EP3180592A1

EP3180592A1 - Method for determining an orthogonality error between two sensor signals

Info

Publication number: EP3180592A1
Application number: EP15750060.4A
Authority: EP
Inventors: Jens GLEISSBERG; Markus Irth
Original assignee: Continental Automotive GmbH
Current assignee: Continental Automotive GmbH
Priority date: 2014-08-14
Filing date: 2015-08-13
Publication date: 2017-06-21
Also published as: US20170153127A1; WO2016024001A1; DE102015215510A1; CN106574851A; KR20170029608A

Abstract

The invention relates to a method for determining a corrected rotation angle (x_t1) of a raw rotation angle (x) recorded using an angle sensor which, on the basis of the raw rotation angle (x), outputs first and second raw rotation angle signals (s_r, c_r) which have a periodic profile and are in an orthogonal relationship with one another, wherein a deviation from the orthogonal relationship between the sensor signals can occur on account of the error (y), having the steps of: determining a correction value (y) by means of a determination unit for determining the correction value (y) and making the correction value (y) available to a correction unit, applying the correction value to at least one of the raw rotation angle signals (s_r, c_r) by means of the correction unit for determining at least one corrected rotation angle signal (s_oc, c_oc), and calculating the corrected rotation angle (x_t1) using at least one of the corrected rotation angle signals (s_oc, c_oc).

Description

Method for determining an orthogonality error between two sensor signals

The invention relates to a method for determining a corrected angle of rotation of a raw angle of rotation detected by means of an angle sensor, an angle sensor arrangement and a drive device for carrying out such a method.

From the prior art, the document DE 10 2010 003 201 AI is known in which a method for Be ^¬ tuning a rotation angle with an angle measuring unit is disclosed. In this document, it is disclosed how the rotation angle can be corrected by means of a correction value so that the influence of a false angle F on the value of the rotation angle is reduced. This is an error, resulting from a not quite exact orthogonality between a sine and cosine-shaped sensor signal of the sensor element resul ^¬ advantage.

In the priority application DE 10 2014 216224.6 a method is further disclosed, which shows a different possibility than in the aforementioned document for determining the Orthogonalitätsfehlers or a correction value for Korri ^¬ yawing the Orthogonalitätsfehlers between the two angle signals.

Proceeding from this, it is an object of the invention to show a possibility with which a corrected rotation angle calculation or determination can be carried out precisely and in a simple manner.

The object is achieved according to a method according to claim 1. Further advantageous embodiments of the method according to the invention are the subject of the dependent claims, which are characterized by expressly be made the subject of the description.

The invention was based on the basic concept of providing a method by means of which the correction value determination as well as the introduction of the correction value into the rotation angle calculation can be integrated as simply as possible into existing processes. On the basis of the method according to the invention, in particular the determination unit, a particularly flexible implementation of the error correction in an already set operation is possible. Furthermore, it is possible to adapt the possibilities for calculating or determining the correction value as required, for example, depending on the requirements for the accuracy of the calculation of the correction value. In contrast to the prior art, in addition to a correction unit, a determination unit is used to determine the correction value. In this way, the way in the investigation unit, as the corrective ^¬ turwert is determined to be set flexibly. The inventive method is characterized in advantageous ^¬ legally developed such that corrected for both Rohdrehwinkelsignale by means of the correction value of rotation angle signals are determined and the corrected rotational angle is calculated using both the corrected rotational angle signals. In this way it is ensured that both raw rotation angle signals are corrected in such a way that they essentially assume their desired value.

The method according to the invention is advantageously further developed in that the corrected rotational angle signals are normalized to normalized rotational angle signals, preferably to a value range between -1 and 1. By means of normalization, the corrected rotational angle signals are processed so that the determination of the corrected rotational angle is independent, of the Scaling the Rohdrehwinkelsignale can be done. The inventive method can thus be applied to a variety of sensor types readily. According to the invention a method for determining a corrected rotary angle (x_tl) of a Rohdrehwinkels detected by an angle sensor (x) and angle sensor signal is pre ^¬ propose, the function of the Rohdrehwinkels (x) first and second Rohdrehwinkelsignale (s_r, c_r) outputs or sensor signals, which have a periodic course and are in an orthogonal relationship to one another, wherein due to the error (y) a deviation from the orthogonal relationship between the sensor signals may occur, comprising the steps:

Determining a correction value (y) by means of a determination unit for determining the correction value (y) and

Providing the correction value (y) to a correction ^¬ unit,

Applying the correction value to at least one of the Rohdrehwinkelsignale (s_r, c_r) by the corrective-tureinheit for determining at least one Corridor ^¬ alloyed rotation angle signal (s_oc, c_oc), and

Calculating the corrected rotation angle (x_tl) based on at least one of the corrected rotation angle signals (s_oc, c_oc).

The inventive method is characterized in advantageous ^¬ legally further developed in that the corrected, preferably normalized, Rohdrehwinkelsignale are continuously accessed for determining the correction value. The so-called on-line determination of the correction value allows an even more precise determination of the correction value. The external effects on the Rohdrehwinkelsignale as well as other influences that lead to a change in Orthogonalitätsfehlers can be taken into account in this way continuously. Also, this way can Also a vibration behavior of the orthogonality error with in the correction value are taken into account. In particular in applications in the automotive sector, for example in the case of steering assistance, this embodiment can be particularly advantageous in order to determine an accurate steering assistance torque.

The method according to the invention is further developed in a particularly advantageous manner in that the determination of the correction value comprises the steps:

Forming a radius signal using the sums of squares of the corrected or normalized raw rotation angle signals,

Determining the 2 * nth harmonic of the radius signal, with n equal to a whole positive number, and

- Determining the error on the basis of a phase-shifted by 90 ° to the rotation angle value of the amplitude at the second harmonic.

This way of determining the error correction is particularly easy to implement and has proven to be a particularly stable and accurate method for determining the error correction. It can be done both online and offline. The advantage of this embodiment lies in the fact that the error can be determined on the basis of the radius signal, which can be determined solely on the basis of the two raw rotational angle signals. Since these signals are necessary anyway for determining the angle of rotation, there is no need to change existing rotational angle sensors. There is no need for a reference sensor signal with which the sensor signals could be individually compared to determine the error in the individual sensor signals. The method can therefore be integrated particularly easily into existing systems, since the electronic means necessary for the evaluation of the sensor signals are present anyway. The mathematical derivation follows as follows. The amplitude of the radius signal can be mapped by the equation e_orth (x) = sin ² (x) + cos ² (x + y), where x is the value of the angle of rotation and y is the value of the error. If the error is y = 0, the aforementioned condition of the addition theorem is satisfied. Inter alia, the amplitude of the radius signal has a maximum at an angle of 45 °, so that the radius signal assumes the following value at x = 45 °: e_orth (45 °) = l-sin (y)

For the determination of e_orth (45 °) from the measured signals, other errors in the stereo output signal e_orth (x) or e must be hidden. This can be done by harmonic analysis using Fourier transform. In this case, the magnitude of the imaginary part of the 2nd harmonic as a result of the Fourier transformation corresponds to the value e_orth (45 °). The advantage here is the complete suppression of other errors in the signal e_orth (x) by the sole consideration of the second harmonic, such as offset, gain or Deachsierungsfehlern because they do not affect the 2nd harmonic. Since the amplitudes are calculated as a result of the Fourier transformation, but y describes the peak-to-peak value (from the minimum value to the maximum value), the magnitude of the second harmonic imaginary part must be doubled: y = aresin [2 * (e_orth, 2 * n . , in the) ]

The inventive method is characterized in advantageous ^¬ legally further developed in that the second or 2 * n-th harmonic of the radius signal is determined by means of a Fourier transformation, preferably discrete Fourier transformation. The discrete Fourier transformation makes it possible to perform the aforementioned method with as little computation as possible.

The inventive method is characterized in advantageous ^¬ legally further developed in that the imaginary part of the second or 2 * n-th harmonic of the Ra ^¬ diussignals is used for calculating the correction value. This embodiment is based on the idea that the error has an immediate effect on the amplitude of the second or an integral multiple of the second harmonic of the radius signal and therefore an analysis of the amplitude of the second harmonic gives a direct indication of the magnitude of the error. In particular, this off ^¬ guide die based on the recognition that the error in the second harmonic of the radius signal occurs with a phase shift of 90 ° to the angle of rotation, so that the imaginary part of the upper shaft results in an exclusion of the error.

The inventive method is characterized in advantageous ^¬ legally further developed in that the correction value using the equation y = Σ [arcsin (2 * e_2 * n, in)] or preferably on the basis y = Σ [2 * e_2 * n, im] is calculated, where e_2 * n, im is the value of the imaginary part of the second or 2 * nth harmonic. Especially with the latter simplification, a possibility is shown with which the computational effort for determining the correction value continues is reducible. It is based on the assumption that the correction value is usually within a small range of values around zero and thus this assumption leads to a simplified calculation ^method for a large number of ^application cases.

The inventive method is characterized in advantageous ^¬ legally further developed in that the value of the imaginary part of the harmonic only at intended angles of rotation (X_ST) is determined. In this way it is ensured that within one revolution the imaginary part of the 2nd or 2nd nth harmonic for an angle is not calculated several times.

The inventive method is characterized in advantageous ^¬ legally further formed, {* 2n / N, 2 * 2n / N, (Nl) * 2n / N 0, l, l} is calculated that the values of the imaginary part of the upper shaft at the rotation angles X_ST = where N is a positive integer. This fixes the calculation of the imaginary component to fixed angles or angular positions. This has the advantage that the loading ^¬ calculation of the imaginary part can be determined independent of the speed with which the rotational angle changes. The inventive method is characterized in advantageous ^¬ legally further developed in that the correction value is determined once for each sensor. The so-called offline calculation or determination of the correction value requires little effort due to the one-time calculation and can also be carried out by means of external calculation units, as a result of which manufacturing costs of sensor arrangements can be kept low.

The object is further achieved by means of a Winkelelsensoran ^¬ order with a sensor unit for detecting the Rohdrehwin kelsignale and an evaluation unit for carrying out a method according to one of the aforementioned embodiments.

Furthermore, the object is also achieved by means of a drive ^{¬ device} with an electric motor, in particular for a steering assistance device, a control device for controlling the electric motor, and with an angle sensor arrangement according to the aforementioned embodiment.

The invention will be described below with reference to an embodiment and with reference to several figures. Show it:

1 shows a basic structure of an angle sensor for an angle sensor arrangement,

Fig. 2 is an exemplary illustration of a

Orthogonality error in the angle sensor,

3 is a block diagram of an angle sensor arrangement for

Carrying out the method according to the invention,

4 is a detailed representation of a determination ^¬ unit from the block diagram of Figure 3,

FIG. 5 is a block diagram for performing a comparison test between an absolute target angle value and the corrected rotation angle; and FIG

Fig. 6-8 is a comparison of the results with and without rotation angle correction.

Some reference numerals in the figures each have a small index, which will be described in the following description by way of substitution of the character "". FIG. 1 shows the construction of an angle sensor 101, which is used in the angle sensor arrangement 100 shown in FIG. The angle sensor 101 has an angle encoder 102. The angle encoder 102 is a rotational angle signal x, which are detected by two sensor elements ^¬ 103s, 103c. Each sensor element 103s, 103c generates from the rotation angle signal x of the angle sensor 103 a raw rotation angle signal s_r and c_r, which is used for calculating or for determining the corrected rotation angle x_tl.

The raw rotation angle signals s_r, c_r are a periodic signal, for example, a sine and a Kosi ^¬ nus signal, which are shifted by 90 ° to each other phases. Because of the orthogonal relationship between the sensor signals, the sensor signals should comply with the condition according to the addition theorem ^¬ sin ² (x) + cos ² (x) = 1, with x as a value of the rotation angle. Due to different causes, which may occur, for example, once in the production of the angle sensor 102 or external influences during operation, which are permanently present, there may be a deviation of the orthogonal relationship between the two Rohdrehwinkelsignalen s_r and c_r. FIG. 2 shows by way of example an error case in which the actual raw rotational angle signal s_r has an orthogonality error and therefore erroneously assumes the course of the curve s_r_err. As a result, the phase shift between the two Rohdrehwinkelsignalen not 90 ° but 90 ° + y. The inventive method allows to determine this error as a correction value y. FIG. 3 shows an angle sensor arrangement 100 with which the method according to the invention can be carried out. The win ^¬ kelsensoranordnung comprises on the one hand the angle sensor 101. This is connected to a correction unit 110. The correction unit 110 is in turn with a normalization unit 130 connected. The normalization unit is connected to an angle calculation unit 140. Further, the angle ^¬ sensor arrangement 100 comprises a determination unit 150 for determining the correction value y. The determination unit 150 is connected to the correction unit 120, and provides this to the corrective ^¬ y turwert available. Connected in the sense of the ^exemplary embodiment means a connection for the transmission of signals or data. Preferably, these are a Da ^¬ tenverbindung. In the figures, the compounds are each represented by an arrow. It is not necessarily necessary that the angle ^sensor arrangement has a standardization unit 140. This is advantageous, however.

The method is carried out in such a way that on the one hand the determination unit 150 determines a correction value y and provides this value to the correction unit 120. The correction value is applied to at least one of the raw rotation angle signals s_r, c_r by means of the correction unit for determining at least one corrected rotation angle signal s_oc, c_oc, from which the corrected rotation angle x_tl is calculated on the basis of at least one of the corrected rotation angle signals s_oc, c_oc. In this embodiment, corrected rotational angle signals s_oc, c_oc are determined for both raw ^¬ rotation angle signals s_r, c_r by means of the correction value (y) and the corrected rotation angle x_tl is calculated by means of both corrected rotation angle signals s_oc, c_oc. Further, this embodiment also includes the step of normalizing the corrected rotation angle signals s_oc, c_oc to normalized rotation angle signals s_n, c_n. The normalization is preferably carried out in such a way that the standardized rotation angle signals s_c, c_n are then within a range of values between -1 and 1.

Furthermore, the exemplary embodiment has the property that the corrected and standardized raw rotational-angle signals s_n, c_n are continuously retrieved to determine the correction value. For this purpose, as shown in FIG. 3, the normalized rotation angle signals s_n, c_n are forwarded to a computing unit 151, in which a radius signal e is calculated by means of the equation s_n ² + c_n ² . The correction value y is then calculated or determined in the determination unit 150 on the basis of the radius signal. FIG. 4 shows the structure of the determination unit 150 in more detail.

The determination unit 150 comprises on the one hand DFT block 152 and on the other hand an integration block 153.

Within the DFT block 152, the second harmonic is calculated from the radius signal e by means of a discrete Fourier transformation. Since only the imaginary component is decisive for determining the orthogonality error or the correction value y, the calculation can be further simplified. Ie. instead of the

Calculation about the equation

can perform the calculation by the equation

if-l l = ^s

be further specified.

However, this is done only for certain angular positions or angles. For this purpose, the current corrected angle of rotation x_tl is retrieved and a check 154 is made as to whether the current corrected angle of rotation x_tl corresponds to a predetermined angle of rotation x_ST stored in a memory. In a first pass, the corrected rotation angle x_tl can also be uncorrected. Preferably, the values of the predetermined rotation angle lie at interpolation points x_ST, which may be defined as follows: x ST = {0,1,1 * 2π / Ν, 2 * 2π / Ν, ..., (Nl) * 2π / Ν}

If this is the case, the calculation of the summand e_2, im is performed, which is illustrated by the arrow 155. The calculation of the sinusoidal signal can also be replaced by the use of a table with the sine values matching the position x_ST. From the imaginary part of the second harmonic e_2, im, a single correction value y_s = arcsin (2 * e_2, im) can then be calculated by means of the equation. Based on the assumption that the values of the imaginary part of the second harmonic are close to zero, a simplification can also be made, namely y_s = 2 * e_2, in

For the compensation of the orthogonality error, the value y must be the sum

Y = Y + Y_s are formed, since the return of the determined Kompensationswert y an orthogonality error of y_s = 0 would result. The initial value of y is 0. To avoid sudden jumps in the angle signal x_tl, the signal y should have a change limitation, which can be easily realized, for example, by limiting the value of y_s. The effectiveness of the method of the invention is represented by the following ^¬ in Figures 5 to eighth FIG. 5 shows a structure with which a simulation of the method according to the invention was carried out. The construction essentially corresponds to the construction of the angle sensor arrangement 100. To compare the effectiveness of the angle sensor arrangement 100 or the method according to the invention, the value of a reference sensor x_ref was used to make a comparison between the value of the reference sensor x_ref and the corrected rotation angle x_tl. The raw angles of rotation x were performed on real resolvers with subsequent simulation of the angle error, ie the angle error was artificially added. The signals x_ref, x_tl, s_r and c_r were recorded. There were different

Orthogonality error used as described below.

During the simulation, the s_r and c_r signals are fed into the model as stimuli. Thus, two systems for Winkelbe ^¬ calculation are realized, which receive exactly the same input data. On the one hand, the model simulation with automatic orthogonality compensation by means of the correction value y. In addition, a control unit without orthogonality compensation was used for comparison.

The angle x_tl calculated by the model can be compared with the reference angle x_ref and forms the angle difference x_diff, comp. In addition, the angle error without

Orthogonality compensation from the measured values of the

Controller calculated as x_diff for comparison.

Both systems will give time to settle for about 3.2 seconds before the simulation performs the automatic orthogonality correction. Thus, the effect can be visualized visually. Figures 6, 7 and 8 show the results for resolvers with injected orthogonality errors

- Figure 6: 0.154 ° el

- Figure 7: 0, 655 ° el

- Figure 8: -0.3 ° el.

The upper diagram shows the angle error without compensation x_diff as a red / dashed line. The com ^¬ pensierte value x_diff_comp is as a blue / solid line. At the time t = 3.75s, the first value for the orthogonality error was determined and the

Orthogonality compensation supplied. The effect is clear to see, the angle error has no dominant 2nd order more.

In the lower diagram, the signal y as the determined orthogonality error is shown as a green / solid line and the signal y_s as a blue / dashed line as the determined residual orthogonality error.

In Figures 6 to 8 can be seen the significant reduction of Orthogonalitätsfehlers. In the application shown here, the remaining angle error no longer has a critical size. The compensation value y was directly applied without smoothing, whereby angle jumps in the change of y occurred particularly in FIG. For this, one can easily see how y_s once assumes the value of y in order then to remain at the level of recognition accuracy. If you smooth y, the signals y and y_s show a transient response.

It should be noted that the compensation value of the Spit ^¬ ze-peak value is the harmonic angle error. He has thus twice the value of the y_s = aresin (e_2, im). Furthermore, is good too recognize that the correction of orthogonality the angular displacement equal to the orthogonality besei ^¬ Untitled.

Below, further advantageous variants are listed, which can be combined with the aforementioned embodiments:

1. A method for determining an error (y) between two sensor signals (sl, s2) in an angle sensor which outputs the sensor signals (sl, s2) in dependence on an angle sensor, which have a periodic course and are mathematically in an orthogonal relationship to one another, wherein, due to the error (y), a deviation from the orthogonal relationship between the sensor signals may occur, comprising the steps of:

Forming a radius signal (e_orth) by means of the sums of squares of the sensor signals,

Determining the 2 * nth harmonic of the radius signal (e_orth), with n equal to a whole positive number, and

Determining the error of a value of the amplitude at the second harmonic phase phase-shifted by 90 ° to the angle of rotation value.

2. The method according to claim 1, characterized by

Determining a frequency component or portions of the radius signal by means of a Fourier transform, and

Determine the error based on the imaginary part of the frequency component or the frequency components of the second harmonic.

3. The method according to claim 2, characterized in that the Fourier transformation is performed by means of a fast and / or discrete Fourier transformation.

4. The method according to claim 2 or 3, characterized by, calculating the error y by means of the equation y = aresin I e_orth, 2 * n. , im i is determined, where e_orth, 2nd, in the imaginary part of the amplitude of the 2 * n-th harmonic of the radius signal maps.

Method according to one of the preceding claims, characterized in that the real part of the 2 * nth harmonic is used as a scale for a scaling error.

Method according to one of the preceding claims, characterized in that the method is used during operation.

Method according to one of claims 1-5, characterized in that the method is carried out before commissioning of an angle sensor, in particular by means of an external computing unit.

Method for determining an error (y) between two sensor signals (sl, s2) in an angle sensor which outputs the sensor signals (sl, s2), which have a periodic course and are mathematically in an orthogonal relationship as a function of an angle sensor, due to the error (y) may cause a deviation from the orthogonal relationship between the sensor signals, comprising the steps of:

Determining a rotation angle (x) from the sensor signals (sl, s2),

Determining a conversion value by performing a Fourier transformation for the determined rotation angle,

Determine the error based on the conversion value. An angle sensor for detecting a rotation angle, comprising a sensor element that outputs the sensor signals as a function of a win ^¬ kelgebers, which have a periodic course and are mathematically in an orthogonal relationship to each other, and

an arithmetic unit for carrying out the method according to any one of the preceding claims.

Claims

claims

1. A method for determining a corrected angle of rotation (x_tl) of a detected by an angle sensor Roh ^¬ rotation angle (x), the first and second Rohdrehwinkelsignale (s_r, c_r) outputs depending on the Rohdrehwinkels (x) having a periodic course and in one orthogonal relationship to each other, wherein due to the error (y) may occur a deviation from the orthogonal relationship between the sensor signals, comprising the steps:

Determining a correction value (y) by means of a determination unit for determining the correction value (y) and providing the correction value (y) to a correction unit,

Applying the correction value to at least one of the Rohdrehwinkelsignale (s_r, c_r) by means of the corrective ^¬ tureinheit for determining at least one corrected rotational angular signal (s_oc, c_oc), and

2. The method according to claim 1, characterized in that for both Rohdrehwinkelsignale (s_r, c_r) by the Cor ^¬ rekturwertes (y) corrected rotational angle signals (s_oc, c_oc) are determined and the corrected turning angle (x_tl) by means of both corrected rotational angle signals (s_oc, c_oc) is calculated.

3. The method according to claim 1 or 2, characterized by normalizing the corrected rotational angle signals (s_oc, c_oc) to standardized rotational angle signals (s_n, c_n), preferably to a range of values between -1 and 1. Method according to one of the preceding claims, characterized in that the corrected, preferably nor ^¬ mierten, Rohdrehwinkelsignale (s_oc, c_oc / s_n, c_n) are continuously retrieved to determine the correction value.

A method according to claim 3, characterized in that the corrected rotation angle signals (s_oc, c_oc) or the normalized rotation angle signals (s_n, c_n) in response to the corrected rotation angle (x_tl) for determining the correction value (y) are retrieved.

Method according to one of the preceding claims, characterized in that the determination of the correction value comprises the steps:

Forming a radius signal (e) by means of the sums of squares of the corrected or normalized raw rotational-angle signals (s_oc, c_oc / s_n, c_n),

Determining the 2 * n th harmonic of the radius signal (e), with n equal to a whole positive number, and determining the error on the basis of a value of the amplitude at the second harmonic phase-shifted by 90 ° to the rotation angle (x).

A method according to claim 5, characterized in that the second or 2 * nth harmonic of the radius signal (e) is determined by means of a Fourier transformation.

Method according to Claim 6, characterized in that the imaginary component of the second or 2 * nth harmonic of the radius signal (e) is used to calculate the correction value.

9. The method according to any one of claims 5 to 9, characterized in that the correction value based on the equation y = Σ 2 * arcsin (e 2 * n, im) or preferably based on

y = Σ 2 * e_2 * n, im

where e_2 * n, im is the value of the imaginary part of the second or 2 * nth harmonic.

Method according to one of claims 6 or 7, characterized in that the value of the imaginary portion of the harmonic wave is determined only at intended rotation angles (x_ST).

Method according to claim 9, characterized in that the values of the imaginary part of the harmonic at the angles of rotation x_ST = {0, 1, 1 * 2n / N, 2 * 2n / N, (N-1) * 2n / N}

is calculated, where N is a positive integer.

Method according to one of claims 1 to 3 and 5 to 8, characterized in that the correction value (y) is set once for each sensor. 13. Angle sensor arrangement with a sensor unit for detecting the Rohdrehwinkelsignale (s_r, c_r) and an evaluation ^¬ unit for carrying out a method according to one of claims 1 to 11. 14. Drive device with an electric motor, in particular for a steering assistance device, a control device for controlling the electric motor, and with an angle sensor arrangement according to claim 12.