EP3278117A1

EP3278117A1 - Method for producing a speed signal of an electric motor

Info

Publication number: EP3278117A1
Application number: EP16706968.1A
Authority: EP
Inventors: Benjamin Kaufner
Original assignee: Schaeffler Technologies AG and Co KG
Current assignee: Schaeffler Technologies AG and Co KG
Priority date: 2015-03-31
Filing date: 2016-01-27
Publication date: 2018-02-07
Also published as: US10352957B2; US20180088145A1; DE102015205772B3; WO2016155713A1

Abstract

The invention relates to a method for the improved speed determination of electric motors while taking into account angle errors of an angular position sensor. A commutation sensor based on a magnetic pole wheel indicator, the angle signal (01) of which has an angle-dependent and thus periodic angle error (03), is used. By differentiation, said angle signal (01) burdened with error is converted into a raw speed signal (04), which has two disturbance components. One disturbance component follows from the angle error (03) and is the disturbance waviness, which is periodic just like the angle error (03). The second disturbance component is formed by a noise component superposed on the raw speed signal (04). In order to correct the raw speed signal (04) burdened with error, half of the period duration of the raw speed signal (04) must be determined. If said half of the period duration is determined, two samples (09, 10) of the raw speed signal (04) lying exactly one half of the period duration of the raw speed signal (04) apart from each other are averaged. A speed signal having no disturbance waviness but rather only the noise component results from these steps.

Description

Method for generating a speed signal of an electric motor

The invention relates to a method for generating a speed signal, which is used for improved speed determination of a rotary

Electric motor is suitable. The improvement in the extraction of a signal representing the speed of the electric motor is achieved by taking into account the application-specific error form. This is a by a

Commutation sensor detected speed signal corrected by a sample averaging. The speed of the electric motor may be an angular velocity or a rotational velocity. The invention also relates to an electric motor assembly in which said method is used.

An accurate speed signal is needed wherever electric motors are used for a drive whose control is to be more accurate than with conventional block commutation. The speed signal can be determined via an angle signal, which can be detected via a commutation sensor. However, the angle signal and thus also the speed signal may have interference, which hinder a precise control of the electric motor. In the prior art, for example solutions are shown to correct the interference by a previously determined correction value.

DE 102 60 862 A1 shows a method for correcting an angular and / or distance-measuring sensor arrangement, in which sinusoidal or cosinusoidal

Measuring signals are evaluated, which by scanning a moving

Measured object to be won. The correction of the angular or phase errors of the measurement signals takes place in that of a plurality of measurement signals

Constants are derived for the estimation and correction of the angular or phase error and / or the amplitude of the measurement signals.

DE 101 33 524 A1 describes a method for correcting a dynamic error of a sensor. This dynamic error superimposes the sensor signal, for example, in the form of periodic fluctuations whose frequency and amplitude are constantly changing change with the speed of the engine. To correct the dynamic error, the sensor output signal is fed to a filter circuit and a correction circuit. The correction circuit receives one or more filtered signals output from the filter circuit and generates a corrected sensor signal from information obtained by comparing the filtered signals with the unfiltered sensor output signal or corrected signals derived therefrom.

DE 10 201 1 105 502 A1 shows a method for adjusting a phase offset between a rotor position sensor and a rotor position of an electrically commutated motor. In this case, the rotor position sensor measures a position of the rotor of the motor, which is controlled during operation with a block commutation. The measured position is compared with an expected position. From the

Difference between measured and expected position, a phase offset is formed, which is used to control the electrically commutated motor. The position of the rotor is measured with an absolute rotor position sensor, which is set in relation to a motor parameter that characterizes the expected position of the rotor.

US 2007/0043528 A1 shows a method and system for measuring a

Speed of a high-speed motor, wherein from the rotor speed of the motor or from a filtered speed of the motor, a rotor period is calculated and from this a number of points in a moving average value is calculated. The object of the present invention is based on the prior art is a method for generating a speed signal of a

To provide electric motor to be made available by which signal interference, so that the generated speed signal better corresponds to the actual speed of the electric motor.

This object is achieved by a method according to the appended claim 1. The inventive method is suitable for use on electric motors, in which a very accurate control is necessary. This is a

Speed signal required, which is permanently determined, so even during a revolution of the rotor of the electric motor, exactly.

As a signal transmitter, a commutation sensor based on a magnetic

Polradgebers are used, which determines an angular signal on the electric motor, from which the speed signal can be obtained. The angle signal determined by the commutation sensor has interference components at least in the form of a noise component and an angle error. The noise component of the angle signal of the commutation sensor occurs due to disturbances in the signal transmission. The angle error is based on system or

Application-specific error characteristics of the commutation sensor. The angle error is angle-dependent and consequently occurs periodically. The

Period duration of the angle error depends on the type of

Commutation sensor, of which the period of the angle signal is dependent. The period of the angle signal and the period of the angle error are thus in direct proportion to each other. In a first step of the method according to the invention, an angle signal of the electric motor is determined by means of a commutation sensor, which is preferably based on a magnetic Polradgeber. The Polradgeber may for example be equipped with a plurality of Hall sensors, which are associated with poles on a fixedly connected to the rotor of the electric motor pole. The angle signal measured by such a commutation sensor inevitably has spurious components in the form of a noise component and an angle error. The angle error is angle dependent and repeats depending on the speed of the

Electric motor with each revolution in the same form, since he on system or

Application-specific error characteristics of the commutation sensor based. Thus, the angle error is periodic. The period of the angle error depends on the type of commutation sensor. Both the period of the angle signal, as well as the period of the angle error can be determined as soon as a plurality of revolutions of the electric motor has been executed. In a next step, the angle signal is converted by differentiation into a raw speed signal. Due to the derivation of the measured angle with time, the angular error which is still relatively small in the angle signal is amplified and causes an increased error of the raw velocity signal, which due to its periodicity is also referred to as "stator ripple". The shape of the angular error of the angle signal and thus also the shape of the Storwelligkeit the raw velocity signal repeats itself periodically over an ideal value with each magnetic pole pair of the commutation sensor. The shape of the

Angular error and the Storwelligkeit is further characterized by a point symmetry, which results from the fact that the angle signal oscillates around the ideal value, similar to a sinusoidal noise signal on the angle signal. The

Period duration of the speed signal periodized by the current ripple depends on the type of commutation sensor and its system and application specific fault characteristics.

Depending on the type of commutation sensor used, it detects, for example, one or two periods of the entire motor revolution per complete revolution

Angle signal.

In a further step of the method, half the period of the

Speed signal determined. For this, the engine has to be at least halfway

Rotation have performed, since half the period is determined from the measured by the commutation sensor samples and is not predetermined. Since the samples and thus also half the period of the raw speed signal can change continuously due to speed changes, half the period is preferably continuously during the operation of the

Electric motor determined. It is simplified but also possible to determine the period only at each jump of the angle signal from + π to -π. Other methods may also be used to determine the correct distance between the two sets of samples, for example, by returning the calculated speed signal. In a next step, at least two samples of the raw velocity signal are averaged, which are exactly half a period apart. These samples represent the values sampled by the commutation sensor. To correct for the interference ripple of the speed signal, two or more samples must be selected, each having a distance of exactly half a period length of the speed signal.

Preferably, the most recent of the at least two

selected samples by the most recent of the provided samples. Due to the periodicity and the point symmetry of the interference ripple of the raw velocity signal, in each case two samples for averaging the raw velocity signal are fundamentally sufficient, so that the signal after the averaging no

Interference has more, but only contains the noise component. This formed average is then provided as the speed signal of the motor. The averaging of the samples takes place z. In the usual way by adding the two velocity values of the samples and dividing by the number of added samples.

A particular advantage of the method according to the invention is that a very good correction of the speed signal takes place, since the interference ripple is completely eliminated and initially only the noise component remains, which can be filtered out in further steps. Another advantage arises from the fact that the speed-dependent averaging is not a full period of the

Noise signal required. Since the inventive method is preferred for the entire operation of the

Electric motor is to deliver a continuous speed signal, the o.g. Procedural steps at each time, at which the

Commutation sensor delivers a sample, carried out continuously in succession as soon as the electric motor has made a first half turn.

In a preferred embodiment of the method according to the invention, a commutation sensor serves to determine the angle signal, which is used per whole Rotation of the electric motor detects one period of the angle signal and two periods of the angle error.

In particularly preferred embodiments of the method according to the invention, the commutation sensor detects an angle signal with an angle error whose period is twice as large as that of the angle signal. As a result, the angle error has two full periods per revolution of the electric motor.

In preferred embodiments of the method according to the invention, the determination of half the period duration of the raw velocity signal is carried out by determining the duration between two value range jumps of the angle signal. The value range jumps preferably occur in each case after a complete revolution of the electric motor, particularly preferably when the angle signal jumps from + π to -π or vice versa. This duration is the period of the angle signal and preferably the duration for one revolution of the electric motor, since between two

Jumps from + π to -π exactly 2π, so one revolution lies. Since the period of the angle signal is preferably half as large as that of the angular error and thus the Störwelligkeit the raw velocity signal, the determined period of the angular signal must be quartered to obtain half the period of the raw velocity signal.

In an alternative preferred embodiment of the invention

Procedure, the determination of half the period of the raw speed signal by the return of the determined and corrected

Speed signal. Since at the beginning of the procedure still no corrected

Speed signal is available, the previously determined and not yet corrected raw speed signal must be used in the first run of the method steps according to the invention to determine half the period. In all subsequent passes of the method steps, the corrected values of the speed signal are used to determine half the period of the raw speed signal. The raw speed signal of the electric motor represents values in the form of numbers of revolutions. Thus, the half period be calculated from half of the reciprocal of the number of revolutions.

In preferred embodiments, more than two samples of the raw velocity signal are averaged to correct the raw velocity signal. In this case, two mean values are averaged over each n samples, which one

Have a distance of exactly half the period of the raw speed signal, where n> 1. Another possibility is to merge all the samples of n pairs in one step, with the respective samples of the pairs being exactly half the period of the raw velocity signal. A particular advantage of this preferred embodiment is that not only the Storwelligkeit but also the noise component of the raw velocity signal can be corrected. For this purpose, preferably n> 5 pairs of samples are selected, more preferably 5 <n <50 pairs, with the respective first samples of the pairs and the respectively second samples of the pairs immediately following each other. Preferably, each of the temporally younger samples of the pairs is the most recent of the provided samples.

Using the method according to the invention, an improved speed-controllable electric motor arrangement can be constructed, which can be used, for example, for the drive of electric vehicles.

Further details, advantages and developments of the invention will become apparent from the following description of preferred embodiments of the invention, with reference to the drawing. Show it:

Fig. 1 is a diagram depicting the ideal and the angle signal measured by the commutation sensor;

FIG. 2 shows a diagram with the raw velocity signal course and the course of the angle error at the commutation sensor; FIG. a diagram with the raw velocity waveform at

Speed change; Fig. 4 is three diagrams showing a moving averaging according to the prior art in the waveform of the angle error;

5 shows three diagrams with an averaging over two samples of the angle error;

6 shows three diagrams with an average of two mean values; FIG. 7 shows four diagrams with a comparison of that shown in FIGS. 4 to 6

methods;

8 is a diagram with an angular waveform.

1 shows the profile of an angle signal 01 measured by the commutation sensor, which shows the magnetic field angle as a function of time and is shown as a continuous line. In contrast, the ideal angle signal 02 is shown as a dashed line. Per revolution from -π to + π sweeps over

Angle signal 01 is a full period. The comparison of the ideal 02 with the measured angular waveform 01 shows that the commutation sensor has a slight

Angular error 03 absorbs, similar to a sinusoidal noise on the

Angle signal around the ideal signal curve 02 oscillates.

In Fig. 2, this angle error 03 is shown as a magnetic field angle difference as a function of time, as well as the raw speed signal 04, which by

Differentiation of the angle signal 01 is obtained. It can be seen here that the ripple to be detected in the course of the angular error 03 becomes increasingly visible in the raw speed signal 04 due to the differentiation, so that the error has a clearer effect. In Fig. 3 it is clear how strong the gain effect of the error in the raw speed signal 04 is. Compared to the actual speed 06 of the

Electric motor occurs at the derived from the angle signal 01 of the commutation sensor raw velocity value an error of about ± 10%. It can also be seen that the frequency of the Storwelligkeit of Roh-speed signal 04 depends on the speed and consequently the inventive determination of half Period of the raw speed signal 04 preferably should be performed continuously for each pass of the method steps in order to achieve an optimal error correction. The three diagrams in FIG. 4 illustrate a method for correcting the raw velocity signal 04 over the moving average based on the course of the angle error 03, which is composed of a plurality of samples 07, as known from the prior art. The averaging here provides a noise-free signal 08, which is represented by a continuous black line, while the Störwell ig speed is maintained, however.

In contrast, the three diagrams in FIG. 5 show the profile of the angular error 03, which according to the present invention is averaged over two samples 09 and 10, ie a sample pair, which are exactly half the period of the angular error 03 apart. This is shown by a dark line

Result signal 12 of this averaging shows a perfect compensation of the

Noise, while the noise is initially maintained almost unchanged.

FIG. 6 shows, in contrast to FIG. 5, a preferred embodiment of the invention, in which not two individual samples are averaged, but to

Averaging two average values over each n, preferably n = 10, samples 09 and 10 are used, which in turn have a spacing of half a period of the angle error 03. In this method results, represented by the continuous black line, a very well-smoothed signal 13, which has neither the noise component nor the interference ripple.

The comparison of the four diagrams in FIG. 7 illustrates the difference between the results of the methods shown in FIGS. 4 to 6 on the basis of real, metrologically detected speed signals. As a result, while the moving-average method with the sliding-average signal 08 solves only the noise component of the raw velocity signal 04, averaging over two individual samples with the noise-free signal 12 corrects the noise ripple of the raw velocity signal 04 and the averaging n pairs of samples with the Smoothed signal 13 as a result removes both the noise component and the noise of the raw velocity signal.

Furthermore, Fig. 7 shows that the time delay in averaging over n pairs of samples 09 and 10 (Τ / 2 + η) is slightly longer than in averaging over two individual samples 09 and 10 (T / 2) and significantly is longer than in the case of pure averaging (n), at least for the averaging values n selected in FIG. 7. If, however, a comparatively good signal were sought with a method according to the prior art, then the temporal filter delay would have to be a multiple of that

Period of the raw speed signal 04 amount. Is the

However, delay constant too large, the raw speed signal 04 can not be used for dynamic reactions. Thus, the speed-dependent averaging proposed by the invention is two

Averages also the delay-optimal method.

FIG. 8 shows a diagram with an angular signal profile 01 according to the invention from -π to + π, wherein the magnetic field angle is shown as a function of time. The value range jumps 14 from + π to -π are indicated by dots and mark the time at which the electric motor starts a new revolution. Consequently, the distance between two value range jumps 14 represents the period of the angle signal 01 again. At each value range jump 14, the speed and thus the period can be redetermined in order to determine therefrom the correct distance of the samples to be used for averaging 09 and 10. From the curve can also be seen that from the angle signal 01 the

Speed change can be derived.

Reference numeral measured angle signal

ideal angle signal

angle error

Raw speed signal actual speed

measured samples

sliding averaged signal first samples of the pairs

second samples of the pairs noise-free signal

smoothed signal

Range jump

Claims

claims

1 . Method for generating a speed signal of an electric motor, which comprises a commutation sensor, comprising the following steps:

- Determining a periodic angle signal (01) of the electric motor by means of the commutation sensor, wherein the angle signal (01) consists of a plurality of samples (07) and has an angle-dependent, periodic angle error (03);

- Transfer of the angle signal (01) in a raw-speed signal (04), which due to the angular error (03) has a Störwelligkeit whose period corresponds to the period of the angular error (03);

- Determination of half the period of the by the noise

periodized raw velocity signal (04);

- Averaging at least one pair of samples (09, 10) of the raw speed signal (04), which are the determined half period of the raw speed signal (04) apart, whereby an average value is obtained; and

- Providing the average obtained as a speed signal.

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

Commutation detected during a complete revolution of the electric motor, a period of the angle signal (01) and two periods of the angular error (03).

3. The method of claim 1 or 2, characterized in that the determination of half the period of the raw speed signal (04) by determining the duration between two immediately successive

Value range jumps (14) of the angle signal (01) takes place.

4. The method according to claim 2, wherein one quarter of the duration between two immediately successive value range jumps (14) of the angle signal (01) is used to obtain half the period of the raw speed signal (04).

5. The method according to claim 1 or 2, characterized in that the determination of half the period of the raw speed signal (04) by a return of each previously provided repeatedly

Speed signal occurs.

6. The method according to any one of claims 1 to 5, characterized in that n> 1 pairs of samples (09, 10) are averaged, wherein in each case the samples (09, 10) of a pair exactly half a period length of the raw velocity signal (04 ) are apart.

7. The method according to claim 6, characterized in that n> 5 pairs

Samples (09, 10) are averaged, wherein in each case the samples (09, 10) of a pair are exactly half a period length of the raw speed signal (04) apart.

8. The method according to claim 6 or 7, characterized in that the respective first samples of the pairs (09) and the respective second samples of the pairs (10) follow one another directly.

9. A speed-controllable electric motor arrangement with a rotary electric motor, which comprises a commutation sensor, and a

A control unit which provides a speed signal of the electric motor and is configured to generate the speed signal by performing a method according to any one of claims 1 to 8.