Method of determining total pitch deviation of position sensor

Abstract

A method for determining a total pitch deviation of a position sensor, the position sensor being equipped with a magnetic disc and a magnetic probe, the magnetic disc comprising pairs of magnetic poles, the method comprising the steps of: - recording the magnetic field strength of one revolution of mechanical rotation, measured by the magnetic probe, as an angular signal as a function of the angle of rotation of the magnetic disc; - determining zero-crossing positions from the recorded angular signal and a determined number of zero-crossing positions; - determining a pole pair length from the zero-crossing positions; and - determining the total pitch deviation from the pole pair length.

Description

Technical Field

[0001] This invention relates to a method for calibrating sensors, and more particularly to a method for calibrating angular position sensors. Background Technology

[0002] An absolute position sensor provides an analog sinusoidal signal corresponding to the angular position of the rotor. More precisely, this sensor comprises a rotor formed by magnetic rings equipped with magnetic poles and a stator equipped with magnetic sensors capable of detecting the magnetic field of each pole.

[0003] As the rotor rotates, the magnetic poles pass in front of the magnetic sensor sequentially. Based on the distance between the magnetic sensor and the magnetic poles, a current is induced in the magnetic sensor. This current forms a periodic signal with a sinusoidal shape, which is a function of time, and the signal strength depends on the distance between the magnetic sensor and the magnetic poles. Based on the known geometry and rotational speed of the sensor (rotor), the change in signal over time can be converted into a change over angle, thus allowing a correlation between time and the angular position of the rotor to obtain a sinusoidal signal that links signal strength to angular position.

[0004] Such absolute position sensors are commonly used in motor control. In the specific case of a belt starter generator, the accuracy of the sensor output signal becomes increasingly important due to the need to correctly control the machine's torque with minimal noise. Furthermore, the battery's alternating current level must be maintained within a fixed limit to prevent a decline in overall vehicle performance.

[0005] One source of inaccurate sensor output is the magnetic coil itself, as it cannot perfectly reflect the rotor's position. This is due to two reasons: first, the magnetic field signal generated by the sensor is not a perfect sine wave; second, each individual period of the sine wave may have a different length.

[0006] To better characterize this magnetic ring, the total pitch deviation (TPD) can be used as a parameter. It assesses the cumulative deviation of the magnetic ring's position by measuring the single pitch deviation (SPD) of each magnetic pole on the magnetic ring.

[0007] The method of determining the TPD of a magnetic coil with magnetic poles is similar to the method used in ABS applications to determine the TPD of gears or mechanical encoders.

[0008] The angular distance between the two closest magnetic poles of the same polarity is defined as the interval. The general method for calculating TPD is described by the following formula:

[0009] The single pitch deviation (SPD) of interval i can be calculated based on the following formula:

[0010] [Formula 1]

[0011]

[0012] in,

[0013] P Theoretical It is the theoretical period of a single angular signal;

[0014] P real (i) is the actual period of the angular signal at interval i.

[0015] It should be noted that the actual period P real (i) is determined by two poles of the same polarity, i.e., the two North Poles or the two South Poles. Similarly, the actual period P... real (i) is determined by the edges of the same signal, i.e., between two rising edges or two falling edges.

[0016] The total or cumulative pitch deviation TPD(i) at interval i can be calculated using the following formula:

[0017] [Formula 2]

[0018]

[0019] The total pitch deviation caused by the rotation of a single mechanical wheel can be calculated using the following formula:

[0020] [Formula 3]

[0021]

[0022] Wherein, Nbpp is the number of intervals contained in one mechanical rotation.

[0023] Based on the current state of technology, the following methods for determining TPD are known.

[0024] After 1.3 rounds of mechanical rotation, an angle signal is generated by the sensor. The angle signal contains more than 324,000 points, of which 81,000 points are locally measured and the rest are interpolated values.

[0025] The zero-crossing samples (hereinafter referred to as "zero-crossing samples") are identified within the angular signal. A zero-crossing sample is equal to the nearest point on either side of the zero-crossing point of the angular signal at the zero intensity level measured by the sensor. The zero intensity level is in the direction measured by the sensor, i.e., the normal direction of the Hall effect sensor.

[0026] The angular position of the zero-crossing point is determined based on each zero-crossing sample.

[0027] The method then involves calculating the length of pole pairs in degrees, which can begin either on the ascending edge or the descending edge.

[0028] Then, the pole pair length is compared with the theoretical pole length and expressed as a percentage of the theoretical pole length to determine the pitch deviation.

[0029] Determine the sum of all pitch deviations for a single mechanical rotation, and then record the peak-to-peak values ​​of the pole pairs.

[0030] The method of using noisy angular signals to determine TPD is problematic. Summary of the Invention

[0031] The present invention aims to provide a method for determining the total pitch deviation of a position sensor, the position sensor being equipped with a magnetic disk and a magnetic detector, the magnetic disk comprising paired magnetic poles, the method comprising the following steps:

[0032] - The magnetic field strength of a mechanically rotating disc, measured by a magnetic detector, is recorded as an angular signal, which is a function of the rotation angle of the magnetic disk;

[0033] - Determine the zero-crossing position based on the recorded angle signal and the determined number of zero-crossing positions:

[0034] - Determine the pole pair length based on the zero-crossing position; and

[0035] - Determine the total pitch deviation based on the pole pair length.

[0036] To determine the zero-crossing position based on the recorded angle signal, the following steps are performed:

[0037] - Identify the zero-crossing samples in the angular signal, and determine the zero-crossing position from the zero-crossing samples;

[0038] - Generate a zero-crossing signal whose intensity at the zero-crossing angle is equal to the difference between the maximum and minimum values ​​of the zero-crossing sample, wherein the maximum and minimum values ​​are located on either side of the zero-crossing angle, and the zero-crossing signal is equal to zero at other locations;

[0039] - The number of zero-crossing positions within the predetermined angular position range is determined by averaging at least the zero-crossing signals within the predetermined angular position range;

[0040] - Filter the angular signal that contains multiple zero-crossing positions exceeding a threshold within a predetermined angular position range to reduce the number of zero-crossing positions;

[0041] - Perform linear interpolation on each zero-crossing sample; and

[0042] - The actual zero-crossing position is determined based on all zero-crossing positions contained in the interval to avoid outliers deviating from the result. The interval is defined as the angular distance between the two nearest poles of the same polarity.

[0043] The following methods can be used to determine the recalculated zero-crossing position: taking the median, mean, weighted summation, using a percentage sample of measurements near the zero point to perform linear polynomial fitting to obtain the zero-crossing point, or taking the midpoint of the maximum / minimum value of the measured zero-crossing position.

[0044] Even if no zero-crossing samples are detected to overlap, a step of filtering the diagonal signal can be implemented to reduce the number of zero-crossing positions.

[0045] Determine the standard deviation of the recalculated zero-crossing position, and issue a warning to the user if the standard deviation exceeds a predetermined value.

[0046] Different methods can be used for different intervals.

[0047] To determine the pole pair length based on the zero-crossing position, the following steps can be performed:

[0048] - Preprocess the angle signal containing the recalculated zero-crossing position based on the strength and rate of change of the angle signal relative to the first zero-crossing position at the start of recording;

[0049] - The length of each pole is determined as the difference between the angular position associated with the rising edge of the angular signal and the angular position associated with the falling edge, with the zero-crossing position associated with the rising edge and the falling edge.

[0050] - The length of a pole pair is defined as the sum of the lengths of the adjacent positive and negative poles in the pole pair.

[0051] To preprocess the diagonal signal, the following steps can be performed:

[0052] - Determine if the angle signal begins at a zero-crossing position with increasing intensity; if so, no processing is required.

[0053] If the corner signal does not originate from a zero-crossing position with increasing intensity, then

[0054] - Determine the first zero-crossing point of the intensity increase; and

[0055] - Move the portion of the angle signal from the beginning to the first zero-crossing position where the intensity increases to the end of the angle signal.

[0056] To determine the total pitch deviation based on the pole pair length, the following steps can be performed:

[0057] - The pitch deviation ( / pitch error) is calculated as the percentage of the difference between the determined pole pair length and the theoretical pole pair length, which is determined by the structure of the sensor's magnetic disk;

[0058] - Determine the cumulative pole pair length deviation (error) during one round of mechanical rotation ( / one mechanical revolution) for each pole pair length;

[0059] - Determine the maximum and minimum values ​​of the cumulative pole pair length deviation, and then determine the cumulative pitch deviation of the magnetic disk by subtracting the minimum value of the cumulative pole pair length deviation from the maximum value of the cumulative pole pair length deviation.

[0060] The number of record points can be increased by inserting data.

[0061] The above method for determining cumulative pitch deviation has the following advantages:

[0062] This method eliminates the need for an encoder with a large number of native points or an interpolation stage that introduces additional errors. Instead, it performs a linear input method within the algorithm to find zero-crossing positions, thereby reducing cost and size.

[0063] This method is robust against angular signal noise.

[0064] The calculations are performed based on a complete mechanical rotation, which corresponds exactly to the actual angular length of the magnetic coil.

[0065] Calculating the TPD value using either the rising or falling edge can provide different indications for the design of the magnetizing yoke (zero-crossing accuracy, coil arrangement, magnetic eccentricity, etc.). Attached Figure Description

[0066] The invention and its advantages can be better understood through a detailed study of a series of specific embodiments. These specific embodiments are given by way of non-limiting examples and are illustrated in the following figures, wherein:

[0067] Figure 1 This invention illustrates the main steps of the method described in this invention;

[0068] Figure 2 Displays the theoretical angle signal and the corresponding zero-crossing signal;

[0069] Figure 3 Display the first corner signal;

[0070] Figure 4 Display the second-angle signal and the preprocessing performed on it; and

[0071] Figure 5 Display the third-angle signal and the preprocessing performed on it. Detailed Implementation

[0072] Figure 1 The method shown for determining the position sensor TPD includes the following steps:

[0073] In step 1, the angular signal of one rotation of the position sensor is recorded to generate a series of sampling points, particularly no fewer than 18,000 sampling points. In one specific embodiment, interpolation can be used to increase the number of sampling points. The angular signal is a periodic signal whose intensity changes as a function of the cumulative rotation angle since the start of signal recording. It should be noted that the angular signal is generated by the magnetic sensor detecting the magnetic field strength as the magnetic disk poles pass by.

[0074] The cumulative rotation angle is determined by an encoding device comprising an encoding disk and a pickup sensor. The encoding disk contains several encoders spaced precisely at fixed angles. The magnetic field of an encoder is acquired whenever the pickup sensor detects it.

[0075] The acquisition (also known as "sampling") of the magnetic field strength at a fixed angular position is determined by this design and is independent of any increase or decrease in rotational speed. This acquisition cannot be achieved through clock setting, because clock setting would cause the magnetic field acquisition to occur after a predetermined time has elapsed. Any increase or decrease in rotational speed will cause a change in the angular position of the acquired magnetic field.

[0076] In the second step 2, the zero-crossing position is determined based on the recorded angular signal; steps 2a to 2d are used to achieve this determination.

[0077] In step 2a, zero-crossing samples in the angular signal are determined, and the zero-crossing positions are determined based on these samples. Thus, a zero-crossing signal is generated, whose intensity at the zero-crossing angular position is equal to the difference between the maximum and minimum values ​​of the zero-crossing samples, located on either side of the zero-crossing angular position. The zero-crossing signal is zero at other positions. Figure 2 Display the theoretical angle signal and its corresponding zero-crossing signal.

[0078] It is necessary to point out that the zero-crossing position corresponds to the angular position where the angular signal crosses the zero (magnetic field) strength base level. In some implementations, the base level may have an offset, which is therefore removed before the calculation begins.

[0079] Because of noise in the angular signal, the zero-crossing position in the signal may differ from the physical position of the magnetic disk at the theoretically corresponding zero-crossing position. More precisely, in this case, multiple zero-crossing positions may occur within a single interval.

[0080] In step 2b, the number of zero-crossing points within a predetermined angular position range is determined by averaging the zero-crossing signals over that range. The number of zero-crossing points within each predetermined angular position range is compared to a predefined threshold. If a range contains a number of zero-crossing points greater than the threshold, that range is considered to contain a large number of clustered zero-crossing points.

[0081] Then, the angular signal within the angular position range where a large number of zero-crossing positions are clustered is filtered to reduce the number of zero-crossing positions.

[0082] The aforementioned range is also checked for overlapping zero-crossing samples to determine whether a diagonal signal filtering step should be performed. This overlap occurs when a portion of one zero-crossing sample coincides with a portion of another zero-crossing sample. This overlap indicates that the diagonal signal is too noisy, or that each interval does not contain enough samples.

[0083] If zero-sample overlap is detected, an error message is sent and the calculation is stopped.

[0084] In step 2c, linear interpolation is performed on each zero-crossing sample. If noise exists in the angular signal, there may be multiple solutions for each interval.

[0085] In step 2d, if it is determined that there are multiple zero-crossing positions within a certain interval, the actual zero-crossing positions are recalculated based on all the zero-crossing positions in that interval. In one implementation, the recalculated zero-crossing positions are determined to avoid outliers deviating from the results.

[0086] More precisely, the zero-crossing position can be recalculated using the following methods: taking the median, mean, weighted summation, using a linear polynomial fit to a percentage sample of measurements near zero to derive the zero-crossing point, or taking the midpoint of the maximum / minimum values ​​of the measured zero-crossing positions. Calculate using these methods, and then determine the standard deviation of the results from all these methods. If the standard deviation is too high, issue a warning to the user.

[0087] Different methods can be applied to different intervals.

[0088] The length of the pole pair is determined in step 3, and steps 3a to 3c make this determination possible.

[0089] In step 3a, the angle signal, which includes the recalculated zero-crossing position, is preprocessed based on the strength and rate of change of the angle signal relative to the first zero-crossing position at the start of recording. Various scenarios of angle signal preprocessing will be further described below.

[0090] In step 3b, the length of each pole, in degrees, is determined as the difference between the angular position associated with the rising edge and the angular position associated with the falling edge of the angular signal, and the zero-crossing position is associated with the rising edge and the falling edge.

[0091] Then, in step 3c, the pole pair length is determined as the sum of the lengths of the adjacent positive and negative poles in the pole pair.

[0092] In step 4, the TPD is determined based on the pole pair length. Steps 4a to 4c make this determination possible.

[0093] In step 4a, the pitch deviation is calculated as the percentage of the difference between the determined pole pair length and the theoretical pole pair length relative to the theoretical pole pair length. The theoretical pole pair length is determined by the structure of the sensor's magnetic disk.

[0094] In step 4b, the cumulative pole pair length deviation under one-round mechanical rotation is determined by using [Formula 2], and thus the vector value of TPD(i) is obtained.

[0095] In step 4c, the maximum and minimum values ​​of the cumulative pole pair length deviation are determined, and then the TPD of the magnetic disk is determined according to the maximum and minimum values ​​of the cumulative pole pair length deviation using Formula 3, thereby obtaining a single value of TPD.

[0096] The different scenarios for performing signal preprocessing in step 3a will now be described. The following description focuses on computations using rising edges; the same principle applies to computations using falling edges.

[0097] Figure 3 The first signal shown comes from signal acquisition that begins at the initial stage of the cycle, and has a zero-crossing position Z1 and a positive rate of change. In other words, signal recording begins when the pole pair is just in front of the magnetic sensor in the position sensor.

[0098] Therefore, the pole pair length is calculated as follows:

[0099]

[0100] Obtaining such a signal in actual measurements is uncommon.

[0101] Figure 4 The second signal shown is from the results collected when the magnetic field (intensity) increases before the zero-crossing position.

[0102] In this case, the portion of the signal containing the final zero-crossing position Z9 is obtained by transposing the angular signal X before the first zero-crossing position. 0 This is obtained by shifting the signal to the end of the recording. This shift alters the recording of a mechanical rotation angle signal, modifying the angle signal to begin at a zero-crossing position and have a positive rate of change. In other words, the resulting angle signal is similar to the first signal described above.

[0103] Therefore, the pole pair length is calculated using the following method:

[0104]

[0105] Figure 5 The third signal shown comes from the acquisition results when the magnetic field (intensity) decreases.

[0106] In this case, the signal includes the portion X1 of the last two zero-crossing positions (Z9, Z10). 0 This is achieved by shifting the portion of the signal containing the first two zero-crossing positions (Z1, Z2) to the end of the recording. Similar to the second signal, this shift alters the mechanical rotation angle signal, modifying it to begin at the zero-crossing position and have a positive rate of change. In other words, the resulting angle signal is similar to the first signal described above.

[0107] Then, calculate the pole pair length using the following formula:

[0108]

[0109] The method described above can determine the TPD value even with noisy signals, and the obtained TPD value represents the actual pitch deviation between the angular signal and the pole pair position.

Claims

1. A method for determining the total pitch deviation of a position sensor, the position sensor being equipped with a magnetic disk and a magnetic detector, the magnetic disk including paired magnetic poles, the method comprising the following steps: - The magnetic field strength of a mechanically rotating disc, measured by a magnetic detector, is recorded as an angular signal, which is a function of the rotation angle of the magnetic disk; - Determine the zero-crossing position based on the recorded angle signal and the number of determined zero-crossing positions: - Determine the lengths of multiple pole pairs based on the zero-crossing positions; as well as - The total pitch deviation is determined by comparing the lengths of the plurality of pole pairs with the theoretical pole pair length, wherein the theoretical pole pair length is determined based on the structure of the magnetic disk.

2. The method according to claim 1, characterized in that: To determine the zero-crossing position based on the recorded angle signal, the following steps are performed: - Identify the zero-crossing samples in the angular signal, and determine the zero-crossing position based on the zero-crossing samples; - Generate a zero-crossing signal whose intensity at the zero-crossing angle is equal to the difference between the maximum and minimum values ​​of the zero-crossing sample, wherein the maximum and minimum values ​​are located on either side of the zero-crossing angle, and the zero-crossing signal is equal to zero at other locations; - The number of zero-crossing positions within the at least predetermined angular position range is determined by averaging the zero-crossing signals within the at least predetermined angular position range; - Filter the angular signal that contains multiple zero-crossing positions exceeding a threshold within the predetermined angular position range to reduce the number of zero-crossing positions; - Perform linear interpolation on each zero-crossing sample; as well as - The actual zero-crossing position is determined based on all zero-crossing positions contained in the interval to avoid outliers deviating from the result. The interval is defined as the angular distance between the nearest two poles of the same polarity.

3. The method according to claim 2, characterized in that, Choose one of the following methods to determine the recalculated zero-crossing position: take the median, mean, weighted summation, use a percentage sample of measurements near the zero point to perform a linear polynomial fit to obtain the zero-crossing point, or take the midpoint of the maximum / minimum value of the measured zero-crossing position.

4. The method according to claim 2 or 3, characterized in that: Even if no zero-crossing samples are detected to overlap, a step is taken to filter the diagonal signal to reduce the number of zero-crossing positions.

5. The method according to claim 2 or 3, characterized in that: Determine the standard deviation of the recalculated zero-crossing position, and issue a warning to the user if the standard deviation exceeds a predetermined value.

6. The method according to claim 2 or 3, characterized in that: Different methods are used for different intervals.

7. The method according to any one of claims 1 to 3, characterized in that, To determine the lengths of multiple pole pairs based on their zero-crossing positions, the following steps are performed: - Preprocess the angle signal containing the recalculated zero-crossing position based on the strength and rate of change of the angle signal relative to the first zero-crossing position at the start of recording; - The length of each pole is determined as the difference between the angular position associated with the rising edge of the angular signal and the angular position associated with the falling edge, with the zero-crossing position associated with the rising edge and the falling edge. - The length of a pole pair is defined as the sum of the lengths of the adjacent positive and negative poles in the pole pair.

8. The method according to claim 7, characterized in that, To preprocess the diagonal signal, perform the following steps: - Determine if the angle signal begins at a zero-crossing position with increasing intensity; if so, no processing is required. If the corner signal does not originate from a zero-crossing position with increasing intensity, then - Determine the first zero-crossing point of the intensity increase; and - Move the portion of the angle signal from the beginning to the first zero-crossing position where the intensity increases to the end of the angle signal.

9. The method according to any one of claims 1 to 3, characterized in that, To determine the total pitch deviation based on a comparison of the multiple pole pair lengths and the theoretical pole pair length, the following steps are performed: - The pitch deviation is calculated as the percentage of the difference between the determined pole pair length and the theoretical pole pair length relative to the theoretical pole pair length; - Determine the cumulative pole pair length deviation of one round of mechanical rotation for each pole pair length; - Determine the maximum and minimum values ​​of the cumulative pole pair length deviation, and then determine the cumulative pitch deviation of the magnetic disk by subtracting the minimum value of the cumulative pole pair length deviation from the maximum value of the cumulative pole pair length deviation.

10. The method according to any one of claims 1 to 3, characterized in that: The number of record points is increased by inserting data.

Images