Method for determining rotor position of an electric machine - Patent Application 20070122997

Abstract

A novel method for determining the rotor position of an electric machine is provided. [Solution] The present invention relates to a method for determining the rotor position of an electric machine without using a rotor position sensor, in which an anisotropic position control and / or position adjustment algorithm using high-frequency excitation of voltage modulation is applied to estimate the rotor angle, and a measurement noise-affected signal with a minimum disturbance gradient tangent is calculated using a least-squares algorithm (6) that can identify high-frequency current derivatives, and an angle control loop (2) is applied to minimize the generated disturbances and jumps in the rotor angle, and a high-frequency injection with optimized amplitude is applied.

Description

[Technical Field]

[0001] The invention relates to a method for determining the position of a rotor of an electric machine according to the preamble of claim 1 . [Background technology]

[0002] The operation of electric machines, e.g., radially or axially arranged permanently excited synchronous machines or current-excited synchronous machines, is achieved by controlling the current in the electric machine by regulating the voltage. A well-known concept for this current control is field-oriented control, which transforms the current and voltage into a rotor-fixed coordinate system (Clarke and Park transformations) based on the position of the permanent magnets. The position of the permanent magnets (also known as the rotor angle) is then usually determined by a rotor angle sensor (e.g., resolver). Sensorless control concepts offer the possibility to omit the sensor and determine the rotor angle using only data from other sensor parameters and machine parameters, in order to save costs and installation space.

[0003] In this case, a well-known possibility for determining the rotor position is to use the anisotropy of the electrical machine. L The degree of rotation is expressed in the rotor-fixed coordinate system L D and L Q It is determined by the inductance of the inverter and varies depending on the operating point due to the effects of magnetic saturation, etc.

[0004]

number

[0005] For anisotropic rotor angle determination, numerous possibilities for injecting an additional signal are known from the literature. This signal can be implemented as a sine wave or a square wave and can be applied as a voltage injection in the rotor-fixed or stator-fixed reference frame. Furthermore, the possibility of proposing a current injection in the rotor-fixed reference frame is also known. All these methods have in common that the injection imposes additional high-frequency losses as well as additional torque ripple on the electric machine, which has a negative impact on the efficiency and noise, vibration and harshness behavior (NHV). The goal is therefore to make the rotor angle estimation better and more robust, while at the same time keeping the injection as small as possible.

[0006] Besides common high-frequency injection, it is known from the prior art that in addition to the original high-frequency injection, voltage modulation with the switching frequency can also be used directly as high-frequency injection. The problem with this method is to accurately identify the current derivative at a specific pointer of the switching period. In the academic literature, this identification is mainly ensured by additional prototype sensor systems, which are not available in the automotive environment. If the derivative can be identified with a certain degree of accuracy, there are various possibilities to extract the rotor angle dependent term. A well-known method is the so-called arbitrary injection (see Non-Patent Document 1 below). This involves extracting the entire current derivative, together with the anisotropic high-frequency part e αβ , isotropic high frequency part

[0007]

number

[0008]

number

[0009]

number

[0010] where R S is the stator resistance,

[0011]

number

[0012]

number

[0013] Subsequently, an angle control loop, also called an angle tracking observer (ATO) (see Non-Patent Document 2 below), is used to minimize the occurrence of oscillations in the rotor angle. The effect of cross-coupling is calculated by a table that depends on the id and iq currents, and the inductances in the d and q directions, and the cross-coupled inductance L obtained from the characteristic data. dq (i d ,i q ) and the compensation term is calculated as follows:

[0014]

number

[0015] Furthermore, according to Non-Patent Document 3 below, the differential coefficient of the high frequency current can be identified by a least squares (LSQ) algorithm. The basic idea here is that the current curve is expressed by a linear equation

[0016]

number

[0017]

number

[0018] By solving the linear regression equation and simplifying the equation by transforming it using the arithmetic mean of the incremental indices and squared pyramid numbers, the equation can be easily implemented, for example, on a freely programmable logic gate array (FPGA).

[0019]

number

[0020] where:

[0021]

number

[0022] Patent Document 1 below describes a method for determining the rotor position of an electric rotating machine powered by a PWM-controlled inverter. The injected voltage determined by the control device is composed of a predetermined voltage and a high-frequency voltage and converted into a corresponding PWM duty cycle, and the inverter is appropriately controlled by these PWM duty cycles. Next, the current curves of the phase currents are detected by measuring at least a first phase current and a second phase current, and this measurement is performed during the last third of each passive switching state in one PWM period. Then, the rotor position is determined according to the detected current curves and the supplied high-frequency voltage. [Prior art documents] [Patent documents]

[0023] [Patent Document 1] German Patent Application Publication No. 10 2019 208 497 A1 [Non-patent literature]

[0024] [Non-Patent Document 1] D. Paulus, P. Landsmann and R. Kennel, "Saliency based sensorless field-oriented control for permanent magnet synchronous machines in the whole speed range", 3rd IEEE International Symposium on Sensorless Control for Electric Drives, 2012 [Non-patent document 2] J. Friedmann, R. Hoffmann and R. Kennel, “A new approach for a complete and ultrafast analysis of PMSMs using the arbitrary injection scheme”, IEEE Symposium on Sensorless Control for Electrical Drives, 2016 [Non-patent document 3] A. Liske, S. Decker and M. Braun, “Resource optimal FPGA implementation of a Least-Squares-Estimator for fast and robust online measurement of current slope and absolute value”, 21th European Conference on Power Electronics and Applications, 2019 [Non-patent document 4] P.Landsmann, J.Jung, M.Kramkowski, P.Stolze, D.Paulus and R.Kennel, "Lowering injection amplitude in sensorless control by means of current oversampling", 3rd IEEE International Symposium on Sensorless Control for Electric Drives, 2012 Summary of the Invention [Problem to be solved by the invention]

[0025] The present invention aims to provide a novel method for determining the position of a rotor of an electric machine. [Means for solving the problem]

[0026] This problem is solved according to the invention by a method for determining the position of a rotor of an electric machine having the features of claim 1.

[0027] Advantageous embodiments of the invention are the subject of the dependent claims.

[0028] In a method according to the present invention for determining the rotor position of an electric machine without a rotor position sensor, an anisotropic position control and / or position adjustment algorithm is applied using a voltage-modulated high-frequency excitation to estimate the rotor angle. Furthermore, according to the present invention, a measurement noise-affected signal with a minimum disturbance gradient tangent is calculated using a least-squares algorithm capable of identifying high-frequency current derivatives, and an angle control loop is applied to minimize the disturbances and rotor angle jumps, and a high-frequency injection with optimized amplitude is applied.

[0029] The present invention provides a method for operating and monitoring a permanently excited synchronous machine without a rotor position sensor, which enables anisotropy-based rotor position estimation optimized for NVH (Noise, Vibration, and Harshness) and efficiency without a rotor position sensor.

[0030] The present invention combines methods for determining the rotor position of an electric machine without using sensors. In particular, a voltage-modulated high-frequency excitation by arbitrary injection is used, the signal affected by measurement noise is calculated on a high-speed computing unit using an LSQ algorithm, and / or disturbances and jumps in the rotor angle resulting from the angle control loop are minimized. By combining these aspects and additionally incorporating a high-frequency injection with optimized amplitude, it is possible to determine the rotor angle anisotropically with reasonable accuracy and robustness until it stops.

[0031] According to the present invention, an anisotropy-based position control and / or position adjustment algorithm is applied with a least-squares algorithm capable of identifying high-frequency current derivatives using modulated excitation without sensors. Additionally, an optimized injection scheme is proposed to ensure stable and robust estimation even at low motor speeds and motor stalls. The combination of all three features allows the algorithm to be used in automotive motor control units.

[0032] Eliminating the rotor position sensor offers advantages in terms of cost, mounting space, and diagnostic effort.

[0033] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. [Brief explanation of the drawings]

[0034] [Figure 1] FIG. 1 is a schematic diagram of a rotor position estimation method for feedback to field-oriented control. [Figure 2] FIG. 1 is a schematic diagram of a rotor position estimation method for redundantly monitoring rotor position. [Figure 3] 10 is a schematic diagram of a characteristic map for selecting the frequency and amplitude of the injection depending on the operating point; [Figure 4] FIG. 2 is a schematic diagram of voltages and currents of an inverter for operating an electric machine. [Figure 5] FIG. 1 is a schematic diagram of an angle control loop. DETAILED DESCRIPTION OF THE INVENTION

[0035] In all the drawings, the same reference numerals are used to designate corresponding parts.

[0036] As explained in the prior art, there is a possibility to estimate the rotor angle based on the anisotropy by using a voltage-modulated high-frequency excitation. A known method here is arbitrary injection. Furthermore, it is known that signals affected by measurement noise can be calculated on a high-speed computing unit (e.g., FPGA) using the LSQ algorithm6, and that the minimum disturbance gradient tangent can be calculated. Finally, the concept of an angle control loop2 is known, which minimizes the disturbances and rotor angle jumps that occur. The essence of the present invention is the implementation of a novel combination of individual aspects not previously included in the prior art, additionally incorporating an amplitude-optimized high-frequency injection, which allows the rotor angle to be determined based on the anisotropy with reasonable accuracy and robustness until it stops. For this purpose, a distinction is made between case 1: rotor position estimation for feedback to field-oriented control (Fig. 1) and case 2: rotor position estimation for redundant rotor position monitoring (Fig. 2). Depending on the requirements of the electric drive and other conditions, either one of the two methods or a combination of the methods can be used.

[0037] Figure 1 is a schematic diagram of a rotor position estimation method for feedback to field-oriented control, and Figure 2 is a schematic diagram of a rotor position estimation method for redundant monitoring of rotor position.

[0038] In case 1, a unique operating point dependent high frequency injection (Figure 3) is used to ensure a good and robust estimation of the rotor position in the case of low modulation, while in case 2 no unique injection is applied since the existing high frequency injection from another algorithm can be relied upon. The injection in case 1 is rectangular, which achieves a nearly constant modulation depth, unlike other shapes (e.g. sinusoidal). Figure 3 shows the frequency f of the injection in case 1. i and amplitude a i Schematic diagram of a characteristic map for selecting the injection frequency f in dependence on the operating point. i and amplitude a i The selection depends on the operating point of torque M, rotation speed n, and DC voltage V DC , switching frequency fs and at least one characteristic map KF is determined depending on the driving mode FM. By incorporating the driving mode FM, it is possible to distinguish, in particular, between efficiency-optimized, comfort-optimized and performance-optimized injection.

[0039] Figure 4 shows the inverter voltage v for driving the electric machine. α and current i α FIG.

[0040] Information about the rotor angle is contained in the anisotropic high-frequency components of the active pointers T1 and T2, which can be extracted by subtracting them from other components, as described in the prior art. Here, the voltage state in which the inverter 1 is not in freewheeling mode is referred to as the active pointer T1 or T2. This voltage state includes six switching states: 100, 110, 010, 011, 001, and 101, where 1 represents a closed switch and 0 represents an open switch at the top of the half-bridge. In this case, T1 always represents the first active pointer of a new switching period, and T2 represents the second active pointer. The freewheeling state is defined as the zero pointer (Nullzeiger: 000 or 111), where T0 always represents the zero pointer at the beginning of a new switching period, and T7 represents the zero pointer midway through the switching period.

[0041] The isotropic high frequency components are calculated using multiplication depending on the d and q currents by the observed active pointers T1, T2 and the calculated voltage in the average admittance characteristic map. The low frequency components are extracted from the long zero pointer T7. Then, the isotropic (average) admittance characteristic map Y Σ (i d ,i q ) is calculated from an inductance characteristic map obtained, for example, from offline characterization of the electric machine.

[0042]

number

[0043] The pointers are outlined for half a switching period on the left side of Figure 4. For the active pointers T1 and T2, the longer of the two active pointers T1 and T2 (T A,use For the zero pointer, T7 (T Z,use ) is selected. If the long active pointer T1, T2 is shorter than a defined threshold, as mentioned, it is known from the prior art that the reconstruction of the intermediate active pointer T1, T2 can be performed by evaluating two zero pointers T0 and T7. For this purpose, points P and Q on the right side of FIG. 4 are reconstructed and the average current derivative is calculated using knowledge of the lengths of the two active pointers T1, T2. Using the definition of the minimum pointer length of the long active pointer T1, T2, it is decided whether to apply direct identification from the long active pointer T1, T2 or identification by reconstruction.

[0044] The pointer selection 3 is the same in the two mentioned cases, only the input signal differs. Based on the input signal, the duty cycle is extracted and the current pointer position is detected. Then, using incrementing and sorting, the pointers are prepared so that the correct pointer (long active pointers T1, T2, long zero pointers T0, T7) is selected. Furthermore, an evaluation is performed to determine whether to apply the reconstruction process for the intermediate active pointers T1, T2. After selection and sorting, the LSQ algorithm 6 is triggered accordingly, and the voltage levels of the individual pointers are provided based on an offline calculation table that compensates for component-specific voltage drops. In this case, the triggering of the LSQ algorithm 6 starts with a delay start time defined depending on the power semiconductor, thereby minimizing current overshoots.

[0045] FIG. 5 is a schematic diagram of the angle control loop 2.

[0046] Using the anisotropic high frequency components of the long / middle active pointers T1, T2 and their associated voltages, a conversion is performed as described in the prior art. The converted signals with similar phase positions are then further processed by angle control loop 2, which extends with an additional PT1 section, to compensate for the effects of cross-coupling.

[0047] The proposed method is not limited to the application of a fixed switching frequency and modulation method, Space Vector Modulation (SVPWM). s mainly affects the angle control loop 2, whereas when applying the discontinuous method, the position or presence of the pointer changes. This is taken into account in the pointer selection 3 block, where appropriate, which allows applying a number of discontinuous methods (GDPWM, DPWMMIN, DPWMMAX, DPWM0, DPWM1, DPWM2, DPWM3).

[0048] In one embodiment, the method for operating a synchronous machine is carried out as shown in FIG. 1, where the synchronous machine is implemented in a polyphase manner and is powered via a rectifier or inverter 1.

[0049] In one embodiment, for minimum high frequency rectangular injection, an offline optimized operating point dependent (torque M, rotation speed n, DC voltage V) DC The specification (commands) of the operating conditions (regarding the load, circuit breaker temperature, and driving mode FM) is carried out by means of several characteristic maps KF (see Figure 3) with application-specific optimization targets. For example, one characteristic map KF can have an optimization target for efficiency, another for comfort, and another for dynamics.

[0050] In one embodiment, the switch positions calculated by the pulse width modulation 4 component are passed to a sensorless algorithm 5, which identifies the relevant pointer characteristics, i.e., pointer length, pointer start point, and pointer end point, and calculates the resulting voltage levels within each pointer.

[0051] In one embodiment, sensorless determination of rotor position is performed using a combination of the methods of arbitrary injection and least squares algorithm6, where long active pointers T1, T2 are applied to determine the total current derivatives, long intermediate zero pointers T0, T7 are applied to smooth the low frequency parts, and offline calculation tables dependent on d-current and q-current are applied for the high frequency isotropic components.

[0052] In one embodiment, depending on the power semiconductors of the inverter 1, a delay start time of the least squares algorithm 6 is defined at the start of each pointer to minimize the effect of current overshoot.

[0053] In one embodiment, a coupled angle control loop 2 is applied to minimize high frequency oscillations in the estimated rotor position signal.

[0054] In one embodiment, a pointer length dependent switch is made between direct calculation from the active long pointers T1, T2 and reconstruction from an average value from the two active pointers T1, T2.

[0055] In one embodiment, variable sequence control of pointer selection 3 and sorting is provided to allow for handling discontinuous modulation schemes.

[0056] In one embodiment, a method for monitoring a permanently excited synchronous machine is carried out, as shown in Figure 2. The synchronous machine is then implemented in a polyphase manner, fed via a rectifier or inverter 1 and operated at least partly by a method based on the general anisotropy of high frequency injection.

[0057] In one embodiment, the switch positions of the gate drivers 7 of the inverter 1 are read back to identify the pointer times and resulting voltage levels within each pointer and to comply with safety related requirements. [Explanation of symbols]

[0058] 1 inverter 2 Angle control loop 3 Pointer selection 4. Pulse Width Modulation 5 Sensorless Algorithm 6 Least Squares Algorithm, LSQ Algorithm 7 Gate Drivers a i Injection Amplitude fi i Injection frequency FM driving mode f s Switching Frequency i α current KF characteristic map M Torque n rotation speed P Points Q Points T0 zero pointer T1 active pointer T2 active pointer T7 Zero Pointer T A,use Selected active pointer T Z,use Selected Zero Pointer V DC DC voltage v α Voltage

Claims

1. 1. A method for determining the rotor position of an electric machine without a rotor position sensor, wherein an anisotropy-based position control and / or position adjustment algorithm with high frequency excitation of a voltage modulation is applied to estimate the rotor angle, comprising: Furthermore, the signal with the minimum disturbance gradient tangent and affected by measurement noise is calculated using a least squares algorithm (6) that is able to identify high frequency current derivatives; Furthermore, an angle control loop (2) is applied to minimize the disturbances and jumps in the rotor angle that occur, Additionally, amplitude-optimized high-frequency injection is applied A method characterized by:

2. The high frequency excitation of the voltage modulation is generated by an arbitrary injection 2. The method according to claim 1, characterized in that

3. The rotor angle estimation may be applied for feedback to field oriented control and / or for redundant monitoring of the rotor angle, depending on the requirements of the electric machine.

3. The method according to claim 1 or claim 2, characterized in that

4. an operating point dependent high frequency rectangular injection is applied for feedback to said field oriented control; The frequency of the injection (f ｉ ) and amplitude (a ｉ ) depends on the operating point of the torque (M), the rotation speed (n), and the DC voltage (V ＤＣ ), switching frequency (f ｓ ) and driving mode (FM) are determined using several characteristic maps (KF).

4. The method according to claim 3, characterized in that

5. By incorporating the driving modes (FM), a distinction is made between efficiency-optimized injection, comfort-optimized injection and performance-optimized injection.

5. The method according to claim 4, characterized in that

6. The information about the rotor angle is stored in the active pointer (T １ , T ２ ) is extracted by differential formation from the anisotropic high frequency components of The active pointer (T １ , T ２ ) is a voltage state in which the inverter (1) for controlling said electric machine is not in freewheeling; The isotropic high frequency components are observed by the active pointer (T １ , T ２ ) and the voltage calculated in the average admittance characteristic map by multiplication depending on the d-current and the q-current, The low frequency component is a long zero pointer (T ７ ) is extracted from The admittance characteristic map is calculated from the inductance characteristic map. The method according to any one of claims 1 to 5, characterized in that

7. The inductance characteristic map is obtained from offline characteristics of the electric machine.

7. The method according to claim 6, characterized in that

8. Long and / or medium active pointers (T １ , T ２ ) and its associated voltages are used to perform a transformation, and the transformed signals with similar phase positions are then further processed by an angle control loop (2) that extends an additional PT1 section to compensate for the effects of cross-coupling.

8. The method according to claim 6 or claim 7, characterized in that

9. The method is applied to a permanently excited synchronous machine applied as a drive machine of a vehicle. The method according to any one of claims 1 to 8, characterized in that

Images