may be necessary to maintain corresponding reserves, such as current reserves, which are associated with additional costs and space requirements.Furthermore, this can result in a reduced lifespan or adverse driving characteristics. Even with unexcited electric machines, precise control requires a detailed machine model and therefore significant computational demands.

[0006] Feedforward control with a detailed machine model, in turn, is associated with high computational effort for fast calculations using this detailed machine model. In particular, calculating the flow rate, which depends on the reference currents, is computationally intensive.

[0007] Therefore, one challenge is to implement the control of an externally excited electrical machine with sufficient precision and limited computing resources.

[0008] This problem is solved by the subject matter of the independent claim. Further applications of the procedure described here, as well as properties, features, embodiments, and advantages, will become apparent from the dependent claims, the description, and the figures.

[0009] It is proposed to use feedforward control based on a machine model for load relief (or decoupling the respective controllers for the stator and rotor in separately excited machines). The voltages for the PWM output, in particular the phase voltage, are calculated accordingly. For calculations using the machine model, determining the desired flux (i.e., the magnetic reference flux) is necessary. There are embodiments that address the control of a separately excited electrical machine. In these, the excitation voltage is also calculated in the feedforward control. The term "voltage calculation" refers specifically to the calculation of the phase voltage in methods for non-separately excited electrical machines (permanently excited electrical machines), and to the calculation of both the phase voltage and the excitation voltage in methods for separately excited electrical machines.

[0010] To reduce computational effort, it is proposed to perform this calculation in a slower time slice, i.e., with a lower repetition rate. The terms "slower" and "lower repetition rate" refer to the time slice and repetition rate (sampling rate) of a feedforward controller, respectively, and to a voltage calculation that uses the reference flux as its input. The calculated flux (dependent on the reference currents or reference current components) can be computed in a slower time slice and passed on in correspondingly equal proportions to the feedforward controller, which is faster (compared to the flux calculation). The flux is then computed in the faster time slice together with a deviation controller (i.e., together in a closed control loop driven by the reference current components). The signal representing the flux is thereby stretched or interpolated over the time slices.The control loop also operates with time slices that are shorter than the time slices used to determine the flux, or with a repetition rate that is faster (i.e., lower) than the repetition rate used to determine the flux.

[0011] The higher temporal resolution of the feedforward control (compared to the flux calculation) relieves the controller and decouples it from the input load. In particular, the slower flux calculation and thus stepwise adjustment (i.e., time stretching or interpolation) have no significant impact, since the deviation controller only receives the changed setpoints with a delay.

[0012] This computationally reduced, yet precise (and therefore computationally intensive) feedforward control offers a significant advantage for separately excited synchronous machines. In general, this control method can also be advantageously used for reluctance synchronous machines or permanent magnet synchronous machines. The inverter's hardware (actuators and sensors) does not require modification.

[0013] A method for controlling the phase voltage of a multiphase electrical machine (such as a synchronous machine, for example, a separately excited synchronous machine, a permanent magnet synchronous machine, or a reluctance motor) is described. The method involves receiving a target torque, preferably together with a target speed. This value(s) can be provided by a higher-level controller configured to process an operator request. Using a predefined mapping, the target torque (and optionally the target speed) is mapped to reference current components. These reference current components represent a reference current, which is typically represented by space vector components. It is possible for the mapping to also map the target torque (and optionally the target speed) to a reference excitation current. In other words, it is possible for the mapping to map the target torque (and optionally the target speed) to a reference excitation current.The inverter maps the target speed to a reference current and a reference excitation current. The term "reference current" refers to a phase current, i.e., a current in the phases of the electrical machine (i.e., in the stator), or to a (multiphase) current supplied by the inverter or flowing between the inverter and the electrical machine.

[0014] The mapping can include a lookup table and / or a model and / or (approximate) equations that implement this mapping. The target value(s) are specified, in particular, by a numerical (time- and value-discrete) signal, such as a digital signal.

[0015] A closed-loop control system regulates the phase voltage of the electric machine (or its stator). The phase voltage is the controlled variable, i.e., the output variable of the control loop. The input variable, or reference variable, of the control loop is the reference current component (or the reference current), which results from the target torque (or target speed) through mapping. The closed-loop control system uses, in particular, a vector representation or component representation (space vector component representation). The vector representation or component representation mentioned here is, in particular, a rotor-fixed representation with an active and a reactive component, for example, in the form of a d,q representation (i.e., a representation in space vector components). It can be provided that, in addition to the phase voltage, an excitation voltage is also controlled, preferably together with the phase voltage. The input variable, i.e.,The reference variable of the control loop then additionally includes the reference excitation current.

[0016] Based on the reference current components, i.e., the reference current (and, if applicable, the reference excitation current), feedforward voltage components (i.e., a feedforward voltage) are generated. This is performed by a feedforward control unit. The feedforward voltage components (and, if applicable, the excitation voltage) are determined using electrotechnical relationships. These relationships are illustrated below as examples (I) - (III). The relationships yield a generator voltage. <w eZ 'P c / Q (due to rotation with angular velocity ro) eZ ) again, give that voltage RI d q again, which is due to the phase current I dfq across the ohmic components R of the stator windings, and / or they output a voltage 'P' c / Q again, which is caused by the time-dependent change of the magnetic flux 'P' c / Qarises. If a separately excited electrical machine is used, then the voltage R can also be determined in the contexts according to formula (III). E I E be reproduced, which are due to an excitation current I E on ohmic components of the excitation winding R E the voltage drop, and / or the voltage 'Pf' that results in the excitation winding from the change in magnetic flux (in the rotor). The feedforward voltage components U d q and, if applicable, also the excitation voltage U E (Excitation pre-control voltage) are derived in particular from the reference current components I according to the following relationships dq determined:

[0017] (I) U d = RI d - la) el W q +

[0018] (

[0019]

[0020] II) U q = RI q + lo> eZ 'Pd + £ 'P. and possibly

[0021] (III) U E = R E I E+ 'Pf

[0022] These pre-control voltage components U d q (Generally: a feedforward voltage) are the result of the feedforward control. Also the excitation voltage U. E , if used, is the result of the feedforward control.

[0023] The feedforward control system has a voltage calculation that maps the reference flux to the feedforward voltage components (feedforward voltage) as shown here. The voltage calculation can also be configured to map the reference flux to the excitation feedforward voltage.

[0024] The feedforward voltage (and optionally the excitation feedforward voltage) is combined with a control loop variable, in particular with a manipulated variable of the control loop. The control loop can include a control element, which preferably receives a control deviation, such as the difference between the reference current components and a feedback signal, for example, actual current components. The output signal of the control element (e.g., a control voltage) is combined with the feedforward voltage, preferably by addition, which can be unweighted or weighted. This combined quantity can be referred to as the manipulated voltage and in particular forms the manipulated variable of an actuator of the control loop. The actuator is in particular implemented by an inverse space vector transformation, i.e.,A transformation converts the combination (sum) of the feedforward voltage and the control voltage (output signal of the control element) from a space vector representation into a time-domain signal. Based on this signal, a pulse-width modulated (multiphase) control signal for the inverter or electric machine can be generated. In other words, the feedforward voltage, obtained through the feedforward control, is added to the voltage signal within the control loop, with the added signal forming the manipulated variable of the control loop, i.e., it forms the basis for the output signal (which may still need to be converted from a numerical value into a pulse-width modulated signal to drive the inverter accordingly). In this way, the feedforward voltage is fed into the control loop. This is particularly relevant for the control of the stator current; the aforementioned quantities are therefore—unless explicitly stated otherwise—stator-related quantities.Furthermore, a control system for the excitation voltage or excitation current may be provided, for which a (further) control element may be included. The reference variable is, in particular, the excitation current (together with the reference current), whereby the excitation feedforward voltage is added to the (excitation) manipulated variable. The excitation feedforward voltage is derived, in particular, from an (excitation) control element that compares the actual excitation current with the target excitation current (reference excitation current) and outputs a corresponding manipulated variable (excitation voltage). The excitation voltage can also be transformed into a pulse-width modulated signal (PWM excitation voltage), which is preferably applied to the excitation coil of the electric machine in order to generate the (actual) excitation current (excitation signal).

[0025] The feedforward control includes a flux calculation. This maps the reference current (i.e., the reference current components) to a magnetic reference flux. A machine model is used for this mapping. The reference flux is obtained when applying the machine model, assuming that the reference current flows within the electric machine. This model represents (i) the relationship between the stator-related reference current and the resulting stator flux, and furthermore (ii) the coupling (i.e., the relationship) between the stator flux (generated by the reference current) and the rotor flux, and can, in particular, represent structural and / or magnetic properties and behaviors, such as saturation effects and / or flux density distributions within the stator and / or rotor of the electric machine.In the case of a separately excited electric machine, the model can further represent the relationship between an excitation signal and the resulting rotor flux. The model can also take into account other operating parameters such as temperature, rotational speed, the inverter's DC supply voltage, the AC voltage component in a DC link upstream of the inverter (DC link ripple current), and others. The model represents the characteristics of the electric machine, particularly its electromagnetic properties, and is used to map an operating point (defined by the reference current and, if applicable, other parameters) to the corresponding flux. Depending on the desired temporal and numerical precision and the number of characteristics considered, this can require considerable computing power.

[0026] To reduce the required computing power without significantly impacting the control behavior, the flux calculation is performed at a lower repetition rate than the voltage calculation. The flux calculation is performed at a first repetition rate, and the voltage calculation at a second repetition rate. The first repetition rate is lower than the second. Therefore, over a given time period, the flux calculation generates fewer results (i.e., magnetic fluxes) than the voltage calculation. This allows the voltage calculation to operate with a higher temporal resolution than the flux calculation. The less frequent flux calculation compared to the voltage calculation does not result in any significant inaccuracy, as the flux changes at a slower rate than the control voltage or feedforward voltage (or feedforward voltage).The excitation feedforward voltage must be adjusted. One embodiment provides that the flux calculation generates current reference fluxes at specific time intervals. For the time between successive reference fluxes, fill values ​​based on the last (most recent) reference flux or on the most recent (multiple) reference fluxes are used for the voltage calculation. The fill values ​​can correspond to the last reference flux until a new reference flux is calculated using the flux calculation. Between two successive reference fluxes generated by mapping, one or more fill values ​​are used. The fill values ​​between two calculated reference fluxes can be the same, or the qualitative change (decrease or increase) and / or quantitative change (difference between successive calculated reference fluxes) can be kept constant when determining the fill values.In particular, the fill values ​​correspond to a combination of the two most recently generated reference flows. The reference flows are generated by mapping. The calculated reference flows and the fill values ​​between them are transmitted to the stress calculation. Using the flow calculation, a reference flow is generated as the initial result value by mapping. This, as well as at least one subsequent fill value, is passed to the stress calculation. This continues until the flow calculation generates a subsequent result value by mapping, i.e., generates another reference flow. After the at least one fill value following the first reference flow (initial result value), the subsequent reference flow is output. The reference flows or the fill values ​​can be submitted with a delay (relative to the generation of the respective values). After mapping (i.e.,(Generation of the reference flows) a delay of the generated reference flows can therefore be implemented. The level of the at least one fill value can correspond to the first result value (i.e., the first of successive reference flows) that is followed by the at least one fill value, can correspond to the subsequent result value (i.e., the second of successive values) that precedes this reference flow, or can correspond to a combination of the flows between which the at least one fill value is provided.

[0027] Embodiments provide that the at least one fill value represents the change from the first result value to the subsequent result value. In other words, the at least one fill value lying between two reference flows (generated by mapping) can represent the change between these flows. In particular, several fill values ​​lying between two (directly consecutive) reference flows can increase or decrease in (preferably equal) steps if the reference flows differ. If there is a decrease between two (directly) consecutive reference flows, the fill values ​​decrease in equal steps from the first reference flow to the (directly) subsequent reference flow. If there is an increase between two (directly) consecutive reference flows, the fill values ​​increase in equal steps from the first reference flow to the (directly) subsequent reference flow.The fill values ​​and the reference flows are preferably delivered at equal time intervals, i.e., in a fixed, constant time grid. This procedure can be described as interpolation. The interpolation is, in particular, a linear interpolation starting from successive reference flows, whereby the fill values ​​are generated by linear interpolation.

[0028] The voltage calculation also outputs the feedforward voltage according to a time grid (or receives the flux calculation values ​​according to a time grid), where this time grid corresponds to the time grid in which the reference fluxes, supplemented by the fill values, are output. However, the time grid in which the reference fluxes are calculated is wider (by a preferably integer factor) than the time grid of the voltage calculation (or than the time grid of the input and output values ​​of the voltage calculation). This factor is the ratio of the second repetition rate (concerning the voltage calculation) to the first repetition rate (concerning the flux calculation).

[0029] It is specifically proposed that the initial result value (or a first reference flow) is generated by mapping and stored until the subsequent result value (or a directly following reference flow) is generated by mapping. Furthermore, at least one fill value is generated that lies between the initial result value and the subsequent result value. After the subsequent result value (i.e., the following reference flow) has been generated, the stored initial result value, the subsequent fill value (or at least one), and the subsequent result value are passed to the stress calculation as reference flows (in that order). This intermediate storage can be performed in a buffer device that is either downstream of or integrated into the flux calculation.

[0030] The calculation of the reference current can be performed using the first (slower) repetition rate or the associated (wider) time grid. This is particularly relevant for the first time grid. Here, too, at least one fill value or at least one intermediate value can be provided between successive calculated reference currents. The fill values ​​and the reference currents (as a combination) are calculated and output according to the second repetition rate. The fill values ​​can be determined in the same way as the fill values ​​between reference flows, except that the reference currents replace the reference flows.

[0031] Alternatively, the reference currents are calculated (and output) using the second repetition rate or the associated time grid. This applies to the stator-related reference currents, i.e., the reference phase currents, and can also include the excitation reference current.

[0032] Mapping the target torque to the reference current (and, if applicable, also to the excitation reference current) can be performed (using the given mapping) at the first repetition rate. The reference current can be interpolated to a discrete-time signal at the second repetition rate. Furthermore, the flux calculation can be performed using the model at the first repetition rate. The calculated reference fluxes can be interpolated to a discrete-time signal at the second (higher) repetition rate. The aforementioned fill values ​​can be used for this purpose. In particular, the calculated reference fluxes can be interpolated to a discrete-time signal with a time grid that is narrower than the grid used to calculate the reference fluxes. The time grid of the interpolated signal corresponds to the second (narrower) time grid. The repetition rate of the interpolated signal corresponds to the second repetition rate.Interpolation changes the time grid from the first to the second time grid. Interpolation also changes the signal's repetition rate from the first (slower) repetition rate to the second. One aspect is that the reference current (possibly also the excitation reference current) provided by the mapping is delayed. This can be achieved with a delay element. The delay element, in particular, delays the signal for a duration that encompasses the mapping of at least two consecutively mapped reference flows. The reference current thus delayed is used as the setpoint of the control loop. This setpoint is fed into a differential element, in which a feedback current (actual current, such as a measured current) is subtracted from the setpoint. The reference current resulting from mapping the target torque is therefore the setpoint of the (closed) control loop and is output with a delay.The delay of the reference current and the delay of the (mapping-derived) reference flux are such that the delayed quantities refer to the same point in time. The control loop's setpoint (the delayed reference current) and the feedforward output (the feedforward voltage) are delayed equally. The control loop's setpoint (the delayed reference current) and the feedforward output (the feedforward voltage) are generated at the same, lower, second repetition rate. In other words, they are generated according to a first time grid that is narrower than the second time grid used for the flux calculation. The control loop also operates (like the voltage calculation) at the narrower (i.e., first) time grid, or at the higher repetition rate, i.e., the second repetition rate. This approach can also be applied to the excitation current.

[0033] The second repetition rate is preferably an integer multiple of the first repetition rate. The second time grid operates with grid intervals that are also an integer multiple of the grid intervals of the first time grid. The flux calculation uses the first time grid, or the first repetition rate. The control loop and the voltage calculation use the second time grid, or the second repetition rate.

[0034] Further embodiments provide that the mapping of the target torque to the reference current operates at the second repetition rate. The reference current is passed to the control loop as a setpoint and to the voltage calculation as an input (not interpolated) at this repetition rate. The flux calculation can receive the reference current as an input at the second (faster) repetition rate, but processes it at the first (slower) repetition rate and outputs the reference flux at the first (slower) repetition rate accordingly (to the voltage calculation). The flux calculation can thus receive a time-averaged signal of the reference current, whereby the time averaging is performed, in particular, with a time window whose length corresponds to the grid time interval of the first time grid.

[0035] The procedure described here can be implemented using a computer program that, when executed on a processor, performs the described procedure. The computer program can be a control program. The computer program has an input interface that allows the input of the target torque. The computer program has an output interface that allows the output of the phase voltage, either in the form of space vector components or in the form of a pulse-width modulated signal (for controlling the inverter switches). The computer program can, in particular, contain code sections that implement the mappings and calculations as well as the control loop; each code section can contain interface sections for the input of the corresponding quantities mentioned here. Specifically, the computer program can represent the (parameterizable) model as a code section (with corresponding interfaces).The delay can also be implemented as a code section (with appropriate interfaces).

[0036] A control device configured to implement the procedure described here may be used. The control device may have an input configured to receive a signal representing the target torque. The control device may have an output configured to output a signal representing the phase voltage for controlling an inverter in the time domain or as space vector components. A feedback input may be provided for inputting a feedback signal to the closed-loop feedback loop described here. The control device may be implemented by a computer program running on a processor, by logic circuits, or a combination thereof.The quantities and signals described here are represented in particular as time-discrete signals and can also be value-discrete, for example in the form of digital signals.

[0037] Finally, an electric vehicle drive can be provided with an inverter, an electric machine connected to it, and a control device. The control device is connected to the inverter and is configured to control the inverter according to the phase voltage. The inverter and the electric machine correspond to the devices described herein with the same designation. The control device provided in the vehicle drive is, in particular, the control device described above. The control device provided in the vehicle drive is configured to implement the procedure described herein.

[0038] Figure 1 shows a symbolic block diagram to illustrate the procedure described here.

[0039] Figure 2 serves as an example to illustrate the different repetition rates or time grids.

[0040] Figure 1 illustrates an example of the control of a phase voltage U. phase described. This is transmitted at an output AU to an inverter IN, which uses the pulse-width modulated signal PWM (which is the phase voltage U). phase The electrical machine EM (corresponds to or is formed from it) controls an electrical machine EM. In the example shown, the electrical machine EM is a separately excited machine, which is excited according to the excitation current UE (with an excitation current resulting from this).

[0041] A target torque Tq ReqThe input is given to an input El of the depicted control system. A predefined mapping (torque mapping) TA maps the target torque Tqpeq to a reference current l. d Ref Figure TA shows which currents are required for the electric machine to produce the desired target torque Tq. Req to generate. In the illustrated embodiment, figure TA represents the target torque Tq. Req also on an excitation reference current 'E_Ref Ob.'

[0042] Figure TA can also take into account at least one additional operating parameter, such as the level of the inverter's DC supply voltage IN, the temperature of the electric machine EM, or similar parameters. The reference current Id.q Ref is shown in a space vector component representation. This applies to all signals or quantities shown in Figure 1 with the index d, q, or d,q.

[0043] The reference current ld,qRef (and the excitation reference current l)E ) is fed into a flow calculation FB, which uses a machine model implemented there to determine the available reference current Id.q Ref (and the excitation reference current l). E) to a reference flow VW. This model-based mapping T'Ref is performed with a first repetition rate. This results in a flow signal provided with a sampling rate corresponding to the first repetition rate. A temporal interpolation "AVW / 4" stretches the flow signal obtained by mapping to a signal VW with a higher sampling rate, corresponding to a second repetition rate (which is larger than the first repetition rate). Figure 2 shows this interpolation of mapping-obtained values ​​E1, E2, E3 (generally: Ei), which represent a reference flow, to a tighter time grid by adding interpolated filler values ​​F. In the example shown, the second repetition rate is four times higher than the first repetition rate, so that a total of 4 values ​​(one mapped value Ei and 3 filler values ​​F) correspond to each mapped value Ei.

[0044] In the example shown, the reference current Id, q Ref obtained by mapping (reference symbol TA) (and also the excitation reference current l) E) also time-interpolated, see block "AW 4", in particular in the same way (with the same temporal division) as the interpolation AVW / 4. The interpolated signal that indicates the reference current (time-interpolated) is denoted by Id, q Ref. The interpolated signal that indicates the excitation reference current (time-interpolated) is denoted by lE_Ref'. Thus, the signals Id, q Ref (as well as lE_Ref') and VW have the same temporal resolution or the same repetition rate (rate at which the respective values ​​are output or transmitted) or the same sampling rate. This resolution is an integer multiple of the temporal resolution of the quantities Id, q Ref (as well as lE_Ref') and VW, which result from mapping (mapping TA) or by performing the flux calculation (FB). The repetition rate or sampling rate of the interpolated signals Id.q Ref (as well as lE_Ref') and VW is an integer multiple (the same as the previously mentioned multiple) of the repetition rate or

[0045] The sampling rate of the quantities Id.q Ref (as well as lE_Ref') and VW, which result from mapping TA or calculating FB, respectively, is the rate at which the calculation FB or the mapping TA is performed (and new values ​​are generated by calculating TB or mapping TA). This sampling rate can be considered the first sampling rate.

[0046] The sampling rate can be referred to as the second repetition or sampling rate. The repetition or sampling rate of the quantities Id.q Ref (as well as lE_Ref') and VW or a subsequent voltage calculation SB can be referred to as the second repetition or sampling rate.

[0047] The flux calculation FB is followed by the stress calculation SB shown. This receives the interpolated quantities Id.q Ref (as well as lE_Ref') and VW.

[0048] The voltage calculation SB operates with the (higher, second) repetition rate, which corresponds in particular to the sampling rate of the aforementioned interpolated quantities. The voltage calculation SB maps the reference flux (VW) to a feedforward voltage Ud.q(Pre). The voltage calculation SB further maps the reference flux (VW) to an excitation feedforward voltage llE(Pre). The quantities llE(Pre) and Ud.q(Pre) can be referred to simply as feedforward voltages.

[0049] The voltage calculation SB is performed at a rate corresponding to the (second) repetition rate. For each (discrete-time) value of the quantity V^Ref, a feedforward voltage Ud.q(Pre) (as well as IWre)) or a respective (discrete-time) value thereof is determined by mapping. In particular, the voltage calculation SB determines the feedforward voltage from the interpolated reference flux VW and the interpolated reference current I. d,q Ref(as well as the interpolated excitation reference current lE_Ref') the feedforward voltage Ud,q(Pre) (or the excitation feedforward voltage llE(Pre)). It can be used for any pair of values ​​of V and I d,q Ref (or lE_Ref') a value of the feedforward voltage Ud,q(Pre) (or the excitation feedforward voltage llE(Pre)) is calculated. Figure 1 shows the relevant relationships that form the basis for this calculation in the form of formulas within block SB, which are explained in more detail in the general description section. The feedforward voltage Ud,q(Pre) (as well as the excitation feedforward voltage llE(Pre)) is provided with a sampling rate that corresponds to the second repetition rate. The second repetition rate corresponds to the second time grid; the first repetition rate corresponds to the first time grid.

[0050] A closed control loop RK has a differential element D, which represents the difference (i.e., the control deviation) between a reference input I. d,q RefThe difference element D is formed between a reference input lE_Ref (k-2) (setpoint) and a feedback signal lEjvieas (k-2) (actual value). Furthermore, the difference element D determines, in particular, the difference (i.e., the control deviation) between a reference input lE_Ref (k-2) (setpoint) and a feedback signal lEjvieas (k-2) as the actual value.

[0051] The difference is fed as a control deviation ("error") to a control element PI. The control element can be a P, I, PI, PD, or PID controller, or another type of controller. The control element outputs a control voltage to an adder element A. The adder element A also receives the feedforward voltage Ud.q(Pre) (and in particular the excitation feedforward voltage llE(Pre)) from the voltage calculation SB. The additive circuit A generally has the task of combining the result of a feedforward control VS (here the feedforward voltage Ud.q(Pre) and, in particular, also the excitation feedforward voltage llE(Pre)) with the output of the control element PI. The combination (addition) of the feedforward variable (feedforward voltage Ud.q(Pre) or llE(Pre)) with the control element output results in the manipulated variable in the form of the manipulated voltage Ud.q or UE. In this way, the feedforward voltage Ud.q(Pre) is fed into the control loop RK as a feedforward variable.Furthermore, the excitation pre-control voltage llE(Pre) is fed into the control loop RK as a pre-control variable in this way. The feed-in point of the pre-control variable U. d,q Pre (or UE^ OIS feed-in to the control loop RK, lies between the control element PI and the actuator SG. The manipulated variable (manipulated voltages Ud,q and UE), which results from the combination of the output of the control element and the feed-in variable (feed-in voltages Ud,q(Pre), UE(Pre)), is transformed into another form. Here, the manipulated voltage Ud,q, which is represented in space vector form, is converted into a voltage signal U phaseThe signal is converted and output at an output AU of the depicted control device, which is used to drive an inverter IN. The excitation control voltage UE can also be converted by the actuator SG into a voltage representation UE', for example by pulse width modulation. This conversion can also be performed in the inverter IN, which can also provide the excitation voltage for the excitation coil of the electric machine EM.

[0052] The voltage signal U phaseThis is also referred to as phase voltage. The phase voltage represents the voltage applied to the individual phases of the electrical machine EM. The phase voltage is multi-phase. In particular, the phase voltage is a pulse-width modulated signal suitable for controlling the inverter IN and especially suitable for controlling the inverter's switches in the open or closed state (pulse-width modulated). Due to the control signal U, the inverter IN outputs phase A power signal is output to the electric machine EM, specifically to its rotor windings. The electric machine shown as an example is a separately excited machine, whereby the excitation voltage UE, which is calculated in the voltage calculation SB and (converted into a pulse-width modulated excitation voltage signal UE') is output to the rotor winding of the electric machine EM, is used to generate the excitation current.

[0053] The phase current I is phase The phase current is detected and fed into the feedback input RE. The phase current is (compared to the quantity I). d,q Ref ) is provided with a delay, which is why the reference symbol for the phase current has the suffix (k-2). The detected or measured phase current I phase The signal is transformed into a space vector representation in the transformation block TR. The signal l is obtained by transforming the phase current. d q Meas (k-2) and is compared to size I d,q Ref delayed (which is why the relevant reference symbol of the phase current bears the suffix (k-2)). This feedback signal l d q Meas (k-2) is input to the differential element D of the control loop RK (as a subtrahend). The differential element D receives the reference current Id.q Ref as a setpoint in delayed form Id.q Ref (k-2). The time-interpolated reference current I d,q Ref The signal is transmitted via the delay element with the reference symbol "delay" and is thereby delayed; this results in I d,q Ref(k-2). This procedure is also used for the detected excitation current l E (k-2) executed, from which (through the transformation TR) a measured excitation current l E meas (k-2) is generated. The transformation TR for the excitation current can be a smoothing process that averages or smooths the current measured at the rotor winding over time, thereby determining the (effective), actual current as the actual value of the excitation or the excitation current. TR (e.g., smoothing / low-pass filtering) results in the measured excitation current l E (k-2) the measured excitation current L (k-2).

[0054]

[0055] In the illustrated embodiment, the process steps or...

[0056] Functional blocks in area G1 have a lower repetition rate than the process steps or functional blocks in area G2. In particular, the mappings and calculations (mapping TA, field calculation FB) in area G1 are executed at a lower repetition rate than the calculations or control in area G2 (such as the voltage calculation SB or the closed-loop control RK). Therefore, temporal interpolation between these areas is provided, which uses the reference current I. d,q Ref (or the errer reference stream) and the reference flow T'Ref are interpolated to a finer time grid. This is symbolically represented in Figure 1 by the notation Alp. ef / 4 (which also refers to the excitation current) or AT'Ref / 4. Accordingly, for synchronization, the delay (see reference symbol "delay") of the reference current Id.q Ref (i.e., the interpolated reference current) is provided to obtain ld,q Ref(k-2). In the same way, the delayed excitation reference current lE_Ref' (interpolated excitation reference current) is derived from the excitation reference current lE_Ref(k-2). The presented feedforward control VS uses a model-based flux calculation FB, which calculates discrete-time flux values ​​at a lower repetition rate than the voltage calculation SB calculates discrete-time feedforward voltage values ​​Ud,q(Pre) (or UE(Pre)).

[0057] In the flux calculation FB, a machine model is used, i.e., a model that represents the electrical and magnetic (as well as electromagnetic) relationships of the electric machine EM. This model is a machine model that describes the relationship between the stator-related reference current I. d,q Ref and the resulting stator flux V'd.q. The machine model also represents the coupling between the stator flux V'd.q and the rotor flux V'E. Furthermore, the machine model, particularly when using a separately excited machine EM, represents the relationship between an excitation signal (UE or IE) and the resulting rotor flux V'E in the electrical machine EM. This can be represented as a function or mapping. V P E (I E ) are designated, where I E This can represent, for example, the target excitation current lE_Ref. When using a separately excited machine EM, the following can therefore be added to the diagram shown. 1 P R ef (I R e f ) also the illustration , P E (I E ). The figure 1 P R e f (I R e f ) can thus stand as a proxy for mapping the target current of the stator to the (stator) flux as well as for mapping the target current of the rotor (excitation current) to the (rotor) flux. The symbolic term, I Re f / 4" refers to the reference current (stator current) or its lower sampling rate than SB, and in particular also to the excitation current. I d,q Ref and lE_Ref are referred to as reference current components, in particular as (stator-related) reference current component I d,q Ref and as excitation current components lE_Ref.

[0058] The temporal interpolation required to adapt to the different velocities is shown in Fig. 2. Figure 2 shows three discrete-time values ​​of the reference flux Ψ. Refin the form of the successive values ​​E1, E2, E3. These are calculated using the displayed repetition rate or sampling rate (via flow calculation FB). To interpolate the time intervals between these calculated (i.e., model-based mapping) flow values ​​E1, E2, E3, fill values ​​F are inserted. In the example shown, the fill values ​​F are determined by the flow values ​​between which they lie. Therefore, to interpolate the fill values ​​F, the calculated values ​​E1, E2, E3 between which the fill values ​​F lie are required. Thus, to determine the fill values ​​F between E1 and E2, these values ​​E1 and E2 are needed. E1 and E2 have the same value, so the fill values ​​between them also have the same value. The fill values ​​after E2 (and before E3) increase at a constant rate of increase because E3 is larger than E2. This results in uniform steps from E2 to E3.After E3, the fill values ​​decrease again because a subsequent value obtained through mapping is smaller than E3. Thus, the fill values ​​reflect the progression of the values ​​obtained through mapping, between which they are positioned. Other embodiments provide that the fill values ​​are determined by the calculated value that precedes them (until a more recent calculated value is available). For example, the fill values ​​after E1 can be defined as values ​​corresponding to the value of E1. Other embodiments may provide that the fill values ​​between two calculated values ​​are the same, and this value can be the average of the two calculated values. The values ​​of the fill values ​​are therefore based on a calculated value that directly precedes them, or their values ​​are based on the calculated values ​​between which they are positioned, or their progression is determined by the calculated values ​​between which they are positioned.

[0059] If the fill values ​​(in terms of their progression and / or their values) are based on several calculated values, then several calculated values ​​are required to calculate the fill values. Therefore, the fill values ​​are only determined once the calculated values ​​have been obtained through mapping. In this context, "calculated values" specifically refers to the values ​​obtained through mapping. Figure 2 thus shows a time-interpolated signal of the flux VW (represented by the calculated values ​​E1, E2, E3) as a (discrete-time) flux signal VW that also includes the fill values ​​F. This results in a sequence of discrete-time values ​​with the values ​​E1–E3, between which the fill values ​​F are specified. The repetition rate of the interpolated flux VW in the example shown is four times the repetition rate of the non-interpolated flux VW.The integer multiple that represents the ratio between the first and second repetition rates is the sum of the number of calculated values ​​E1 - E3 and the number of filler values ​​F, where the sum is divided by the number of calculated values ​​E1 - E3. The numerator and denominator of the ratio refer to the same time interval (or time intervals of equal length).

Claims

Patent claims 1. Method for controlling the phase voltage (U) phase ) a multiphase electrical machine with the steps: Receiving a target torque (Tq) Req ) and mapping the target torque (TqReq) to a reference current (I) d,q Ref ) using a predefined mapping (TA); Rules of phase voltage (U) phase ) as a controlled variable according to a closed control loop (RC), whose reference variable is defined by the reference current (I d,q Ref ) is provided for, whereby a feedforward control (VS) based on the reference current (I) d,q_ref ) a pilot voltage (U d,q Pre ) generated, which are fed into the control loop (RK), whereby the feedforward control (VS) - includes a flow calculation (FB) that uses a machine model to determine the reference flow (I d,q Ref ) on an associated magnetic reference flux (Ψ Ref ) depicts, and the feedforward control (VS) - includes a stress calculation (SB) that uses the reference flux (Ψ Ref ) on the feedforward voltage (U d,q Pre ) depicts, where the flux calculation (FB) is performed with a first repetition rate that is lower than a second repetition rate with which the stress calculation (SB) is performed.

2. The method of claim 1, wherein the flow calculation (FB) uses a reference flow (Ψ Ref ) as an initial result value (E1) by mapping and passes the initial result value and at least one fill value (F) to the stress calculation (SB) until the flux calculation generates a subsequent result value (E2) by mapping, wherein the magnitude of the at least one fill value (F) corresponds to the initial result value (E1), the subsequent result value (E2), or a combination thereof.

3. Method according to claim 2, wherein the at least one fill value represents the change from the initial result value (E1) to the subsequent result value (E2).

4. A method according to claim 2 or 3, wherein the initial result value (E1) is generated by mapping and stored until the subsequent result value (E2) is generated by mapping, the at least one fill value (F) being generated between the initial result value (E1) and the subsequent result value (E2), and, after the subsequent result value (E2) has been generated, the stored initial result value (E1), the following at least one fill value (F), and the subsequent result value (E2) are stored as a reference flow (Ψ). Ref ) is submitted to the voltage calculation (SB).

5. Method according to one of the preceding claims, wherein the mapping (TA) of the target torque to the reference current (I) d,q Ref ) is performed at the first repetition rate.

6. Method according to one of the preceding claims, wherein the mapping of the target torque (Tq) is performed. Req ) on the reference current (I d,q Ref) is performed using the given mapping (TA) with the first repetition rate and the reference current (I d,q Ref ) are interpolated temporally to a time-discrete signal with the second repetition rate.

7. Method according to one of the preceding claims, wherein the reference current provided by mapping (I d,q Ref ) is delayed and the reference current (I) is thus delayed d,q Ref (k-2)) is used as the reference variable of the control loop RK.

8. Method according to any of the preceding claims, wherein the second repetition rate is an integer multiple of the first repetition rate.

9. Computer program configured to execute the method according to any one of the preceding claims when executed on a processor.

10. Control device configured to execute the method according to any one of claims 1-8, wherein the control device has an input (El) configured to receive a signal representing the target torque (TqReq) and an output (AU) configured to output a signal representing the phase voltage (U). phase ) configured to control an inverter (IN) in the time domain or reproduces as space vector components, and configured with a feedback input (RE) for inputting a feedback signal (I) phase ) a feedback loop of the closed control loop (CC).

11. Electric vehicle drive with an inverter (IN), an electric machine (EM) connected thereto and with a control device according to claim 10, wherein the control device is connected to the inverter (IN) in a controlling manner and is configured to control the inverter (IN) according to the phase voltage (U). phase ) to head towards.

Images