Method and apparatus for controlling an electric machine

Abstract

A method is provided for controlling an electric machine to deliver a requested torque, wherein the electric machine comprises a plurality of sub-machines and each sub-machine is driven by a corresponding one of a plurality of inverters. The method comprises the steps of selecting, based on the requested torque, one of a first mode and a second mode; allocating, in accordance with the selected mode, a torque to each of the plurality of sub-machines such that the sum of the allocated torques equals the requested torque; and controlling the inverter of each sub-machine so that the sub-machine generates the allocated torque. In the first mode, all sub-machines are allocated with a non-zero fraction of the requested torque, and in the second mode, at least one of the plurality of sub-machines is allocated with a torque equal to zero. The method is characterized in that any current flowing through windings of a sub-machine with an allocated torque equal to zero is interrupted in the second mode.

Description

[0001] Method and Apparatus for Controlling an Electric Machine

[0002] TECHNICAL FIELD

[0003] The present invention relates to a technique for controlling electric machines having one or more sub-machines.

[0004] BACKGROUND OF THE INVENTION

[0005] In many applications of electric machines, including traction for electric vehicle, the output torque of an electric machine is to be controlled throughout a large range of rotational speed. Especially in the context of electric vehicles, the efficiency by which the electric machine can convert electrical into mechanical power is of utmost interest. Conventional electric machines, however, can provide a high efficiency only for a comparatively limited range of torque and rotational speed. In a low-torque regime at low to medium speed, efficiency is generally suboptimum. This is particularly disadvantageous for electric vehicles during highway driving or extra-urban driving where speeds are relatively constant and traction torque is only a fraction of the full capability of the machine.

[0006] Torque modulation is a conventional technique for improving the electric efficiency in a low-torque range. Patent document US 2012 / 0123624 A1 , for instance, discloses a technique for intermittently operating an electric machine so that intervals with relatively large output torque and intervals with zero output torque alternate each other, thus generating the desired low output torque as an average over time. This technique may be referred to as torque pulse-width modulation (torque PWM). If the torque generated during the active intervals is adapted so as to yield maximum efficiency at the given rotational speed, overall electric efficiency can be improved accordingly. Similar techniques are also disclosed in patent documents WO 2013 / 089515 A1 and JP 3716534 B2. The disadvantage of torque PWM is that, depending on the modulation frequency, vibrations and noise are unavoidable and / or that an uneven acceleration of the vehicle may be perceived as annoying.

[0007] An alternative approach for improving electric efficiency of electric vehicles in a low-torque regime is to provide two, differently sized electric motors that are either operated individually or jointly, depending on torque requirements. Different motors may be adapted to different torque regimes and selecting among these motors and their combinations may improve overall electric efficiency, albeit only at the expense of increased mechanical complexity. An example for this approach is disclosed, for instance, in patent documents CN 103273857 A and JP H07-131994 A.

[0008] Electric machines, such as reluctance machines and synchronous machines with permanent magnets, are generally provided with a plurality of coil sets that are arranged on the stator in a circumferential direction. An electric machine with, say, 12 coils (six pole pairs) may be operated as a three-phase machine, wherein each phase comprises four series-connected coils that are circumferentially distributed along the stator. If each of these coils is provided with separate connectors, the currents through these coils may be controlled independently of each other and the same machine may also be operated as a four-phase machine or a six-phase machine with three coils or with two coils connected in series for each phase, respectively. Further, an electric machine with N*M pole pairs, may also be considered as an electric machine comprising N sub-machines with M pole pairs each and each sub-machine may be controlled with a dedicated inverter independently of the other sub-machines. Each of the N sub-machines may thus occupy a fraction of 3607N of the stator circumference.

[0009] Patent document US 10,069,450 B2, for instance, discloses a brushless DC motor with 12 stator coils that are split into two separate 3-phase sub-motors. In a normal operating mode, the two sub-motors operate together as one motor to output a normal operating power of the motor. In an emergency operating mode, when one sub-motor has become faulty and generates a braking torque, the normal sub-motor is operated to generate a torque to compensate for the braking torque produced by the faulty sub-motor.

[0010] Patent document US 8,749,192 B2 discloses an electric machine for electric vehicles. The electric machine comprises eight coil sets with each coil set having three coil sub-sets that form a three-phase sub-motor. Each of the eight sub-motors is controlled independently of the other sub-motors.

[0011] SUMMARY OF THE INVENTION

[0012] Conventional techniques for controlling an electric machine thus leave room for improvement in terms of efficiency, manufacturing costs and usability. Therefore, it is an object of the present invention to provide a method and an apparatus for controlling an electric machine that can achieve higher electric efficiency in a low-torque regime without the problems caused by the conventional techniques.

[0013] This is achieved by the features of the independent claims. Preferred embodiments are the subject matter of the dependent claims.

[0014] The inventor(s) realized that in a conventional electric vehicle, even when the electric machine is not producing any torque, losses are generated by the alternating current in the stator winding, which heats not only the inverter, but also the winding, the stator steel, the rotor magnets and the rotor yoke. At low load, these losses represent a substantial part of the total loss and are therefore important for the overall vehicle consumption in standard cycles, such as the WLTP, as well as during highway driving and extra-urban driving, where speeds are relatively constant and traction torque is only a fraction of the full capability of the machine. These effects are even more pronounced in inverter-motor systems with low inductance, such as direct drive machines that are used in in-wheel applications for electric vehicles.

[0015] The inventor(s) thus realized that the electric efficiency of a motor-inverter system can be improved by disconnecting the inverter during zero-torque intervals. This may be particularly beneficial for electric machines with several sub-machines in a low-torque regime, in which the requested torque may be allocated to only a subset of the submachines so that the motor-inverter systems of the remaining sub-machines may be deactivated. In this manner, electric losses can be decreased, compared to a situation where all of the motor-inverter systems are actively performing a pulse-width modulation of the coil currents (current PWM). Moreover, the active motor-inverter system may be operating at a point at which the requested torque can be generated with higher efficiency as compared to a situation where the requested torque was divided evenly over all submachines.

[0016] It is thus the particular approach of the present invention to operate an electric machine comprising two or more sub-machines, depending on the requested torque, in either a first mode, wherein all sub-machines are jointly operated to generate the requested torque, or a second mode, wherein the requested torque is generated by only a subset of the submachines and the inverters of the other sub-machines are deactivated.

[0017] According to a first aspect of the present invention, a method is provided for controlling an electric machine to deliver a requested torque, wherein the electric machine comprises a plurality of sub-machines and each sub-machine is driven by a corresponding one of a plurality of inverters. The method comprises the steps of selecting, based on the requested torque, one of a first mode and a second mode; allocating, in accordance with the selected mode, a torque to each of the plurality of sub-machines such that the sum of the allocated torques equals the requested torque; and controlling the inverter of each sub-machine so that the sub-machine generates the allocated torque. In the first mode, all sub-machines are allocated with a non-zero fraction of the requested torque, and in the second mode, at least one of the plurality of sub-machines is allocated with a torque equal to zero. The method is characterized in that any current flowing through windings of a sub-machine with an allocated torque equal to zero is interrupted in the second mode.

[0018] Preferably, in the second mode, the inverter of each sub-machine with an allocated torque equal to zero is turned off and / or disconnected from windings of the corresponding submachine. In this manner, electric losses within the windings and / or the inverter of the deactivated sub-machines are eliminated.

[0019] According to a preferred embodiment, in the second mode, the at least one sub-machine with an allocated torque equal to zero is permuted, preferably periodically, among the plurality of sub-machines in order to equalize a thermal load within the electric machine. Preferably, permuting the at least one sub-machines with an allocated torque equal to zero among the plurality of sub-machines comprises allocating a zero torque to a first subset of the plurality of sub-machines and a non-zero torque to the other sub-machines during a first interval of time, and allocating a zero torque to a second subset of the plurality of submachines different from the first sub-set and a non-zero torque to the other sub-machines during a second interval of time. Problems caused by local overheating or an inhomogeneous temperature distribution within the electric machine may thus be avoided.

[0020] Preferably, permuting the at least one sub-machines with an allocated torque equal to zero among the plurality of sub-machines comprises gradually increasing the torque allocated to a first sub-machine from zero to a non-zero target value while decreasing simultaneously the non-zero torque allocated to a second sub-machine to zero. In this manner, while keeping the sum of the torques equal to the requested torque, smooth operation of the electric machine can be ensured and ripples in the output torque avoided.

[0021] Preferably, a duration of an interval in which the torque allocated to the first sub-machine is gradually increased from zero to the non-zero target value while the non-zero torque allocated to the second sub-machine is simultaneously decreased to zero is significantly shorter than a duration of an interval in which a torque equal to zero is allocated to the first sub-machine or the second sub-machine, preferably at least by a factor of 10 or 100. In this manner, efficiency can be improved by making sure that the electric machine is operating most of the time with the optimum allocation of torque to its sub-machines.

[0022] Preferably, all sub-machines are allocated with the same non-zero fraction of the requested torque in the first mode. All sub-machines are thus subjected to the same thermal and mechanical load and temperatures will be equally distributed throughout the electric machine.

[0023] Advantageously, the first mode is selected when the requested torque exceeds a first predefined threshold value, and wherein the second mode is selected when the requested torque falls below a second predefined threshold value lower than the first threshold value. Electric efficiency is thus improved in the low-torque regime and hysteresis prevents rapid switching forth and back between the two modes of operation.

[0024] Preferably, the first threshold value and / or the second threshold value are a function of a rotation speed of the electric machine. Since electric efficiency may not only depend on torque but also on rotational speed, efficiency can be further improved if this dependency is taken into account for switching between the two modes of operation.

[0025] Preferably, the second mode is only selected when a rotation speed of the electric machine exceeds a predefined minimum value and / or falls below a predefined maximum value. This may be beneficial in a situation where field weakening is used at high rotational speeds or during start-stop operation of an electric vehicle in order to avoid unnecessary switching of the inverters.

[0026] In a preferred embodiment, the method further comprises the step of selecting, in the second mode and based on the requested torque, a number of sub-machines to which a non-zero torque is allocated, said number being greater than zero and less than the plurality of sub-machines. This allows for further improving electric efficiency over a wider range of requested torque, in particular if the requested torque exceeds the torque that can be generated in an efficient manner by a single sub-machine.

[0027] According to a second aspect of the invention, a controller for an electric machine with a plurality of sub-machines is provided. Each sub-machine is driven by a corresponding one of a plurality of inverters. The controller is configured to perform all steps of a method according to the first aspect of the invention.

[0028] According to a third aspect of the invention, a motor-inverter system is provided that comprises an electric machine with a plurality of sub-machines; a plurality of inverters, each inverter being configured to drive a corresponding one of the plurality of submachines; and a controller according to the second aspect of the invention.

[0029] According to a fourth aspect of the invention, an electric vehicle with a motor-inverter system according to the third aspect is provided.

[0030] In the context of the present disclosure, the electric machine may comprise NxM pole pairs forming N sub-machines having M pole pairs each. The number of phases is preferably equal to three, but a different number of phases is possible, in particular four, five or six phases. The number of sub-machines N is at least two and not limited to any particular number. Advantageous implementations may include two, three, four and eight submachines.

[0031] BRIEF DESCRIPTION OF DRAWINGS

[0032] The present invention is illustrated by way of example, and not by way of limitation, in the accompanying drawings in which:

[0033] Fig. 1 shows an efficiency map of a conventional motor-inverter system;

[0034] Fig. 2 shows an efficiency map of a motor-inverter system according to an embodiment of the present invention;

[0035] Fig. 3 shows a block diagram of a motor-inverter system with three sub-machines according to an embodiment of the present invention;

[0036] Fig. 4 shows a waveform of a torque transition between two sub-machines according to an embodiment of the present invention; and

[0037] Fig. 5 shows a waveform of a torque generated in the improved efficiency area by a motor-inverter system with two sub-machines according to an embodiment of the present invention.

[0038] DETAILED DESCRIPTION

[0039] Fig. 1 shows the electric efficiency of a conventional motor-inverter system as a function of rotational speed and torque. As can be seen, there is an optimum point of operation (“sweet spot”) with an efficiency of more than 93% for a speed in the range of 600 RPM to 800 RPM and a torque of about 400 Nm. For torques lower than 300 Nm, inverter-induced losses become dominant and the efficiency drops dramatically.

[0040] In order to improve efficiency in the low-torque regime, the electric machine is split into independently controlled sub-machines. All sub-machines are substantially identical, especially in terms of their electromechanical properties, such as power rating, torque generating capabilities, etc. Instead of operating all sub-machines in parallel in order to jointly generate the requested torque, at least one of the sub-machines is deactivated and the one or more remaining sub-machines are driven in order to generated the requested torque. In this manner, inverter induced electric losses in the windings of the deactivated sub-machines and the corresponding inverters are eliminated. Moreover, the torque per sub-machine is increased if the requested torque is to be generated by a lower number of sub-machines, so that the point of operation of the active sub-machines may be closer to the sweet spot. By varying the number of active and inactive sub-machines in accordance with the requested torque and rotational speed, the electric efficiency of the electric machine may thus be improved.

[0041] Fig. 2 shows the electric efficiency of a motor-inverter system according to an embodiment of the present invention as a function of rotational speed and torque. For torques higher than about 300 Nm, the efficiency map is substantially identical to that shown in Fig. 1. In the low-torque regime below 300 Nm, however, efficiency is significantly improved, as it is apparent from the contour lines being stretched out in the direction of lower torque values. This is achieved, in this example, by switching to a single sub-machine operation (improved efficiency mode of operation) when the requested torque drops below a certain threshold value.

[0042] During single sub-machine operation, all but one sub-machines are deactivated and the requested torque is generated by the remaining sub-machine only. The inverter of each of the deactivated sub-machines is turned off, pulse-width modulation of the winding current is stopped, all windings of the sub-machine are disconnected and all currents flowing therein are interrupted. All inverter induced electric losses in the deactivates sub-machines and their inverters are thus eliminated.

[0043] When the requested torque increases again, operation is switched back to full submachine operation (regular mode of operation), wherein all sub-machines are driven in order to jointly generate the requested torque. In order to avoid a rapid switching forth and back between these two modes of operation, hysteresis is implemented by using two different threshold values for switching in the two different directions. Since electric efficiency is also dependent on rotational speed, the threshold value may vary as a function of rotational speed as well. Fig. 2 shows the threshold values for switching to single sub-machine operation (lower bold line) and to full sub-machine operation (upper bold line) as a function of the rotational speed. The area below these lines is the torque / speed domain in which the electric machine is operated in the single sub-machine mode. This area may be delimited by a high-speed limit so that there is no switching to single sub-machine operation if the rotational speed exceeds this limit. This may be advantageous when field weakening is used and the inverter of a sub-machine cannot be turned off due to high induced voltage, as it may be the case for a synchronous machine with permanent magnets. In the case of a reluctance machine, there would be no such limit or would be imposed at a higher speed by system efficiency. There may be also a low-speed limit below which there is no switching to single sub-machine operation. This may be useful for electric vehicles because unnecessary inverter switching can be avoided during start-stop operation, i.e., during short periods of time in when the requested torque is either comparatively high or equal to zero so that no substantial gain in efficiency can be obtained anyway.

[0044] In the above example, the electric machine comprises two sub-machines and switching is performed between a single sub-machine mode wherein only one sub-machine is active, and a full sub-machine mode wherein all sub-machines are jointly operated. However, the invention is not limited to this particular example and may also be applied to electric machines with more than two sub-machines. In such a case, more than two modes of operation may be defined with different numbers of active sub-machines, depending on the requested torque and / or the rotational speed. This allows for further improving the electric efficiency, because a torque higher than the torque that can be delivered by a single sub-machine may still be generated more efficiently by only two sub-machines than by the entire electric machine. By adaptively activating a variable number of sub-machines out of a total of three or more sub-machines, electric efficiency may thus be improved for a large range of torque and speed requirements.

[0045] Specifically, the number of activated sub-machines (and inverters) may depend on the balance between the joule losses generated in the winding, which increase with a square of the torque, and the losses caused by current PWM modulation (current ripples and related effects), which are relatively independent of the torque in the low torque region (<25% of full torque). The number of activated sub-machines for achieving maximum efficiency may be provided in form of a table or a map as a function of the requested torque or as a function of the requested torque and the rotational speed. Fig. 3 is a block diagram that illustrates the architecture of a motor-inverter system according to an embodiment of the present invention. The inverter 100 receives a command from a vehicle control unit (VCU, 10) which is internally converted into a motor torque command in block 110. The motor torque command is fed to sub-machine controller 120 which allocates the requested torque to the sub-machines 210-1 , 210-2 and 210-3. The key functionality of the sub-machine controller 120 is to distribute the requested motor torque between the sub-machines so that the sum of all requested sub-machine torques equals the requested motor torque. Signals indicating the amount of torque allocated to each sub-machine are forwarded to the respective sub-machine inverter 130-1 , 130-2 and 130-3, which controls the current flowing through the windings of the respective submachine. The control of said currents may be performed, for instance, by applying field- oriented control (FOC). Real time signals indicating motor speed and temperature of each sub-machine are fed back to the sub-machine controller 120, where they are used, together with static motor characteristic data, to decide how the requested torque is to be allocated to the sub-machines.

[0046] Depending at least on the requested torque, the sub-machine controller 120 may thus select one of the following modes of operation: an improved efficiency mode in which only one or some, but not all, sub-machines provide torque, and a regular mode in which all sub-machines contribute substantially equally amounts to the total output torque. Said operation modes may also include transitioning states, where torque generation is transferred from among the sub-machines.

[0047] According to a further aspect of the invention, thermal load on the individual sub-machines is equalized in the improved efficiency mode by regularly switching the active submachines among the entire set of sub-machines. This can be achieved, for instance, by periodically alternating the active sub-machine and the inactive sub-machine in an electric machine with a total of two sub-machines in a mode in which only one sub-machine is generating the requested torque. Similarly, in a machine with three or more sub-machines, the active sub-machine may be cycled periodically among the three or more submachines. Generally speaking, the subset of one or more, but not all, active sub-machines may be permuted within the set of all sub-machines.

[0048] Said switching, cycling or permuting of active sub-machines is performed on a thermal timescale Tth, i.e. the timescale on which heating and cooling of the sub-machines take place. This timescale is generally in the order tens of seconds or hundreds of seconds, and thus many orders of magnitude longer than the timescale on which the current through one of the windings is modulated by the inverter and still several orders of magnitude longer than the timescale on which the magnetic field is rotating in the electric machine.

[0049] When the generation of torque is switched from one sub-machine to another sub-machine during the improved efficiency mode of operation, the output torque of the first submachine is gradually decreased while the torque generated by the second sub-machine is gradually increased so that the total output torque remains constant (or follows the torque requested by the torque command). In this manner, smooth operation of the electric machine is guaranteed and there are no ripples in the output torque.

[0050] Fig. 4 shows an example waveform of the torque generated by a first sub-machine SM1 and a second sub-machine SM2 before, during, and after a transition of torque generation from SM1 to SM2. The requested torque (solid line) is constant in this example.

[0051] In the period, in which the generated torque is gradually transferred from a first submachine to a second sub-machine, both sub-machines are jointly activated and are thus likely not operating at the optimum point within the torque / speed diagram. The duration of this transition phase Tswis thus made as short as possible, preferably less than 10 %, or even less than 1 % of the interval Tth in which the torque is generated by only one of the two sub-machines. The duration of the transition phase Tswmay thus be in the order of tens or hundreds of milliseconds.

[0052] At the same time, Tswis based on the inverter calibration parameters of the FOC motor control and motor parameters to achieve smooth torque transition with minimized torque overshoots / undershoots. The inverter switching frequency, maximum motor electrical frequency, motor inductance and motor torque time constant are the main hardware powertrain constraints which limit the transition time. The inverter software may thus use state-of-the-art calibration parameters to optimize the inverter control to the powertrain transition time. These parameters are FOC PI gains and the torque ramp up / down capability of the motor [Nm / s],

[0053] Fig. 5 shows a further example waveform of the torque generated by two sub-machines during the regular mode of operation and the improved efficiency mode of operation.

[0054] In the first interval before time t=2, the requested torque (solid line) is allocated in equal shares to the first sub-machine SM1 (dotted line) and the second sub-machine SM2 (dotted line).

[0055] At time t=2, the requested torque drops below the threshold (lower border of the shaded area) for transitioning to the improved efficiency mode of operation. Consequently, the requested output torque is no longer distributed among the sub-machines but allocated, for a limited amount of time, to only one of the two sub-machines. In the interval between t=2 and t=6.5, the second sub-machine SM2 is thus deactivated and torque is generated by the first sub-machine SM1 only.

[0056] The active sub-machine SM1 is becoming increasingly warmer than the non-operating submachine SM2. At about t=6.5, i.e., after a “thermal time” Tth, torque generation is gradually transferred from the first sub-machine SM1 to the second submachine SM2 in order to equalize the thermal load among the sub-machines. The second sub-machine SM2 thus generates the entire output torque until time t=11 , when torque generation is transferred back to the first sub-machine SM1. In this manner, a thermally balanced operation of the sub-machines can be achieved and problems caused by local overheating and unequally distributed temperatures are avoided.

[0057] At time t=13.9, the requested torque exceeds the threshold (upper border of the shaded area) for transitioning to the regular mode of operation. Hence, the torque generated by the first sub-machine is gradually decreased while the torque generated by the second sub-machine is gradually increased until both sub-machines generate equal shares of the requested output torque at time t=14. Since the requested output torque still increases until time t=15, also the shares allocated to the two sub-machines is increased accordingly, until a plateau in the requested output torque is reached at time t=15.

[0058] Thermal balancing of the sub-machines during the improved efficiency mode of operation can be achieved in several ways. A straight forward solution is using a predetermined time Tth for periodically switching among the sub-machines. This time may be related to the thermal time constant of a single sub-machine. Alternatively, the switching among submachines may be performed on the basis of temperatures measured at each of the submachines. In this manner, sub-machines with the lowest and the highest measured temperatures may be activated and deactivated with a higher priority, thus effectively reducing temperature discrepancies among the sub-machines.

[0059] During single sub-machine operation, all but one sub-machines are deactivated and the requested torque is generated by the remaining sub-machine only. The inverter of each of the deactivated sub-machines is turned off, pulse-width modulation of the winding current is stopped, all windings of the sub-machine are disconnected and any currents flowing therein are interrupted. In order to support the above torque transition between two submachines, the FOC performed by the sub-machine inverters may thus implement an on- the-fly PWM restart, as it is known in the art of the motor control. The on-the-fly function enables / disables the PWM activation at any motor speed.

[0060] Summarizing, the present invention provides a method and an apparatus for controlling an electric machine that can achieve higher electric efficiency in a low-torque regime without the problems caused by the conventional techniques. This method, which is also referred to as advanced coasting control method for electric vehicles, is particularly advantageous for electric vehicles using in-wheel sub-machine electric motors.

Claims

Claims1. A method for controlling an electric machine to deliver a requested torque, said electric machine comprising a plurality of sub-machines, each sub-machine being driven by a corresponding one of a plurality of inverters, said method comprising: selecting, based on the requested torque, one of a first mode and a second mode; allocating, in accordance with the selected mode, a torque to each of the plurality of sub-machines such that the sum of the allocated torques equals the requested torque; and controlling the inverter of each sub-machine so that the sub-machine generates the allocated torque; wherein, in the first mode, all sub-machines are allocated with a non-zero fraction of the requested torque, wherein, in the second mode, at least one of the plurality of sub-machines is allocated with a torque equal to zero, characterized in that any current flowing through windings of a sub-machine with an allocated torque equal to zero is interrupted in the second mode.

2. A method according to claim 1 , wherein, in the second mode, the inverter of each sub-machine with an allocated torque equal to zero is turned off and / or disconnected from windings of the corresponding sub-machine.

3. A method according to claim 1 or 2, wherein in the second mode, the at least one sub-machine with an allocated torque equal to zero is permuted, preferably periodically, among the plurality of sub-machines in order to equalize a thermal load within the electric machine.

4. A method according to claim 3, wherein permuting the at least one sub-machines with an allocated torque equal to zero among the plurality of sub-machinescomprises allocating a zero torque to a first subset of the plurality of sub-machines and a non-zero torque to the other sub-machines during a first interval of time, and allocating a zero torque to a second subset of the plurality of sub-machines different from the first sub-set and a non-zero torque to the other sub-machines during a second interval of time. A method according to claim 3 or 4, wherein permuting the at least one submachines with an allocated torque equal to zero among the plurality of submachines comprises gradually increasing the torque allocated to a first submachine from zero to a non-zero target value while decreasing simultaneously the non-zero torque allocated to a second sub-machine to zero. A method according to claim 5, wherein a duration of an interval in which the torque allocated to the first sub-machine is gradually increased from zero to the non-zero target value while the non-zero torque allocated to the second submachine is simultaneously decreased to zero is significantly shorter than a duration of an interval in which a torque equal to zero is allocated to the first submachine or the second sub-machine, preferably at least by a factor of 10 or 100. A method according to claim 1 , wherein all sub-machines are allocated with the same non-zero fraction of the requested torque in the first mode. A method according to claim 1 , wherein the first mode is selected when the requested torque exceeds a first predefined threshold value, and wherein the second mode is selected when the requested torque falls below a second predefined threshold value lower than the first threshold value. A method according to claim 8, wherein the first threshold value and / or the second threshold value are a function of a rotation speed of the electric machine.A method according to claim 1 , wherein the second mode is only selected when a rotation speed of the electric machine exceeds a predefined minimum value and / or falls below a predefined maximum value. A method according to claim 1 , further comprising the step of selecting, in the second mode and based on the requested torque, a number of sub-machines to which a non-zero torque is allocated, said number being greater than zero and less than the plurality of sub-machines. A controller for an electric machine, said electric machine comprising a plurality of sub-machines, each sub-machine being driven by a corresponding one of a plurality of inverters, said controller being configured to perform all steps of a method according to any of claims 1 to 11 . A motor-inverter system comprising an electric machine with a plurality of sub-machines; a plurality of inverters, each inverter being configured to drive a corresponding one of the plurality of sub-machines; and a controller according to claim 12. A motor-inverter system according to claim 13, wherein the electric machine has NxM pole pairs forming N sub-machines having M pole pairs each. An electric vehicle with a motor-inverter system according to claim 13.

Images