Method for correcting an angular error in sensorless detection of a rotation angle of an electric machine rotor using an injection method

Abstract

A method for correcting an angular error in sensorless detection of a rotation angle of a rotor of an electric machine using an injection method is disclosed, comprising: processing a current Q-current I q by means of a first filter element and a current D-stream I d by means of a second filter element; determining an amount of the processed D-stream I d_hp and determining a moving average of the magnitude; determining a squaring of the moving average; calculating a product of the processed Q-stream I q_hp and the processed D-current I d_hpand determining a moving average of the product; and calculating a quotient of the squaring and the moving average of the product, and providing the quotient as angular error data. Also disclosed is a computing unit for executing the procedure and a vehicle with an electric machine and the computing unit.

Description

[0001] The present invention relates to a method for correcting an angular error in sensorless detection of the rotation angle of a rotor of an electric machine by means of an injection method. The invention further relates to a computing unit for carrying out the method and a vehicle comprising an electric machine and the computing unit.

[0002] US 2020 / 0 169 204 A1, EP 2 897 282 A1, DE 103 44 024 B4 and the article Yi, Bowen [et al.]: A new signal injection-based method for estimation of position in interior permanent magnet synchronous motors. In: IET Power Electronics, Vol. 13, 2020, No. 9, pp. 1865-1874, ISSN 1755-4535 describe methods for correcting angular errors.

[0003] Methods for sensorless angle measurement of an electric machine rotor using injection techniques are known. In these methods, a current amplitude along an estimated Q-axis is evaluated, which is generated by a voltage injection into an estimated D-axis. Subsequently, an angular error is determined, and the estimated angle is corrected by this error.

[0004] Based on this, the object of the present invention is to provide an advanced technique for correcting an angular error when using an injection method. In particular, dependencies of the angular error are to be minimized and / or compensated in order to improve the estimation of a rotation angle.

[0005] This problem is solved by the features of the independent claims. Advantageous embodiments are the subject of the dependent claims.

[0006] A method for correcting an angular error in sensorless detection of the rotation angle of an electric machine rotor using an injection method has been disclosed. The method comprises the following steps: processing a current Q-current I q by means of a first filter element and a current D-stream I d by means of a second filter element; determining an amount of the processed D-stream I d_hp and determining a moving average of the magnitude; determining a squaring of the moving average; calculating a product of the processed Q-stream I q_hp and the processed D-current I d_hp and determining a moving average of the product; and calculating a quotient of the squaring and the moving average of the product, and providing the quotient as angular error data.

[0007] The injection method can, for example, be configured to superimpose an electrical signal (voltage signal) onto a voltage used to drive the electric machine. The voltage can also be a control signal for the electric machine. The electrical signal can, for example, be injected into the estimated D-axis of the control signal.

[0008] The control signal can, for example, be a modulated control signal, particularly one that exhibits pulse-width modulation (PWM). The frequency of the electrical signal can depend on the frequency of the control signal. Specifically, the frequency of the electrical signal can be determined according to the Nyquist-Shannon sampling theorem or the Whittaker-Kotelnikov-Shannon sampling theorem.

[0009] The filter element can, for example, include a high-pass filter designed to cut off or attenuate frequencies below a cutoff frequency. Furthermore, the filter element can also include a band-pass filter designed to cut off or attenuate frequencies below a lower cutoff frequency and above an upper cutoff frequency.

[0010] The angular error data can, for example, include information about an angular error between an estimated Q-axis and an actual Q-axis.

[0011] The current amplitude in the Q-axis can be proportional to an angular error that depends on operating points and various parameters of the electric machine. Furthermore, the angular error can also depend on parameters of the injection method.

[0012] The disclosed method can therefore be used to determine and take into account the angular error in the control of the electric machine.

[0013] Furthermore, the method can be used to keep parameters affecting the current amplitude constant. In particular, it can reduce the dependence between the amplitude of the voltage injection along the D-axis and / or the frequency of the injection voltage.

[0014] The procedure may further include calculating a product from the quotient and compensation data, and providing the product as angular error data.

[0015] The compensation data can, for example, have values ​​designed to compensate for and / or eliminate dependencies on inductances of the electrical machine.

[0016] Accordingly, the angular error can be evaluated independently of the operating points of the electrical machine, so that the angular error is, for example, independent of a change in the differential inductance.

[0017] The compensation data can have at least one value L. komp include, which is based on a calculation from provided differential inductance values ​​of the D-current L dd and differential inductance values ​​of the Q current L qq based.

[0018] The method can particularly preferably linearize a detected angular error, for example an angular error signal, or angular error data, relative to an actual angular error of the electric machine. This can be especially useful at operating points where the electric machine does not behave linearly due to iron saturation effects.

[0019] This can occur particularly at high Q currents, since the differential inductances (L) then dd and L qq ) change the most.

[0020] Accordingly, the compensation data can be designed to account for dependencies on the differential inductance in the Q-axis L. qq and / or the differential inductance in the D-axis L dd to compensate for and / or eliminate.

[0021] The calculation can be based on the following formula: Lkomp=Lqq / (Ldd−Lqq).

[0022] The compensation data can include lookup table values, which provide compensation values ​​and / or scaling values ​​for the provided differential inductance values ​​of the D-current L. dd and / or for the differential inductance values ​​of the Q current L qq exhibit

[0023] Using lookup table values ​​allows for better parameter compensation and simplifies calculations. Therefore, using lookup table values ​​can also result in faster processing compared to methods that do not use them.

[0024] The procedure can also include determining angle data from the angle error data.

[0025] The angular data can be configured, for example, to be processed by an αβ filter and / or directly supplied to a tracking controller. The processed angular data can preferably include an angular velocity and / or a rotation angle.

[0026] Also disclosed is a computing unit that is configured and programmed to execute the disclosed method. The computing unit can be, for example, a microcomputer, a computer system, a networked computer system, and / or a cloud-based computer system. The computing unit is configured to acquire and process current values ​​from the electrical machine for executing the method.

[0027] Also disclosed is a vehicle with an electric machine and the disclosed computing unit.

[0028] The present invention is described in detail below with reference to the figures. These show: Fig. 1 a circuit arrangement with an angle estimator and a modulation module; Fig. 2. A construction of the angle estimator from Fig. 1; Fig. 3 another circuit arrangement with an angle estimator and a modulation module; Fig. 4. A construction of the angle estimator from Fig. 3; and Fig. 5 a diagram showing the course of a ratio of L q to L d represents an angular error.

[0029] The present disclosure is described below with reference to preferred embodiments and the figures. However, this description of the disclosure should not be considered exhaustive.

[0030] Fig. Figure 1 shows a circuit arrangement 1 with an angle estimator 20 for correcting an angular error in a sensorless detection of a rotation angle of a rotor of an electric machine using an injection method and a modulation module 18 for generating an offset voltage value U αβ .

[0031] A transformation module 10 processes a current value I αβ and an estimated angle value to calculate a current value I dqto determine which value is provided to the angle estimator 20. The current value I is based on this. αβ based on measured current values ​​of the electric machine. The transformation module 10 can, in particular, use a d / q transformation (also called Park transformation) to determine the current value I. dq determine.

[0032] The angle estimator 20 processes the current value I dq , to determine angular error data which are provided to a filter module 14.

[0033] Filter module 14 determines an angular velocity value and a filtered angle value, which are output and provided to angle module 16 and modulation module 18. Filter module 14 can determine the angular velocity value and the filtered angle value, in particular, using an αβ filter.

[0034] Angle module 16 processes the angular velocity value and the filtered angle value to determine the estimated angle value, which is then provided to transformation module 10. Angle module 16 can be specifically configured to compensate for delays, latencies, or processing times within the system.

[0035] Modulation module 18 processes the angular velocity value and the filtered angle value to calculate the offset voltage value U. αβ to determine which one is designed to correct the angular error. The offset voltage value U αβ can be provided to the electric machine, e.g., to a control system for the electric machine.

[0036] Fig. Figure 2 shows a setup of the angle estimator 20. Fig. 1, where the current value I dq in a current D-stream I d and a present Q-current I qis divided. The current D-current I d is processed from a first filter element 21 to a D-stream I d_hp processed. The current Q-current I q is processed by a second filter element 22 into a Q-stream I q_hp processed.

[0037] The first filter element 21 and the second filter element 22 are configured as high-pass filters. Both filter elements 21 and 22 can also each be configured as band-pass filters.

[0038] The processed D-current I d_hp is provided to a first determination element 23.1 and a first calculation module 24.1. The first determination element 23.1 determines an amount of the processed D-stream I. d_hp The first calculation module, 24.1, calculates a product of the processed Q-current I. q_hp and the processed D-current I d_hp .

[0039] A second determining element 23.2 determines a moving average of the absolute value and a third determining element 23.3 determines a squared value of the moving average.

[0040] A fourth determining element 23.4 calculates a moving average of the product. A second calculation element 24.2 calculates a quotient of the squared value and the moving average of the product. Furthermore, the second calculation element 24.2 provides the quotient as angular error data.

[0041] Fig. Figure 3 shows a preferred circuit arrangement 1 with an angle estimator 20 for correcting an angular error in sensorless detection of a rotation angle of a rotor of an electric machine, a compensation module 12 and a modulation module 18 for generating an offset voltage value U αβ .

[0042] A transformation module 10 processes a current value I αβand an estimated angle value to calculate a current value I dq to determine which value is provided to the angle estimator 20. The current value I is based on this. αβ based on measured current values ​​of the electric machine. The transformation module 10 can, in particular, use a d / q transformation (also called Park transformation) to determine the current value I. dq determine.

[0043] The compensation module 12 processes the current value I dq , to determine a differential inductance value of the D-current L dd and a differential inductance value of the Q current L qq to determine these inductance values ​​L dd , L qq are then provided to the angle estimator 20. The compensation module 12 can then determine the differential inductance value of the D-current L. dd and / or the differential inductance value of the Q current L qq Preferably determined using lookup table values.

[0044] The angle estimator 20 processes the current value I dq , to determine angular error data which are provided to a filter module 14.

[0045] Filter module 14 determines an angular velocity value and a filtered angle value, which are output and provided to angle module 16 and modulation module 18. Filter module 14 can determine the angular velocity value and the filtered angle value, in particular, using an αβ filter.

[0046] Angle module 16 processes the angular velocity value and the filtered angle value to determine the estimated angle value, which is then provided to transformation module 10. Angle module 16 can be specifically configured to compensate for delays, latencies, or processing times within the system.

[0047] Modulation module 18 processes the angular velocity value and the filtered angle value to calculate the offset voltage value U. αβ to determine which one is designed to correct the angular error. The offset voltage value U αβ can be provided to the electric machine, e.g., to a control system for the electric machine.

[0048] Fig. Figure 4 shows the setup of an angle estimator 20 made of Fig. 3, where the current value I dq in a current D-stream I d and a present Q-current I q is divided. The current D-current I d is processed from a first filter element 21 to a D-stream I d_hp processed. The current Q-current I q is processed by a second filter element 22 into a Q-stream I q_hp processed.

[0049] The first filter element 21 and the second filter element 22 are configured as high-pass filters. Both filter elements 21 and 22 can also each be configured as band-pass filters.

[0050] The processed D-current I d_hp is provided to a first determination element 23.1 and a first calculation module 24.1. The first determination element 23.1 determines an amount of the processed D-stream I. d_hp The first calculation module, 24.1, calculates a product of the processed Q-current I. q_hp and the processed D-current I d_hp .

[0051] A second determining element 23.2 determines a moving average of the absolute value and a third determining element 23.3 determines a squared value of the moving average.

[0052] A fourth determining element 23.4 calculates a moving average of the product. A second calculation element 24.2 calculates a quotient of the squared value and the moving average of the product.

[0053] A fifth determining element 23.5 determines a compensation value with the differential inductance value of the D-current L. dd and the differential inductance value of the Q current L qq .

[0054] A third calculation element 24.3 calculates a product of the quotient and the compensation value and provides the product as angular error data.

[0055] Fig. Figure 5 shows a diagram that illustrates a ratio of L q to L d over an angular error. The diagram includes a first graph 31, which shows the ratio of L q to L d represents the angular error.

[0056] In this case, a second graph 32 represents a compensation value and / or scaling value in relation to the first graph 31, which is based on a calculation of Lqq / (Ldd - Lqq).

[0057] The second graph 32 can represent a preferred application area for the method for compensating angular errors in the range of 0 to 40°. Reference symbol list 1 Circuit arrangement 10 Transformation Module 12 Compensation Module 14 Filter module 16 Angle Module 18 Modulation Module 20 angle estimators 21 first filter element 22 second filter element 23.1 first determining element 23.2. second determining element 23.3 third determining element 23.4 fourth determining element 23.5 fifth determining element 24.1 first calculation element 24.2 second calculation element 24.3 third calculation element 31 first graph 32 second graph

Claims

[1] Method for correcting an angular error in sensorless detection of a rotation angle of a rotor of an electric machine by means of an injection method, comprising: Processing a current Q-stream I q by means of a first filter element and a current D-stream I d by means of a second filter element, Determining the amount of the processed D-stream I d_hp and determining a moving average of the amount; Determining the squaring of the moving average; Calculating a product from the processed Q-stream I q_hp and the processed D-current I d_hp and determining a moving average of the product; and Calculating a quotient from the squaring and the moving average of the product, and providing the quotient as angular error data. [2] Method according to claim 1, characterized by, that the procedure further includes calculating a product from the quotient and compensation data and providing the product as angular error data. [3] Method according to claim 2, characterized by that the compensation data has at least one value L komp include, which is based on a calculation of provided differential inductance values ​​of the D-current L dd and differential inductance values ​​of the Q current L qq based. [4] Method according to claim 3, characterized by , that the calculation is based on the following formula: Lkomp=Lqq / (Ldd−Lqq). [5] Method according to claim 3 or 4, characterized by , that the compensation data includes lookup table values ​​which provide compensation values ​​and / or scaling values ​​for the provided differential inductance values ​​of the D-current (L) dd ) and / or the provided differential inductance values ​​of the Q current (L qqexhibit. [6] Method according to any one of claims 1 to 5, characterized by that the procedure further includes determining angle values ​​from the angle error data. [7] Computing unit which is trained and programmed to execute the method according to any one of claims 1 to 6. [8] Vehicle with an electric machine and the computing unit according to claim 7.

Images