Patent Grant
8648556
The present invention concerns the field of control of permanent magnet synchronous machines (PMSM).
A permanent magnet synchronous machine (PMSM) includes a stator and a rotor. Generally, the stator includes windings connected in a star shape and the rotor includes permanent magnets.
A PMSM is habitually powered by an inverter which enables the currents' ripple factor and torque of the machine to be reduced.
A PMSM has a high torque and very low inertia. In addition, it has relatively low inductances, which leads to fast responses of the currents and therefore of the torque.
It is therefore very advantageous to use PMSMs in the motor of high-power and high-performance actuators, notably in on-board aircraft systems.
Inverter 111 powers PMSM 103 using direct voltage. It enables amplitude and frequency voltages which can be adjusted by control device 101 to be imposed on the terminals of PMSM 103.
Control device 101 is configured to control vectorially the voltages of inverter 111 using electrical feedback data 106, and above all a precise indication of position θ of the rotor. This information habitually originates from a position or speed sensor positioned on the shaft of the machine.
However, there are also sensorless PMSM control devices (see, for example, the publication of Babak Nahid-Mobarakeh et al., entitled “Analysis of convergence of sensorless control laws of PMSMs based on estimation of the emf,” Revue Internationale de Génie Electrique [International Electrical Engineering Review], Article Vol 6/May 6, 2003-pp 547-577-doi:10.3166/rige.6.545-577).
It will be noted that the description concerning the device for sensorless control is essentially taken from the publication of Babak Nahid-Mobarakeh et al, above.
Generally, the electrical equations of a PMSM in a stationary frame of reference associated with the stator are described by:
where νc, νb, νc are the voltages of the stator phases, R is the resistance of the stator phases, ia, ib, ic are the currents of the stator phases, and ψa, ψb, ψc are the fluxes through the stator windings.
The PMSM can also be modelled very simply in a rotating frame of reference associated with the rotor.
By applying the Concordia transformation T32 and the Park transformation to the system of equations above the electrical equations in Park frame of reference d-q can be expressed as follows:
where νd, νq and id, iq are the direct and quadrature voltage and current components, L is the stator inductance, Ω is the speed of rotation of the rotor (i.e. the angular speed of frame of reference d-q), p is the number of pairs of poles of permanent magnets of the rotor, and ed, eq are the “emf” electromotive force components in frame of reference d-q defined by the following relationships:
ed=0
eq=pΩψf
where ψf is the flux of the magnets through the direct equivalent circuit.
Given that the position of rotor θ and angular speed Ω are not measured, frame of reference d-q cannot be located, and the components of the electrical magnitudes in this frame of reference are unknown.
Habitually, to resolve this problem, a rotating estimated frame of reference δ-γ is defined, the position and speed Ωc of which are known. Axis Oδ of estimated frame of reference δ-γ forms an angle
relative to stationary axis Oα and an angle φ relative to axis Od. Angle φ indicates the displacement between axes Oδ and Od.
The problem of sensorless vector control then consists in determining angular speed Ωc such that the difference of position φ between and θ is cancelled out.
The electrical equations of the PMSM in estimated frame of reference δ-γ can then be written as follows:
where νδ, νγ, and iδ, iγ are the voltage and current components in frame of reference δ-γ, Ωc the angular speed of frame of reference δ-γ, and eδ, eγ the components of the emf in frame of reference δ-γ defined by the following relationships:
eδ=pψfΩ sin φ
eγ=pΩψf sin φ
Generally, to control the sensorless PMSM, components eδ, eγ of the emf are estimated in rotating estimated frame of reference δ-γ. If the latter coincides with frame of reference d-q associated with the rotor, the direct component of the emf in the rotating estimated frame of reference becomes zero. This gives a criterion which enables position and speed Ωc of rotating estimated frame of reference δ-γ to be corrected in order that it can be synchronised with frame of reference d-q associated with the rotor. After this, the position and speed of the rotor can be deduced directly from the position and speed of estimated frame of reference δ-γ.
The problem of sensorless vector control can then be summarized by determining a control law defining angular speed Ωc and the components of stator voltages νδ, νγ in frame of reference δ-γ which ensures that φ is constantly maintained at zero, and the components of currents iδ, iγ at their references iδref, iγref determined by reference torque Γref.
This control device includes a torque-currents converter 137, vector control means 119, and a model of inverter-MSAF assembly 114 in frame of reference δ-γ.
Converter 137 accomplishes the transition from the torque to the current by transforming value of the reference torque (or set torque) Γref into the corresponding reference currents iδref, iγref in frame of reference δ-γ.
Vector control means 119 determine a control law to control inverter-MSAF assembly 114 whilst ensuring that φ is constantly maintained at zero (φf=0). This control law defines angular speed Ωc and the components of stator voltages νδ, νγ in frame of reference δ-γ as a function of the components of currents iδ, iγ obtained from the measurements of the back currents, and reference currents iδref, iγref.
A sensorless control device is particularly robust, since it has one fewer detection elements. A sensorless control device is thus simpler to produce, and can have a longer lifetime than a control device with sensors.
However, a position sensor is generally very precise and, consequently, a control device using a position sensor can regulate the voltages of the inverter powering the MSAF with greater accuracy than a sensorless control device.
The purpose of the present invention is consequently to provide a device to control an MSAF having optimum reliability and which is extremely safe, which are major preoccupations in aeronautics.
The present invention concerns a device for controlling a permanent magnet synchronous machine, “PMSM”, including a stator and a rotor, and powered by an inverter, where the control device includes:
This enables the availability of the PMSM in degraded mode to be increased by ensuring that the machine operates satisfactorily in the event of a malfunction of the sensor. It will be noted that this device favours control of the PMSM with the measurement sampled by the sensor, and switches to sensorless control only when an anomaly of the sensor has been detected, whilst making it possible to avoid a substantial difference between the two positions at the moment when control with sensors switches to sensorless control.
Advantageously, the estimation means include:
The control device according to the invention thus enables the PMSM to be controlled in the event of a malfunction of the sensor according to a control law having an area of global convergence which restricts convergence to the single desired operating point whatever the rotor position relative to the stator.
According to one embodiment of the invention, the said speed estimator includes a first estimator configured to determine a prior estimate {circumflex over (Ω)} of the rotational speed according to estimated component êγ of the emf associated with axis γ and a predetermined physical parameter Kf depending on the characteristics of the rotor's permanent magnets, according to the following formula:
and in that the non-linear corrector is configured to regulate the rotational speed by introducing a term which corrects said estimated value {circumflex over (Ω)} of the rotational speed according to the following formula:
where b is a predetermined operational setting, sign({circumflex over (Ω)}) is the sign of said estimated value {circumflex over (Ω)} of the rotational speed, êδ is the emf associated with axis δ, and where K is a non-linear factor which depends on the sign of the emf êδ associated with axis δ and on a coefficient ξ predetermined by means of the following formula and conditions:
K=1ξ·sign(êδ) where 0<ξ<1 and
Thus, the non-linear corrector allows the rotor's real frame of reference to be approached by constantly maintaining electromotive force eδ associated with axis δ at zero, by making all undesired operating points unstable. This causes a rapid convergence towards the desired operating point, whilst allowing a reversal of the rotational speed.
The said adjustment means are configured to execute a PI between measured rotor position θm and estimated rotor position {circumflex over (θ)}.
The adjustment means can advantageously include inhibition means to inhibit the adjustment means when a malfunction of the said sensor is detected.
If a malfunction is detected the correction made by the adjustment means is advantageously inhibited, since measured rotor position θm is probably false.
Advantageously, the said estimation means include initialization means to reinitialize estimated rotor position {circumflex over (θ)} with a last rotor position estimate {circumflex over (θ)}0 before the detection of a malfunction of the sensor.
This enables transitional oscillations to be prevented, and the torque value to be kept constant when control with sensors is switched to sensorless control.
The invention also covers a permanent magnet synchronous machine PMSM including a control device having the above characteristics.
The invention also covers an actuator in an aircraft including a PMSM having the above characteristics.
The invention also concerns a method for controlling a permanent magnet synchronous machine, “PMSM”, including a stator and a rotor, and powered by an inverter, where the control method includes the following steps:
The control method may also include the following steps:
The control method may also include the following steps:
where b is a predetermined operational setting, sign({circumflex over (Ω)}) is the sign of said estimated value {circumflex over (Ω)} of the rotational speed, êδ is the emf associated with axis δ, and where K is a non-linear factor which depends on the sign of the emf êδ associated with axis δ and on a coefficient ξ predetermined by means of the following formula and conditions:
K=1−ξ·sign(êδ) with 0<ξ<1 and
The invention also covers a computer program including instructions for the implementation of the above control method.
Other characteristics and advantages of the invention will appear on reading the preferential embodiments of the invention made in reference to the attached figures, among which:
PMSM machine 3 habitually includes stator windings 5 connected in a star shape with isolated power and a rotor 7 with permanent magnets 9 of symmetrical constitution with p pairs of poles (of which a single pair is represented here p=1).
PMSM 3 is powered by an inverter 11 which imposes voltages νa, νb, νc, at the terminals of stator windings 5. The inverter-PMSM assembly according to a triphase model is represented in simplified form by block 13.
Control device 1 includes a position sensor 15, electrical measuring means 17, and control means 19.
Position sensor 15 is a resolver (for example, a Hall effect sensor, or any other type of resolver) installed on PMSM 3 to sample accurately measurement θm of rotor position 7. The position can, of course, also be determined indirectly by measuring the speed of rotation of the rotor instead of its position. In this case the position sensor can include means to measure the rotational speed, and an integrator to determine the position.
Electrical measurement means 17 are configured to measure electrical feedback data and, more specifically, to measure stator currents ia, ib, ic of PMSM 3.
Means of control 19 receive signals concerning the rotor position, signals concerning stator currents ia, ib, ic measured by electrical measurement means 17, and data concerning reference torque Γref and/or reference rotation Ωref.
Control means 19 include a transformation interface 21 between the triphase model of inverter-PMSM assembly 13 and a two-phase model in a Park frame of reference. This transformation enables the physical magnitudes of a triphase model to be transformed to a two-phase model, and vice versa, according to the position of rotor 7.
Thus, control means 19 can control or inspect the operating point of PMSM 3 (i.e. the operating point desired or set by reference torque Γref and/or reference rotation Ωref) as a function of the position of rotor 9, the predefined parameters (Γref and/or Ωref), and the electrical feedback data.
In accordance with the invention, control device also includes estimation means 23, a malfunction detector 25 and a transition switch 27.
Estimation means 23 are configured to determine an estimated position {circumflex over (θ)} of rotor 7 in estimated Park reference frame δ-γ. As will be seen in greater detail below with reference to of estimated rotating frame of reference δ-γ such that it is synchronised with frame of reference d-q associated with the rotor (see also
Malfunction detector 25 is configured to detect a possible malfunction of sensor 15. In particular, malfunction detector 25 can, for example, consist of a malfunction signal which is generated or delivered by sensor 15 itself when it malfunctions.
Switch 27 is configured to connect control means 19 either to estimation means 23, or to position sensor 15, depending on whether malfunction signal S does or does not indicate that sensor 15 is malfunctioning.
More specifically, while the malfunction detector does not indicate any malfunction of position sensor 15, switch 27 maintains the connection between control means 19 and position sensor 15, in order that control means 19 receive measured position θm of rotor 7. Conversely, when the malfunction detector indicates that position sensor 15 is malfunctioning, switch 27 then connects control means 19 to estimation means 23, in order that control means 19 receive estimated position {circumflex over (θ)} of rotor 7.
Thus, as soon as sensor 15 malfunctions, switch allows a transition from control using sensors to sensorless control of PMSM 3. This enables the availability of PMSM 3 in degraded mode to be increased. Naturally, as soon as position sensor 15 is repaired PMSM 3 can once again be controlled with position sensor 15.
It will be noted that
This diagram shows that estimation means 23 include an electromotive force estimator 31, a speed estimator 33, and an integrator 35. In addition, control means 19 include a torque-current converter 37 and current regulator 39 in addition to transformation interface 21.
Torque-current converter 37 transforms the value of reference torque Γref into the corresponding reference currents iδref, iγref in estimated Park frame of reference δ-γ.
In addition, transformation interface 21 transforms stator currents ia, ib, ic measured by electrical measurement means 17 into components of currents iδ, iγ in Park frame of reference δ-γ.
In addition, current regulator 39 receives reference currents iδref, iγref from torque-current converter 37 and current components iδ, iγ in frame of reference δ-γ from transformation interface 21 to determine the components of stator voltages νδ, νγ in frame of reference δ-γ corresponding to the reference voltages of inverter 11. Transformation interface 21 receives these components of stator voltages νδ, νγ according to the two-phase model, and transforms them into reference voltages ν′a, ν′b, ν′c of inverter 11 according to the triphase model.
Sensorless vector control consists in estimating angular speed Ωc such that difference of position φ between and θ is cancelled out (see
However, given that component eδ of the emf in axis δ tends towards zero when φ tends towards zero (eδ=pψfΩ sin φ), constant maintenance of position difference φ at zero can be replaced by constant maintenance of eδ at zero.
This estimate consists in resolving the following electrical equations in estimated frame of reference δ-γ:
Consequently, electromotive force estimator 31 receives the components of currents iδ, iγ from transformation interface 21, the components of stator voltages νδ, νγ from current regulator 39, and speed of rotation Ωc of the rotor from speed estimator 33, in order to estimate, as a function of these magnitudes, components êδ, êγ of the emf in estimated frame of reference δ-γ.
Speed of rotation Ωc of the rotor is estimated in a closed loop by speed estimator 33 according to estimates êδ, êγ of the emf determined by electromotive force estimator 31, and by maintaining component êδ constantly at zero. Speed of rotation Ωc of the rotor is naturally initialized by a predetermined initial value Ωc0.
Advantageously, speed estimator 31 uses a non-linear corrector to determine rotational speed Ωc according to a control law having an area of global convergence including a single point of asymptotically stable equilibrium in the Lyapunov sense. This point of equilibrium is equal to the operating set point of MSAF 3.
According to this example, the functional diagram of speed estimator 33 includes a first speed estimator 43, a comparator 45, first and second sign indicators 47 and 49, an adder 51, and a non-linear corrector 53.
The purpose of comparator 45 is to compare component êδ with its reference component eδref=0. The purpose of first speed estimator 43 is to determine a prior estimate {circumflex over (Ω)} of the rotational speed according to estimated component êγ. The purpose of first sign indicator 47 is to indicate the sign of prior estimate {circumflex over (Ω)} of the rotational speed, if it is assumed that sign({circumflex over (Ω)})=sign({circumflex over (Ω)}ref), where Ωref is the set rotational speed. The purpose of second sign indicator 49 is to indicate the sign of component ê6. The purpose of non-linear corrector 53 is to introduce non-linear terms in order to make all undesired points of convergence of the control law unstable, or to prevent convergence to any undesired solutions. Finally, the purpose of adder 51 is to add the non-linear terms to prior estimate {circumflex over (Ω)} in order to determine rotational speed {circumflex over (Ω)}c.
First speed estimator 43 calculates the quotient between component of emf êγ associated with axis γ and a predetermined physical parameter Kf depending on the characteristics of the rotor's permanent magnets, using the following formula:
where Kf=pψf.
According to a particular embodiment, non-linear corrector 53 introduces a corrective term which is a function of sign sign({circumflex over (Ω)}) of prior estimate {circumflex over (Ω)} of the rotational speed, of a predetermined operating parameter b, physical parameter Kf, component êδ of the emf associated with axis δ, and finally a non-linear factor which depends on the sign of component êδ and a predetermined coefficient ξ using the following formula:
and K=1−ξ·sign(êδ).
Adder 51 then adds the above corrective term to prior estimate {circumflex over (Ω)} in order to determine rotational speed Ωc using the following formula:
It will be noted that by analysing the stability of the control law expressed by rotational speed Ωc using the above formula (see the publication of Babak Nahid-Mobarakeh et al., “Analysis of convergence of sensorless control laws of PMSMs based on estimation of the emf,”), it can be seen that all the trajectories in phase space φ-Ω converge towards the desired point of equilibrium (φ=0, Ω=Ωref) for the following conditions:
0<ξ<1 and
Operational parameter b is advantageously between 0 and 3 (0<b≦3) and preferably close to 1.
The above control law enables all trajectories in phase space to be prevented from converging towards all undesired points of equilibrium by making certain points of equilibrium unstable, and moving the other points in phase space far enough away to prevent them. This in particular enables the problem of non-observability inherent to electrical equations at a rotational speed close to zero to be overcome.
Moreover, the dependence of the corrective term on the sign of rotational speed sign({circumflex over (Ω)}) enables the trajectories in phase space φ-Ω to converge towards the desired point, whatever the sign of set rotational speed Ωref, allowing speed reversal without any problem.
Thus, with the above conditions, whatever the initial coordinate point (−π≦φ≦π, Ω=Ω0), all trajectories in phase space φ-Ω converge towards the desired point of equilibrium.
In other words, even if at the start the initial position error is of the order of π, the trajectory rapidly converges towards the operating point using the set torque and rotational speed values.
In addition, even if at start-up the initial point has a rotational speed of the sign opposite the set speed, the position error rapidly converges towards zero, enabling the PMSM rapidly to attain a permanent speed equal to the set torque and rotational speed values.
When the rotational speed has been determined by speed estimator 33, integrator 35 integrates rotational speed Ωc from speed estimator 33 to determine estimated rotor position {circumflex over (θ)}.
Furthermore, in order to enable uniform and accurate transition between control with sensors and sensorless control, estimation means 23 can include means to adjust estimated rotor position {circumflex over (θ)} continuously.
Adjustment means 61 can thus include a position comparator 63 to compare rapidly rotor position θm measured by position sensor 15 with estimated rotor position {circumflex over (θ)} from integrator 35, a PI filter or gain multiplier 65 to accomplish a counter-reaction in order that the integration does not diverge and an additional integrator 67 to synchronise estimated rotor position {circumflex over (θ)} with measured rotor position {circumflex over (θ)}m and a second comparator 69 between the output of additional integrator 67 and estimated rotor position {circumflex over (θ)} to correct {circumflex over (θ)}c the rotor position.
It will be noted that the comparison and the counter-reaction occur continuously to prevent estimated value {circumflex over (θ)} of the rotor position from diverging since, while position sensor 15 is in use, estimated value {circumflex over (θ)} would be in an open loop. When a malfunction of position sensor 15 is detected the last estimated value of rotational speed Ωc from speed estimator 33 at that time is injected in integrator 35 at the moment when the control is switched over.
Consequently, if a switchover does indeed occur, between control with sensors and sensorless control, the difference between the last measured value {circumflex over (θ)}m and estimated value {circumflex over (θ)} is advantageously very small.
After the transition has occurred there is no requirement for the counter-reaction as the value of position sensor 15 is incorrect.
Indeed,
Initialisation means 73 include a memory to record the last value of the rotor position (estimated {circumflex over (θ)} or corrected {circumflex over (θ)}c) which is still correct.
Indeed, when a malfunction is detected malfunction signal S activates a trigger on, for example, a falling signal which results in the integration being reinitialised by the last estimated rotor position {circumflex over (θ)}o.
Thus, at the time of the switch between control with sensors and sensorless control the torque at the time of the transient state between the two forms of control has no oscillations.
It will be noted that the different elements of the control device can include processing or calculation means having one or more computer programs including code instructions for implementing the control method according to the invention when the computer program or programs are executed by these various elements.
Consequently, the invention also covers a computer program product which can be implemented in the different elements of the control device, where this program includes code instructions able to implement a method according to the invention as described above.
The system including the PMSM and control of it according to the invention can advantageously be used in actuator motors in on-board aircraft systems. As an example, it can be used in the compressor, ventilation system, thrust reversers, doors and many other types of aircraft equipment.
| Number | Date | Country | Kind |
|---|---|---|---|
| 09 58879 | Dec 2009 | FR | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP2010/069406 | 12/10/2010 | WO | 00 | 7/20/2012 |
| Publishing Document | Publishing Date | Country | Kind |
|---|---|---|---|
| WO2011/070165 | 6/16/2011 | WO | A |
| Number | Name | Date | Kind |
|---|---|---|---|
| 4777422 | Slicker et al. | Oct 1988 | A |
| 5969499 | Shaffer | Oct 1999 | A |
| 6172498 | Schmidt et al. | Jan 2001 | B1 |
| 7002318 | Schulz et al. | Feb 2006 | B1 |
| 7492118 | Burgstahler et al. | Feb 2009 | B2 |
| 7888893 | Nagura et al. | Feb 2011 | B2 |
| 8248039 | Cao et al. | Aug 2012 | B2 |
| 8253365 | Yeh | Aug 2012 | B2 |
| 20050140324 | Kunzel et al. | Jun 2005 | A1 |
| 20060043921 | Nagura et al. | Mar 2006 | A1 |
| 20080106230 | Burgstahler et al. | May 2008 | A1 |
| 20100181945 | Nagura et al. | Jul 2010 | A1 |
| 20100327585 | Cao et al. | Dec 2010 | A1 |
| 20110089882 | Yeh | Apr 2011 | A1 |
| Number | Date | Country |
|---|---|---|
| 2001-112282 | Apr 2001 | JP |
| Entry |
|---|
| International Search Report and Written Opinion issued Jul. 7, 2011 in PCT/EP2010/069406 with English language translation. |
| Babak Nahid-Mobarakeh, et al., “Analyse de convergence des lois de commande sans capteur des MSAP fondées sur l'estimation de la fém”, RS-RIGE, vol. 6, No. 5, 2003, pp. 547-577 with partial English translation. |
| Number | Date | Country | |
|---|---|---|---|
| 20120280641 A1 | Nov 2012 | US |
