Method for determining a blocked rotor in an injection-signal-based sensorless operation of a synchronous machine, control unit, synchronous machine and vehicle

Abstract

The invention relates to a method for determining a blocked rotor during injection-signal-based sensorless operation of a synchronous machine, comprising the following steps: Applying a three-phase alternating voltage, which includes an injection signal, to stator windings of the synchronous machine according to an estimated rotation angle φ est ; Detect a three-phase current flowing in the stator windings in response to the application of the alternating voltage; Transform the detected current into a first current vector that corresponds to a first angle φ est +45° in a dq coordinate system, and a second current vector corresponding to a second angle φ est-45° in a dq coordinate system; determining the contributions of the injection signal to the lengths of the first and second current vectors; determining the difference between these contributions as an error signal; and determining the amplitude of the error signal. The method also includes the following steps: comparing the amplitude of the error signal with a predetermined threshold; and determining the blocked rotor if the amplitude of the error signal exceeds the predetermined threshold.

Description

The present invention relates to a method for determining a blocked rotor, i.e. a slipping of the rotor, during an injection-signal-based sensorless operation of a synchronous machine, in particular a permanently excited synchronous machine. The present invention also relates to a control unit which is designed and programmed to carry out the method, to a synchronous machine which has the control unit, and to a vehicle which has the synchronous machine.In sensorless or sensorless operation, i.e. sensorless control / regulation, of a synchronous machine, such as a permanently excited synchronous machine, a rotation angle sensor with which the current rotation angle of a rotor of the synchronous machine is detected is dispensed with. Instead, currents and detected or estimated voltages detected in stator windings are used to infer the rotation angle and the rotational speed of the rotor of the synchronous machine via a model of the synchronous machine. Below a speed threshold of the absolute speed, it is necessary to feed in a so-called injection signal in order to assist the identification of the angle of rotation and the speed in this speed range. Starting with a stationary rotor, the current rotation angle can be determined by an initialization routine. With this injection method, the synchronous machine can be operated in the lower speed range until a switchover into a model-based sensorless algorithm takes place.The initial angle of rotation can be determined by presetting an injection signal, i.e. a high-frequency oscillation in the d- and q-voltages, for a specific number of points on a voltage circuit path, in that the exciting voltage amplitude leads into a resulting current amplitude. In the case of circular excitation, the d-q coordinates must produce an ellipse in the d-q coordinate system due to an anisotropy of the synchronous machine in the current as a result of the d-q coordinates. The major axis of the ellipse corresponds to the d-axis or d-direction. This direction of the main axis thus describes the initial value of the angle of rotation.In this case, a longer major axis of an ellipse, which consequently corresponds to the current curve in the d-q coordinate system, can be found by an iterative controller approach. By applying the injection signal to the three-phase alternating voltage present at the stator windings, a three-phase current can be measured in the stator windings. The current is then, however, not transformed into the imaginary d-direction, but into an imaginary d+45°-direction and an imaginary d-45°-direction. This makes it possible to achieve significantly greater amplitudes in the current in comparison with an evaluation of the currents in the d- and q-directions, since the current in the q-direction disappears when the real d-axis is found. The detection thus becomes more robust to disturbances and especially in dynamic situations. Then, the amplitudes corresponding to the injection signal are determined and the two amplitudes for +45° and -45° are compared with each other. The estimated rotation angle is then corrected such that the two amplitudes become equal for +45° and -45°. The sign of the correction is selected such that the longer half axis, i.e. the d axis, is found. If the sign is reversed during the correction, the shorter half axis, i.e. the q axis, is found. With this approach, not only can the initial rotation angle be found, but the synchronous machine can be operated with the found rotation angle.Moreover, in a time-sequential sequence, the shorter and then the longer half axis of the ellipse can be determined first. The two half axes found are then set in relation to one another, wherein an angle between the half axes is intended to move in a range of 90°±ε°. Epsilon is set to a few degrees. In practice, the longer half axis is then entirely used to operate the system sensor-free.Similarly, the shorter and longer half axes of the ellipse can be observed by using two different injection frequencies sequentially to find the angles for the shorter and longer half axes.In the injection-signal-based approach for initializing the rotation angle, the anisotropy of the synchronous machine is used, as described above. The anisotropy of a synchronous machine, in particular a permanently excited synchronous machine, refers to the differences in the inductances in the d- and q-direction, which ultimately leads to the ellipsoidal current curve in the d-q coordinate system.For robust and stable operation, the injection-based method for sensorless operation of a synchronous machine requires pronounced anisotropy. In the case of magnets bonded to the rotor, in contrast to buried magnets, the inductances Ld and Lq are more similar to one another, and therefore the anisotropy is consequently smaller. The ellipse is then less pronounced and consequently more circular.In particular, smaller synchronous machines, such as are used in pump drives, for example, have an often less pronounced anisotropy because of the use of bonded magnets. Especially in the case of demanding maneuvers, such as, for example, sharp torque jumps or sudden changes in rotational speed, the synchronous machine, which is often designed to be limit-limiting for the application, can thus get out of step in the worst case and / or even slip completely in the case of a spontaneous blockage. A sensorless control or regulating algorithm usually has model-based measures for smoothing the determined rotational angle and rotational speed signals by prediction. This may result in the control algorithm seeing a rotating rotor while the rotor is in reality locked.It is therefore the object of the present invention to provide a suitable measure which makes it possible to determine a blocked rotor, i.e. a slipping of the rotor, during an injection-signal-based sensorless operation of a synchronous machine.The object is achieved by the subject matters having the features according to the independent claims. Advantageous embodiments are the subject of the dependent claims.A method according to the invention for determining a blocked or slipping rotor during an injection-signal-based sensorless operation of a synchronous machine, in particular a permanently excited synchronous machine, has the following steps: First, a step for applying a three-phase alternating voltage, which comprises an injection signal, to stator windings of the synchronous machine in accordance with an estimated rotation angle φ est. takes place. That is, a high-frequency injection signal is superimposed on the three-phase alternating voltage. Thereafter, a step of detecting a three-phase current flowing in the stator windings in response to the application of the AC voltage is performed. Then, a step of transforming the detected current into a first current vector corresponding to a first angle φ est+45 ° in a d-q coordinate system and a second current vector corresponding to a second angle φ est-45 ° in a d-q coordinate system follows. Consequently, the detected currents are transformed not to the d- and q-axes, but to directions deviating by ±45° from the d-axis. A step for determining a portion of a length of the first current vector caused by the injection signal and a step for determining a portion of a length of the second current vector caused by the injection signal are then carried out. Consequently, it is determined which length component of the first and the second current vector is generated by the superimposed injection signal. When these length components are determined, a step of determining a difference between the component of the length of the first current vector and the component of the length of the second current vector as an error signal follows. This means that the error signal corresponds to the difference of the above-determined length components of the two current vectors. Further, a step of determining an amplitude of the error signal is then carried out. There then follows a step for comparing the amplitude of the error signal with a predetermined threshold value and a step for determining the locked rotor if the amplitude of the error signal exceeds the predetermined threshold value.Consequently, a determination of the locked rotor is made based on the error signal. The error signal arises because the ellipse rotates about the d-q coordinate system with the rotor and the actual, i.e. fixed, rotation angle of the rotor no longer matches the estimated, i.e. rotating, rotation angle. The error signal oscillates at twice the estimated rotational speed, since the actual rotational angle and the estimated rotational angle come to lie twice on top of one another per revolution of the ellipse.Since the two current vectors are transformed to the two angles φ est+45 ° and φ est-45 °, a pronounced error signal can also be obtained in the case of a less pronounced ellipse, i.e. in the case of a less pronounced anisotropy in the synchronous machine. In addition, by setting the threshold value to a predetermined value, an allowable angular deviation between the actual rotation angle and the estimated rotation angle can be predetermined.It is conceivable that the blocked rotor can be determined if the threshold value is exceeded by the amplitude of the error signal a predefined number of times. Accordingly, the rotor-fixed d-q coordinate system was rounded over by the ellipse for a half of the number of times. Consequently, the locked rotor can be reliably detected.The portions of the lengths of the two current vectors caused by the injection signal can preferably be determined by means of a band pass filter. Consequently, the high frequency components generated by the injection signal can be reliably detected.Advantageously, the amplitude of the error signal can be determined using a PLL (Phase-Locked Loop) filter. As already mentioned, the error signal oscillates at twice the estimated speed, so that a PLL approach is well suited for filtering the amplitude from the error signal.The synchronous machine can be designed to drive a pump. As mentioned at the beginning, synchronous machines for driving pumps are often designed to be tight in terms of output, so that slipping of the rotor can occur in the event of steep changes in torque and / or rotational speed. The method according to the invention is therefore particularly suitable for synchronous machines of this type.Furthermore, the synchronous machine can be a permanently excited synchronous machine in which magnets are bonded to the rotor. In this type of synchronous machine, the anisotropy is less pronounced because of the magnets bonded on. By transforming the current vectors to the two angles φ est+45 ° and φ est-45 °, a pronounced error signal can nevertheless be determined and consequently a reliable determination of the blocked rotor can take place.Preferably, a binary determination signal may be set to 1 to indicate a locked rotor. The binary determination signal indicates whether or not the locked rotor is present. Thus, at a value of 0, the determination signal indicates that the rotor is rotating. At a value of 1, it is indicated that the rotor is locked or slips. The determination signal may also be output to cause respective control processings. The determination signal is set to 0 again if the amplitude of the error signal is below the threshold value again.Advantageously, re-initialization of the rotation angle can be performed when a locked rotor is determined. In order to determine the determination of the blocked rotor, the determination signal can be used with advantage. Accordingly, the re-initialization may be implemented as an interrupt routine triggered in response to the determination signal set to 1 in a controller of the synchronous machine.The present invention also relates to a control unit which is designed and programmed to carry out the method according to one of the above aspects. The control unit is therefore designed to acquire measured values required for carrying out the method and to output control commands to the synchronous machine. Consequently, the above-mentioned advantages can be achieved by the control unit.The present invention also relates to a synchronous machine comprising a stator and a rotor rotatable relative to the stator. The synchronous machine also has the control unit according to the above aspect. In particular, the synchronous machine is a permanently excited synchronous machine. The synchronous machine can be used here for driving a pump. In addition, magnets can be bonded to the rotor of the synchronous machine. Since the synchronous machine has the control unit according to the invention which is designed and programmed to carry out the method according to the invention, the above-mentioned advantages can be obtained.A vehicle according to the invention has the synchronous machine according to the above aspect. The synchronous machine can be used in the vehicle for driving a pump. The magnets can be bonded to the rotor of the synchronous machine. By implementing the control unit according to the invention, i.e. the method according to the invention, reliable detection of the slipping of the rotor can consequently be detected.The present invention will be described in detail below with reference to the figures. The following are shown: FIG. 1 shows time diagrams which show simulated profiles of measured values when using the method according to the invention; and FIG. 2 shows time diagrams which show measured curves of measured values when using the method according to the invention.Hereinafter, an embodiment of the present invention will be described with reference to the figures.First, the determination of an error signal, on the basis of which a determination of a blocked rotor takes place, will be described. First, a three-phase alternating voltage is applied to a stator of a synchronous machine, in particular a permanently excited synchronous machine. The three-phase alternating voltage is applied on the basis of an estimated angle of rotation φ est, which is estimated by a sensor- or sensorless control algorithm. The three-phase alternating voltage comprises an injection signal which is a high-frequency signal superimposed on the three-phase alternating voltage. Consequently, in the stator windings, a three-phase current having a portion caused by the injection signal is obtained.The three-phase current is measured and transformed into a first current vector corresponding to a first angle φ est+45 ° in a d-q coordinate system and a second current vector corresponding to a second angle φ est-45 ° in a d-q coordinate system. It is assumed here that the d-q coordinate system rotates est with the estimated rotation angle φ.Subsequently, a portion of a length of the first current vector caused by the injection signal and a portion of a length of the second current vector caused by the injection signal are determined using a band pass filter. From this, a difference between the proportion of the length of the first current vector and the proportion of the length of the second current vector is determined as an error signal. Furthermore, an amplitude of the error signal is then determined using a PLL (phase-lock loop) filter.When the rotor is blocked or slipping, the elliptical current curve present in the synchronous machine due to an anisotropy now rotates about the rotor-fixed d-q coordinate system at the estimated rotational speed. Due to the rotation, the two portions differ in the lengths of the current vectors, so that the error signal occurs. Since two superimpositions of the ellipse with the rotor-fixed d-q coordinate system occur per round, the error signal oscillates with twice the value of the estimated rotational speed. This situation is to be illustrated below on the basis of simulated and measured time diagrams.FIG. 1 shows time diagrams which show simulated profiles of measured values when using the method according to the invention.The curve of a measured angle of rotation (reference sign 1) and the curve of an angle of rotation estimated by a sensorless control algorithm (reference sign 2) in a range from 0 to 2π are shown in an uppermost diagram. It can be seen that the curves are divided into two. Initially, the rotor slips, whereby the estimated rotation angle 2 no longer matches the actual rotation angle 1. The rotation angle 2 estimated by the sensorless algorithm continues to rotate at a constant speed. However, the rotor may not follow a rotating field generated by applying the three-phase AC voltage having the estimated rotation angle to the stator windings. Due to the constant skipping of the rotating field over the rotor, the latter performs a sinusoidal movement, since the rotor is briefly accelerated and then decelerated again at each skipping. After a re-initialization and a short transient process, the estimated rotational angle 2 and the actual rotational angle 1 then agree again from approximately 0.58 s.The second diagram shows a profile of an actual rotational speed (reference sign 3) and a profile of an estimated rotational speed (reference signs 4). The estimated rotational speed 4 continues to assume a constant rotational speed of 200 1 / min. However, the actual speed 3 oscillates sinusoidally about zero speed due to the skipping of the rotating field across the rotor before reinitializing and can be accelerated to a higher speed of 300 1 / min.In the third diagram, the error signal (reference numeral 5) is shown, which clearly oscillates sinusoidally about a value of zero. The frequency of the error signal 5 is twice as high as the estimated rotational speed 4 and the sinusoidal oscillation in the actual rotational speed 3, since there are two solutions per round-trip of the blocked rotor. Thus, the error signal 5 can also be referred to as a second-order oscillation.The third diagram also shows the amplitude of the error signal (reference numeral 6). For determining the amplitude 6 of the error signal, a PLL filter, for example, is suitable, which finds the second-order amplitude 6 from the error signal 5. It can be seen in the third diagram that the error signal 5 and consequently also the amplitude 6 return to a value of zero a short time after the reset. Accordingly, rotation of the rotor coincides with rotation of the ellipse.The bottom diagram shows a profile of a binary determination signal (reference numeral 7). It can be seen that the determination signal 7 is already set to 1 at the beginning on the basis of an excess of a threshold value, not shown, by the amplitude of the error signal. At approximately 0.55 s, a re-initialization then takes place by reducing the estimated rotational speed 2 (see second diagram). The rotor can thus be captured again and then accelerated to a desired rotational speed. After a short settling time, the actual rotation angle 1 and the estimated rotation angle 2 again coincide (see first diagram). Consequently, the actual rotational speed 3 and the estimated rotational speed 4 also correspond again. Since the amplitude 6 of the error signal is then below the threshold value, the determination signal 7 is set to 0 again.FIG. 2 shows time diagrams which show measured curves of measured values when using the method according to the invention.In the bottom diagram, the actual rotation angle (reference sign 10) is shown and it can be clearly recognized that the rotor is blocking twice. During this time, a rotation of the rotor is still assumed for the rotation angle estimated by the sensorless algorithm (reference sign 11) on the basis of internal model-based predictions.The error signal (reference numeral 8) is shown in the top diagram. The calculated amplitude of the error signal (reference numeral 9) is shown in the second diagram. It can clearly be seen that the value of the second-order amplitude rises steeply when the rotor is blocked. Depending on the selection of the threshold value (not shown), the determination signal can signal the blockage of the rotor already after a short time.By using the method according to the invention, a blockage or slipping of a rotor can consequently be reliably determined. Then, an appropriate countermeasure such as re-initialization of the rotation angle may be taken.The present invention is further implementable in a controller. The control unit is then designed and programmed to execute the method according to the invention. The control unit is therefore designed to acquire measured values required for carrying out the method and to output control commands to the synchronous machine.The control unit can be used for operating a synchronous machine which has a stator and a rotor which can be rotated relative to the stator. In particular, the synchronous machine can be a permanently excited synchronous machine. The synchronous machine can be used here for driving a pump. In addition, magnets can be bonded to the rotor of the synchronous machine.The synchronous machine can be used in a vehicle. The synchronous machine can be used in the vehicle for driving a pump, and the magnets can be bonded to the rotor of the synchronous machine.List of reference characters1 Actual rotation angle 2 Estimated rotation angle 3 Actual rotation speed 4 Estimated rotation speed 5 Error signal 6 Amplitude of the error signal 7 Determination signal 8 Error signal 9 Amplitude of the error signal 10 Actual rotation speed 11 Estimated rotation speed

Claims

Method for determining a blocked rotor during an injection signal-based sensorless operation of a synchronous machine, in particular a permanently excited synchronous machine, comprising the following steps: applying a three-phase alternating voltage comprising an injection signal to stator windings of the synchronous machine according to an estimated rotation angle φ est, detecting a three-phase current flowing in the stator windings in response to the application of the alternating voltage, transforming the detected current into a first current vector corresponding to a first angle φ est+45 ° in a d-q coordinate system and a second current vector corresponding to a second angle φ est-45 ° in a d-q coordinate system, determining a portion of a length of the first current vector caused by the injection signal, determining a portion of a length of the second current vector caused by the injection signal, determining a difference between the portion of the length of the first current vector and the portion of the length of the second current vector as an error signal, and determining an amplitude of the error signal, characterized in that the method comprises the following steps: comparing the amplitude of the error signal with a predefined threshold value, and determining the blocked rotor if the amplitude of the error signal exceeds the predefined threshold value.Method according to Claim 1, wherein the portions of the lengths of the two current vectors caused in each case by the injection signal are determined by means of a band-pass filter.The method of any of claims 1 or 2, wherein the amplitude of the error signal is determined using a PLL filter.Method according to one of Claims 1 to 3, wherein the synchronous machine is designed to drive a pump.Method according to one of Claims 1 to 4, wherein the synchronous machine is a permanently excited synchronous machine in which magnets are adhesively bonded to the rotor.The method of any of claims 1 to 5, wherein a binary determination signal is set to 1 to indicate a stuck rotor.The method of any one of claims 1 to 6, wherein a re-initialization of the rotation angle is performed when a locked rotor is determinedA control unit configured and programmed to perform the method according to any one of claims 1 to 7.Synchronous machine, in particular a permanently excited synchronous machine, which has a stator and a rotor which can be rotated relative to the stator, characterized bythe control unit according to Claim 8.A vehicle comprising the synchronous machine according to claim 9.

Images