WO2021037533A1

WO2021037533A1 - Method for operating an electric machine

Info

Publication number: WO2021037533A1
Application number: PCT/EP2020/072364
Authority: WO
Inventors: Maximilian MANDERLA
Original assignee: Robert Bosch Gmbh
Priority date: 2019-08-28
Filing date: 2020-08-10
Publication date: 2021-03-04
Also published as: DE102019212886A1

Abstract

The invention relates to a method for operating an electric machine (100) which is supplied with voltage by a converter with which a DC voltage is converted into an AC voltage, wherein phase voltage values (U_a, U_b, U_c) to be applied to the electric machine (100) are determined in the course of a model predictive control (150) with a model (155) of the electric machine and then applied, wherein physically achievable maximum values are used as thresholds for the phase voltage values in the model predictive control (150).

Description

description

title

Method for operating an electrical machine

The present invention relates to a method for operating an electrical machine which is supplied with voltage by means of a converter fed from a DC voltage circuit, as well as a computing unit and a computer program for its implementation.

State of the art

Electrical machines, in particular variable-speed electrical machines, can be operated as a motor or generator on an inverter that is fed by a DC voltage circuit, or on a converter that has a DC voltage intermediate circuit. The voltage values at the terminals of the electrical machine are thus limited by the direct voltage or the two potentials of the direct voltage (intermediate) circuit.

A typical way of regulating the voltage in such an electrical machine is field-oriented regulation, in which a space vector representation, in particular in d-q coordinates, is used. This type of control is based on the so-called zero condition, according to which a star point (if present) of the load is not connected to the neutral conductor. As a result, the sum of the phase currents is always zero. In the case of a three-phase electrical machine, only two coordinates, typically referred to as d and q, are sufficient for controlling and regulating the electrical machine.

Disclosure of the invention According to the invention, a method for operating an electrical machine as well as a computing unit and a computer program for its implementation are proposed with the features of the independent claims. Advantageous refinements are the subject matter of the subclaims and the description below.

The invention relates to a method for motor or generator Be driving an electrical machine that is supplied with voltage by means of a converter that converts a direct voltage into an alternating voltage. The converter can be an inverter in which an alternating voltage with a corresponding number of phases is generated only from a direct voltage. Likewise, it can also be a converter, in particular a frequency converter, further in particular a so-called two-level converter, in which an alternating voltage is initially converted into a direct voltage and this then again into an alternating voltage with a frequency and generally changed Amplitude is converted. In the latter case, one speaks of a so-called DC voltage intermediate circuit in the converter. The converter is controlled in particular in pulse width modulation or PWM.

As already mentioned, a typical way of regulating the voltage in such an electrical machine is field-oriented regulation, in which a space vector display, in particular in dq coordinates, is used. From the mentioned limitations of the phase voltages (in fixed Koordina th) a polytopic or polygonal voltage limitation which rotates with the electric field frequency (time variant, ie changing with time) results. With a two-level converter and a three-phase electrical machine, this voltage limit is a regular hexagon. For a more detailed explanation of this, reference is also made to the description of the figures. Although the proposed method is useful for a three-phase electrical machine, the proposed method can also be used with a different number of phases, for example five or seven. This time variance represents a technical control difficulty because, depending on the current rotor position, different limits or maximum values for the phase voltage values, i.e. those values of the voltage that can be applied to the phases of the electrical machine, can be relevant and active. To avoid this difficulty, the permissible polytopi cal control area for the voltage in the (field-oriented) control can be approximated by a (rotation- and time-invariant) inscribed circle (in the polygon) with a maximum diameter. For a more detailed explanation, reference is made here to the description of the figures.

Modulation-related mean voltage values can then (during a regulation cycle) be selected so that they always lie in the mentioned inscribed circle. However, this leads to a non-utilization of the available, maximum electrical voltage and thus, depending on the operating point, to losses in dynamics and / or efficiency.

In the proposed method, phase voltage values to be applied to the electrical machine are determined within the framework of a model predictive regulation with a model of the electrical machine as manipulated variables and then applied to the corresponding connections or phase voltage connections of the electrical machine. In model predictive control, physically achievable maximum values are used as limits for the phase voltage values. These maximum values for the phase voltage values are, in particular in a space vector representation or in ab coordinates, a regular polygon, the number of corners of the polygon depending on a number of the phases of the electrical machine, in particular corresponding to twice the number of phases . In the case of a three-phase electrical machine, it is a regular hexagon, for example. The inscribed circle is not used for the maximum values, but the actual physical maximum values are used. A torque of the electrical machine or a current, for example, can be used as a control variable in model predictive control, but this then in particular in dq coordinates. In this way, the above-mentioned disadvantages when using the inscribed circle for the maximum values can be avoided or at least reduced, since the physically existing polytopic restrictions can be fully exploited by the control. Because of the model predictive control, the use of these actual or physical limits is, as has been shown, much easier than with a conventional control. Particular advantages result, in particular, from the rise time that can be achieved and the associated speed of control in the case of transient processes (torque changes) both in the basic setting range and in the field weakening range.

In the case of model predictive control, a (dynamic) model of the system to be controlled - in the present case a model of the electrical machine - is stored. A special feature here is that the phase voltages that can be applied to the machine terminals are used as inputs or manipulated variables and not (as is usual) the voltages in the rotating d-q system. The phase voltages represent the hexagon or, in general, the polytope of the maximum possible voltages.

In model predictive control, target or reference values (or setpoints), such as currents or torques to be set, are used in each control step, that is, with each call, several numerical simulations (over a limited timeframe Horizon) of a plant model with varying manipulated variables, ie voltages, performed. By means of online optimization, that voltage is selected which achieves the best performance (eg fast or low-energy achievement of the reference value) taking into account the voltage limits, ie the physically achievable maximum values. This is where the model, target values in the form of a cost function, for example, and voltage limits come together. The voltage optimized in the hexagon or polytope (best solution) is finally output and connected to the electrical machine. More detailed explanations of such a model of an electrical machine can also be found, for example, in "Schröder, D .: Electrical Drives - Basics, Springer 2009 (Chapter 6.5, p.391)".

Furthermore, maximum utilization of the available electrical voltage is also possible in transient scenarios and there is also better dynamics and faster operating point changes (transient processes). In addition, there is no need to switch to special overmodulation methods, instead a uniform control approach is possible. In addition, only a reduced application effort is required compared to classic control methods with switchover between operating areas (with and without overmodulation).

Furthermore, it is possible to specify sensible voltage directions (with regard to dynamics, efficiency and the like) when the voltage limit is reached by solving an optimization problem in real time, on which the model predictive control is based. Control targets can also be prioritized in an adaptable manner during runtime. In addition, there is a certain robustness through feedback in contrast to control procedures.

With regard to an entire electric drive system in, for example, a vehicle, there are also further advantages. For example, the higher dynamics of the control system enable better and higher-frequency damping of mechanical drive train vibrations, as well as faster gear changes in multi-speed transmissions. In addition, better utilization of the electric drive is possible through a lower supply of control voltage (lower field weakening), which brings cost advantages with it. It is therefore possible to change the control parameters online and without any additional effort, i.e. an adaptation depending on the operating strategy (dynamics vs. efficiency) is possible.

With the proposed method or with the model predictive control used, knowledge of a current angle of the rotor is useful. For this purpose, the current angle of the rotor can be measured using a sensor or a rotor position sensor can be measured. However, the current angle can also be estimated, for example using an observer. It is also conceivable to use both variants to protect the other.

A computing unit according to the invention, e.g. a control unit of a motor vehicle or a control unit or a control or regulating unit for an electrical machine, is set up, in particular in terms of programming, to carry out a method according to the invention.

The implementation of a method according to the invention in the form of a computer program or computer program product with program code for performing all method steps is advantageous, since this causes particularly low costs, especially if an executing control device is used for other tasks and is therefore available anyway. Suitable data carriers for providing the computer program are, in particular, magnetic, optical and electrical memories, such as hard drives, flash memories, EEPROMs, DVDs, etc. A program can also be downloaded via computer networks (Internet, intranet, etc.).

Further advantages and embodiments of the invention emerge from the description and the accompanying drawings.

The invention is illustrated schematically using an exemplary embodiment in the drawing and is described below with reference to the drawing.

Brief description of the drawings

FIG. 1 shows schematically a circuit with an electrical machine, in which a method according to the invention can be carried out.

FIG. 2 shows schematically a space vector illustration to explain a method according to the invention. FIG. 3 schematically shows a sequence of a method according to the invention in a preferred embodiment.

FIG. 4 shows schematically a sequence of a model predictive control.

FIG. 5 schematically shows voltage and current curves when using a method according to the invention in a preferred embodiment.

FIG. 6 shows schematically a voltage profile when using a method according to the invention in a preferred embodiment.

Embodiment (s) of the invention

In Figure 1, a circuit with electrical machine 100 with Ro tor 103 is shown schematically, in which a method according to the invention can be carried out. Here, a converter 110 is provided which has three half bridges each with two switches, once Si and S2, once S3 and S4 and once S5 and S ₆ , to which a DC voltage U _dc is applied.

The switches can in particular be semiconductor switches, for example MOSFETs or IGBTs. The converter 110 can, for example, be part of a computing unit 115 embodied as a control unit for the electrical machine. A capacitor C is also provided. Between each two switches there is a tap for one of the three phases U, V and W, which are connected to corresponding connections (not shown) on a stator winding of the electrical machine 100.

In FIG. 2, a space vector representation is shown schematically to explain a method according to the invention. For this purpose, the axes a and ß are provided as real and imaginary parts in a stator-fixed ab coordinate system. By means of the Clarke transformation or a / b transformation, the axes U, V, W are converted into a two-axis coordinate system with the axes a, β, where, by definition, the axis U coincides with the real axis a. The sum of the three phase currents is always zero at any point in time.

Each half bridge in the circuit shown in Figure 1 can assume two different switch positions. For the half-bridge formed by switches Si and S2, this would be, for example, S1 closed and S2 open as the first position ("high-side" or "1") and S1 open and S2 closed as the second position ("low-side" or "0"). Since three half bridges are necessary for a three-phase three-phase system, there are eight possible switch positions and thus eight switching states. Each switch position results in a different voltage constellation between the phases U, V and W and thus also a different voltage space vector.

The two switch positions in which either all three upper or all three lower switches are closed ([111] or [000]) are an exception. With these switch positions, all three phases are short-circuited. This means that no voltage can be measured between the phases. These two voltage vectors are referred to as zero voltage space vector and are indicated in FIG. 2 by the vectors Uo and II7. The remaining six possible switch positions result in the positions in the stator-fixed a-ß coordinate system identified by the vectors or voltage space vectors Ui to U _Ö.

Each voltage space vector also generates a specific alignment of the flux density distribution in the electrical machine. In order to be able to commutate an electrical machine continuously (sinusoidally), the six voltage indicators Ui to U _{Ö are} not sufficient, since voltage space indicators must be switched to the electrical machine with any angles and amounts.

To achieve this, the basic principle of pulse width modulation is applied. If, for example, a voltage _{space vector U a is to be} output, which is exactly half the angle of the voltage space vector Ui and U2, but the maximum This can be achieved by outputting the voltage space vector Ui alternately with the voltage space vector U2.

The ratio of the two times is decisive for the resulting voltage space vector. In the example given, the two times al must be selected to be of equal length in order to obtain the desired voltage space vector. Due to a typically existing low-pass effect of the stator windings, there is an averaged current in the electrical machine and thus the desired space vector, the desired alignment of the magnetic flux density.

For variable operation of the electrical machine, it is usually necessary to be able to choose the amplitude of the output voltage, i.e. the amount of the voltage space vector, as desired. In order to achieve this, the zero-voltage space vector can be used. If, for example, the voltage space vector U _{b is} to be output, the ratio of the output times of the voltage space vector Ui and U2, as in the previous example, must be the same. In order to be able to reduce the amount of the resulting voltage space vector, an additional time is required in which a zero voltage space vector is output.

As can also be seen from the above explanation, all voltage space vectors with the maximum possible amount that can be set describe a hexagon, which is designated here with U _max for the maximum values that can be achieved. In the general case of a multiphase electrical machine, a polygon or generally a polytope results accordingly.

For motorized operation of the machine, the voltage space vector must rotate with the rotor. The (rotating) voltage regulation can be simplified by transforming it into a coordinate system fixed to the rotor (so-called d / q transformation or Park transformation). _{For this purpose, the phase voltages U a} , U _b and U _c applied to phases U, V and W according to FIG

transformed into the voltage _{values or voltages U d} and U _q in the space vector representation, where cp is the current rotor angle.

When controlling the machine or the converter, ie when transforming d / q back into ce / ß, it must be taken into account that, as explained, the maximum voltage varies with the rotor position, i.e. is time-variant. As already mentioned, this time variance represents a technical control difficulty, so that up to now an inscribed circle with maximum diameter has been approximated to the polygon in the ab coordinates and used for the control. This achieves a time invariance of the maximum voltage. Such an inscribed circle is _{denoted by U max in} FIG.

In Figure 3, a sequence of a method according to the invention is shown schematically in a preferred embodiment. The model predictive control or the corresponding controller 150 with a model 155 of the electrical machine receives reference values R (setpoint values) and measured values M as input variables. _{Values for the phase voltages U a} , U _b and U _{c are} output to the electrical machine as output variables (manipulated variables). A target value is in particular the torque.

Instead of the usual calculation of U _d and U _q , the phase voltage vector in space vector representation, the phase voltages U _a , U _b and U _{c are} optimized in a direct manner within the framework of the proposed model predictive control. By utilizing the angle-dependent relationship mentioned above, these are directly incorporated into the respective system model.

In addition, an angle cp of the rotor 103 is measured, for example by means of a sensor 101, and a measurement is transmitted to the controller 150, a e.g. Position in a permanent magnet synchronous machine. Alternatively, an observer 102 can also be used to estimate or determine the angle, here denoted by f ', for example a flux angle in an asynchronous machine.

In Figure 4, a sequence of a model predictive control is shown schematically for more detailed explanation of the proposed method. For this purpose, a general state X and a general manipulated variable Y over time t are applied.

Characteristic for the model predictive control is a pre-calculation (so-called prediction) of the future system behavior on a sliding horizon H and the necessary optimization of possible voltage curves (as manipulated variables). At a point in time X _k , the future behavior is shown in the upper diagram with a dashed line over a horizon H, optimized with regard to a manipulated variable, which then defines a certain value or course of the manipulated variable (voltage) in the lower diagram with a dashed line shown, results. This determined, optimal manipulated variable is then set.

The horizon is then shifted backwards by the time period At and such an optimization is carried out again _{for this new horizon H at the point in time X k + i.} The actual course of the state X and the actually or ultimately used values for the manipulated variable Y are shown with solid lines.

For this calculation of the future courses, knowledge of future angles cp is required. For this reason, the (dynamic) model of the electrical machine can be expanded to include a prediction of the angular position. With an electrical angular velocity w and a controller sampling time dT, an estimated position angle is obtained for a prediction of k steps

By inserting the above-mentioned coordinate transformation taking into account the calculation of the angle cp in the model of the electrical ma- machine in dq coordinates results in an angle-dependent system description of the controlled system with currents I _d and l _q and phase voltages U _a ,

Ub and Uc.

This model together with the angle prediction is explicitly taken into account in the context of the model predictive control and allows the inclusion of the maximum (physical) limits U _max in the form of simple box restrictions for the phase voltages, as shown in FIG. This allows the use of simple, real-time-capable optimization methods within the framework of model predictive control.

In Figure 5, voltage and current curves are shown schematically when using a method according to the invention in a preferred embodiment. For this purpose, a voltage U and a current I are plotted over time t. U _a , U _b and U _c or I _a , l _b and l _c indicate the actual phase voltages or phase currents. U _q and U _d or I _d and l _q indicate the phase voltages or phase currents (after appropriate conversion) in space vector coordinates, U * _q and U * _d or I * _d and l * _q indicate the associated setpoints.

The typical and exemplary measurement results shown here apply to a suddenly changed torque request with a setpoint jump at t = 0. The new torque or current level is reached after just a few milliseconds, although the phase voltages are temporarily limited.

FIG. 6 schematically shows a voltage profile when using a method according to the invention in a preferred embodiment, which corresponds to the situation shown in FIG. For this purpose, the voltage values U _a and Uß are plotted against one another in the space vector representation, where the maximum values already shown in FIG. 2 are drawn _{in at U max.}

It becomes clear here that the largest possible inscribed circle (according to FIG. 2) is left during the transition phase and the maximum overmodulation is used (The edge of the hexagon is reached). Thus, overall, efficient operation of the electrical machine is possible.

Claims

Expectations

1. A method for operating an electrical machine (100) which is supplied with voltage by means of a converter (110) through which a direct voltage (U _dc ) is converted into an alternating voltage, in which the electrical machine (100) Phase voltage values to be applied (U _a , U _b , U _c ) are determined in the context of a model predictive control (150) with a model (155) of the electrical machine as manipulated variables and then applied, with the model predictive control (150) as limits for the Phase voltage values physically achievable maximum values (U _max ) are used.

2. The method according to claim 1, in which the maximum values (U _max ) for the phase voltage values (U _a , U _b , U _c ) are shown in a space vector representation as a regular polygon, with a number of corners of the poly gons of one The number of phases (U, V, W) of the electrical machine (100) depends, in particular twice the number of phases ent speaks.

3. The method according to claim 1 or 2, wherein a current angle (cp) of a Ro tors of the electrical machine (100) is determined and is taken into account in the model predictive control (150).

4. The method according to claim 3, wherein the current angle (cp) of the rotor (103) is measured using a sensor (101).

5. The method according to claim 3 or 4, wherein the current angle (cp ¹ ) of the Ro tors (103) is estimated.

6. The method according to any one of the preceding claims, wherein a two-level converter is used as the converter (110).

7. The method according to any one of the preceding claims, wherein an electrical machine (100) with three, five or seven phases is used.

8. Computing unit (115) which is set up to carry out all method steps of a method according to one of the preceding claims.

9. A computer program which causes a computing unit (115) to carry out all method steps of a method according to one of claims 1 to 7 when it is executed on the computing unit (115).

10. Machine-readable storage medium with a computer program according to claim 9 stored thereon.