AT516531A1 - Method for determining the voltage induced in the windings of an electric machine at a constant speed by measuring the induced voltage during the run-out of this electric machine - Google Patents

Abstract

The invention relates to a method for determining the voltage induced in the windings of an electric machine during a constant speed by measuring the induced voltage during the run of the electric machine, comprising the steps of: driving the electric machine at a speed n corresponding to a period T, switching off of the drive and operating the electric machine in the outlet in a generator operation, detecting the induced voltage u (t) at least one phase during the discharge of the electric machine, preferably by digital sampling at a constant sampling frequency, wherein the measured voltage values u (t) an increasing Period duration T and a falling amplitude, conversion of the measured voltage values u (t) in corrected voltage values Ukorr (a) with a constant period and constant amplitude with respect to the rotation angle of the electric machine. This conversion preferably takes place by means of a correction function f (Ti) for converting the voltage values u (t) into periodic voltage values u (a) with a constant period T with respect to the rotation angle of the electric machine, and a conversion of the periodic voltage values u (a) by means of a Correction function aa (a) in the corrected voltage values Ukorr (a) with a constant period and constant amplitude with respect to the rotation angle of the electric machine.

Description

Method for determining the voltage induced in the windings of an electric machine at a constant speed by measuring the induced voltage during the run-out of this electric machine

The invention relates to a method for determining the voltage induced in the windings of an electrical machine during a constant speed by measuring the induced voltage during the outflow of this electric machine.

State of the art

A measurement of the induced voltage in the windings of an electrical machine, which is also referred to as counter-EMF or counter-electromotive force, preferably takes place at a constant rotational speed of the electric machine. From the voltage curve of the measured induced voltage machine characteristics can be determined, which allow an evaluation of the properties and characteristics of the electric machine.

In certain cases, it is technically complicated to drive the electrical machine externally in order to measure the induced voltage. Sometimes this is not possible during the production process. In motor operation, in some arrangements, portions of the back EMF of one phase may be measured, but not a complete periodic waveform.

In order to measure the course of an electrical revolution, or a mechanical revolution, in these cases, the measurement of the induced voltage and the determination of the machine characteristics in the outlet of the electrical machine must be determined. In other words: If the electric machine is brought to speed by self-propulsion, for example, the self-propelled drive is switched off and the induced voltage is measured during the coasting phase of the electric machine. Since the induced voltage at the outlet of the machine shows both a decreasing speed (= increasing period) and a falling amplitude, an accurate determination of the

Machine characteristics, in particular the characteristic values obtained by a frequency analysis, not possible.

The induced stress can not be imaged in the discharge of the machine with previous methods with sufficient accuracy in order to obtain it from e.g. to derive a frequency analysis of the voltage values. For these frequency studies it is necessary to measure the induced voltages in the windings at constant speed.

Disclosure of the invention

It is the object of the invention to specify a method with which the reconstruction of the induced voltage (back EMF) of an electrical machine is possible by means of measurements during the outflow of the electrical machine in order to be able to determine therefrom accurate machine parameters.

This object is achieved by a method having the features of claim 1.

Preferred embodiments and further advantageous features of the invention are specified in the dependent claims.

The method for determining and reconstructing the induced voltage in the windings of the electric machine by measuring the induced voltage during the discharge of the electric machine begins with the fact that the electric machine is operated at a speed n, preferably the nominal speed, either by self-propulsion, ie during engine operation or third-party drive. This speed n corresponds to a period T. After reaching the speed n, the drive is turned off. The electric machine now runs out, for example in a load-free generator mode. While the electric machine is coasting, it passes from the initial speed n to a standstill. During the discharge of the electric machine, the induced voltage u (t) according to the invention is detected on at least one phase, preferably by digital sampling. Particularly preferably, the digital sampling is carried out at a constant frequency. The measured voltage values u (t) have an increasing period T and a falling amplitude due to the decreasing speed of the machine. In a further step, the measured voltage values u (t) are converted into corrected voltage values Ukorr (t) with constant period duration and constant amplitude with respect to the rotation angle a (t) of the electric machine.

Preferably, this conversion comprises a step during which the measured voltage values u (t) are converted by a correction function f (Ti) into periodic voltage values u (a) with a constant period T with respect to the rotation angle a (t) of the electric machine. In a further step, the periodic voltage values u (a) are then converted by means of a correction function aA (t) into corrected voltage values uk0rr (a) with constant period duration and constant amplitude with respect to the rotation angle of the electrical machine. These reconstructed voltage values ukorr (a) correspond to the induced voltage of the electrical machine during operation at a constant speed n. From these corrected voltage values ukorr (a), the machine parameters can be determined with sufficient accuracy in a known manner. For the sake of simplicity, an engine having a rotor with a pole pair will be considered below, so that the mechanical rotation angle ameCh (t) corresponds to the electrical rotation angle aei (t), so that both by means of a rotation angle a (t), with a (t) = ameCh (t) = aei (t), can be designated. However, the invention can also be applied to a rotor with any number of pole pairs.

The method according to the invention makes it possible to improve the restoration of the back EMF from measured values in the outlet of the electrical machine. The resulting voltage-time signal can be used to determine more accurate machine characteristics, which can be obtained in particular by a frequency analysis of the back EMF.

The method is less sensitive to the rate of decay of the measured induced voltage and number of recorded revolutions so that a greater range of motor types can be evaluated using this method. ; According to the invention, the zeros of the actual mechanical revolutions are first obtained from the measured amplitude characteristic of the back EMF during the outflow of the electrical machine. For the exact restoration of the back EMF, it is indispensable to take advantage of the circumstance that a mechanical revolution in the voltage signal shows clearly recurring zero crossings, so that these zero crossings are used as nodes for the rotational angle calculation and calculation of the actual period of one revolution (rotational speed n) can.

The measured voltage-time signal is examined for zeros in order to convert therefrom the time base of the voltage signal to the corresponding rotation angle a (t) of the electrical machine. In this case, only the zero points are primarily relevant, on which the rotor has traveled a full 360 ° turn. The voltage signal in multi-pole machines has several zero crossings in one period. Each zero crossing has a repetition after a mechanical 360 ° rotation. These further periodic-mechanical zero crossings are used to form a correction function f (t), and integrated into the rotation angle-time function so as to obtain additional support points for calculating the correction function. By using an approximation method, for example a spline method, to describe this rotation angle time function, the equivalent rotation angle of the electrical machine can then be determined for each time point of the measured time signal. The function thus obtained then represents the amplitude curve in the form u (a), i. The measured voltage values are now displayed in the form of a voltage-rotation angle function.

After the conversion of the time base and creation of the function u (a), the sampling for this function is no longer equidistant. This is due to the fact that a fixed sampling frequency was used in the digitization, which now results in corresponding "compression" of the sampling signal at different sampling intervals with respect to the angle of rotation a (t), ie different ones Intervals of the samples over the rotation angle a (t). For a further correction of the amplitude of the measured signal or of the corrected signal u (a), the number of measuring points can be increased by a resampling method and the sampling can be recalculated equidistantly.

Thus, one is able to obtain the decay behavior of the amplitude of the measured values determined. For this purpose, the amplitude is determined for each necessary angle value of a period and its mechanical repetitions, and the decay behavior is determined in the form of a correction function aA (a). This function aÄ (a) has as many nodes as there are mechanical periods in the recorded signal. By using an approximation method, for example a spline method, to describe this decay function aA (a), the amplitude value is determined at the time of the first measured value. The function Ukorr (a) is determined from the functional set of the amplitude values of the measured back EMF. This function corresponds to a measurement of the back EMF at a constant speed n.

To determine the initial rotational speed n of the machine, the associated angular velocity is calculated from the correction function f (Ti) via the initial slope, and thus the rotational speed of the back EMF for the first measured value acquired can be determined. From the speed n and the function uk0rr (a), the back EMF can be reconstructed and used for the calculation of motor characteristics. In particular, the method according to the invention can be used to determine such engine parameters that can be obtained from a frequency analysis.

For simplicity, the method is described only for a measurement of a motor phase. However, it can also be used accordingly for multi-phase motors and the measured values of several phases of an electrical machine.

In the case of polyphase motors, it is thus also possible to apply the method individually for each motor phase so that the motor parameters can be determined more accurately with the aid of statistical methods.

Furthermore, in multiphase motors, the sequence of zero crossings of the voltages induced in the phase windings can be used to verify the internal wiring of the phase windings of the motor. For such an extension of the method according to the invention, it is necessary that the direction of rotation is known. This can be achieved for example via a third-party drive, wherein the motor is not driven by means of its own phase windings. For example, in a measurement of the voltages induced in the phase windings, after registering a positive zero crossing Uo of a phase U, a corresponding positive zero crossing V0 of a phase V and then a positive zero crossing W0 of a phase W are registered. If the sequence of zero crossings Uo, V0 and W0 coincides with the expected sequence in the selected direction of rotation, the interconnection of the phase windings can then be assumed to be correct. The measurement can be determined for example by means of a third-party drive at a constant speed or in the outlet and therefore also be combined with the inventive method. Crucial here is only that the direction of rotation of the engine is known.

The invention will be described in more detail in the form of an embodiment with reference to the drawings. This results in further features and advantages of the invention.

Brief description of the drawings:

Figure 1 shows schematically an electrical circuit diagram of a phase of a brushless DC motor together with a schematic circuit for driving.

FIG. 2 schematically shows a voltage-time profile of the voltage induced in the winding of FIG. 1 at the outlet of the electrical machine. FIG. 3 shows a diagram of the mechanical machine angle as a function of time.

FIG. 4 shows a refined diagram of the mechanical machine angle as a function of time.

FIG. 5 shows diagrams for the conversion of the voltage-time curve according to FIG. 2 in a voltage-rotation angle diagram.

FIG. 6 shows a diagram of the voltage as a function of

Rotation angle and a corresponding correction function for the voltage amplitudes.

FIG. 7 shows a diagram of the amplitude-corrected voltage as a function of the angle of rotation.

FIG. 8 shows a diagram of the voltage induced in the phases U, V and W of a three-phase motor as a function of the mechanical rotor rotation angle.

FIG. 9 shows a diagram of those induced in the phases U, V and W.

Voltage as a function of the mechanical rotor rotation angle for the same engine as in Figure 8 in the opposite direction.

Description of a preferred embodiment of the invention:

Figure 1 shows schematically a strand of the electric machine 10 having a phase winding 12 which is connected to a drive circuit with four power switches S1 to S4. The power switches S1-S4 are arranged in a known manner in an H-bridge circuit and connect the winding 12 alternately with positive or negative operating voltage in dependence on the mechanical position of the rotor. During generator operation, the winding 12 generates an induced voltage u during rotation of the rotor, the amplitude and period duration of which depend on the rotational speed of the electrical machine.

FIG. 2 shows a typical voltage curve as a function of time for a voltage induced in the winding 12 at the outlet of the motor. Since the electric machine is not driven at the outlet, the speed decreases steadily and thus the period between the individual revolutions becomes larger (compare T1, T2, T3). At the same time, the amplitude of the voltage u decreases over time t.

A frequency analysis of this voltage signal u (t) shown in Figure 2 is not possible due to the changing amplitude and period T.

The invention is now based on reconstructing the induced voltage u (t) as if the third-party electric machine were operated at a constant speed.

According to the invention, the electric machine is operated without load in generator operation in the outlet. The drive in the outlet is done by the own flywheel of the rotor of the machine, whereby speed and amplitude of the induced voltage drop steadily during the measurement.

The voltage-time diagram of the voltage u (t) induced in the winding is distorted by the changing speed n during the measurement and the simultaneously decreasing amplitude of the induced voltage due to the speed change of the rotating mass. In order to restore the induced voltage as if measured in the driven mode at constant speed n, it is necessary to correct the voltage measurement values u (t) and the time sampling of the measurement data. In non-synchronous sampling of the measurement channels, e.g. in multiplex mode, the time offset of the data is to be taken into account and corrected, for example when several phases of an electrical machine are measured in one measuring run.

However, to simplify the inventive principle, only the measurement of one phase will be described in the illustrated example. This general description can also be applied to the measurement of several phases of the electric machine.

When measuring the induced voltage in several phases, it must be taken into account that the phases are not necessarily exactly symmetrical. Due to winding inequalities an imbalance is to be expected, as well as by a non-symmetrical structure of the mechanical system or the magnet system.

Furthermore, due to the system-induced induced voltage due to the hammer shape of the stator, the pole transitions, the saturation effects of the magnet system no pure sinusoidal voltage is generated, so that is to be expected with harmonics. The zero crossings of the curves of the induced voltage are also not symmetrical but follow the mechanical inequalities of the arrangement of the hammers.

The course of the magnetic flux over the angle of rotation is influenced by the mechanical structure.

The measured induced voltages of the individual phases are the phase voltages. In a star connection of the electrical machine can not always be measured to the neutral point of the windings. The measured induced phase voltages are the sum of the individual induced pole-pair windings and can not be separated individually and must be treated as a sum.

The above-described problems in the measurement in the outlet complicate an exact simulation of the induced voltage from the data of the induced voltage measured in the outlet. Therefore, an approximation method is used which simulates as accurately as possible the induced voltage from the measurement and the induced voltage in the outlet.

First, the measured voltage in the outlet is converted to a constant speed. The procedure assumes that the measured data in the undisturbed outlet can be measured. An outage malfunction due to non-RPM events would result in errors in this procedure.

In a first step, the zero-crossings, which correspond in each case to one period of one revolution T1, T2, T3 of the electrical machine, are respectively determined from the voltage-time diagram according to FIG. The goal is to determine from the zero crossings of the voltage signal u (t) the time base of the phase voltage with respect to one mechanical revolution. In this case, only the zeros or zero crossings are relevant in which the rotor has covered a complete revolution, see T1, T2, T3, etc.

FIG. 3 shows a diagram in which the rotation angle a (t) is represented mechanically in the unit [rad] over the time t. There are always the corresponding revolution periods Ti = T1, T2, T3, shown, each corresponding to a full mechanical revolution, ie an angle of 2π, 4 π, 6π. It can be seen that this relationship between the time t and the mechanical angle [rad] is not linear, but due to the growing period Ti at the outlet of the electric machine represents a curved curve with decreasing slope.

From individual support values, ie (T1, 2π) (T2,4π), the function f (Ti) can now be created by an approximation method. With this function, one can then convert the time axis t into the rotation angle a (t) into radians, that is, convert the temporal voltage measured values to the mechanical positions of the rotor.

By using additional support points, this temporal correction function can be improved, as shown in FIG. The induced voltage of the back EMF has even more zero crossings, which can be taken into account. Each zero crossing has a periodicity which is due to the mechanical rotation of the electric machine. Thus, this periodicity is present in the measurement signal for each zero crossing. These additional zero crossings can serve as additional vertices for the function f (Ti) and improve the accuracy of this function. Of these further zero crossings, their periodic distance, that is to say the mechanical angle of multiples of π, is known, but not the angle to the other zero crossings, which are illustrated for example in FIG. Such an angle di, with i = 0, 1, 2, 3,..., Is shown in FIG. For example, d0 describes the angle between a zero crossing at time to and an additional zero crossing at time t0 '. The periods t0, t1, t2,..., Correspond to the periods T1, T2, T3,... Of FIG. 3. The mechanical angle between the zero crossings of the entire periods T1, T2, T3, The additional zero crossings can be determined by means of the zero-crossing curve according to FIG. The times of the further zero crossings now yield an angle which can be inserted into the function f (Ti). Theoretically, this should be

Insertion angle to the zeros or zero crossings of the first zero-crossing curve to be constant, but by the used approximation function, this angle can vary for each zero crossing. It has been found that all inserted angles of the zero-crossing curve averaged over a median give the best value for a constant insertion angle. Extended by these vertices of all possible zero crossings, the accuracy for forming the function f (Ti) is substantially improved.

The voltage values u (t) are now related to the mechanical rotation angle by means of this correction function f (Ti). The time axis of the voltage measured values is converted into mechanical angles using the function f (Ti). The amplitudes are not changed in this step.

FIG. 5 shows the voltage values u (t) in the upper diagram and the voltage values u (a) converted to the mechanical angles in the lower diagram.

In a next step according to FIG. 6, a decay function aA (a) must now be determined. The amplitude of the voltage measurement values u (a) is constantly decreasing due to the falling speed in the outlet of the electric machine. To convert these voltage readings to a constant speed, the amplitude of function u (a) must be converted to a constant speed using a correction function aa (a). The nodes of the decay function aA (a) are each considered at an angle Δ after the periodic zero crossings of the function u (a). In principle, additional additional interpolation points can be considered at an angle Δ after additional zero crossings of the induced voltage. For the amplitude correction, the associated voltage value is determined for each amplitude value to be calculated over the angle axis and for each period. From these interfaces, a function is calculated with the angle value 0 rad, that is, the first zero point in the voltage signal, extrapolated and thus the amplitude at the rotational speed n (t = 0) can be reconstructed. Here, it must be determined with which sampling rate the voltage curve is resampled - - w w w m in order to be able to interpolate the amplitude as accurately as possible for each desired angle value. The periodic amplitude values are used to calculate the decay function θΔ (α). Preferably, a spline approximation is used here. The amplitude of the voltage function u (a) can then be extrapolated to the angle value 0 rad by the determined decay function a & (a). The more nodes are used to calculate the decay function, the more accurate the calculation of the amplitude value of the voltage u at time t = 0 or for u at the speed n. It is also possible to calculate a plurality of decay functions aa (a) at different values of Δ, so that, for example, an average decay function for amplitude reconstruction can be determined by means of statistical methods.

After the curve u (a) has been reconstructed, the speed n of this reconstructed curve can now be determined at time t = 0. The time information is only in the sampled signal and is based on the value of the sampling rate. By means of the conversion function f (Ti) used, the speed of the reconstructed back EMF curve can be determined on the basis of the slope of the conversion function at the zero point (a = 0, t = 0). The speed n then results from n = (da / dt) * 1 / (2 Pi) at the time t = 0.

Due to the approximation method used, preferably a spline interpolation, for the individual functions, the interpolation points are exactly hit by this function, the interim values having deviations from the real value due to the interpolation. To reduce these deviations in the reconstruction of the measured data with constant speed, you should use as many nodes as possible, so at least all zero crossings and other nodes, to reconstruct the curve.

A higher number of rated periods when the electric machine is coasting also makes the calculation of the amplitudes to constant speed more accurate.

Preferably, for a good accuracy of the amplitude reconstruction should

Amplitude can not be calculated to the first zero crossing. In this case, the approximation function for determining the amplitude values has no further interpolation point before the first zero crossing, so that an extrapolation with this approximation function is more inaccurate than if we were to distinguish between two known ones; Interpolated interpolation points. For this reason, the amplitude reconstruction should, for example, be placed between the two zero crossings on the second period in order to be able to interpolate the amplitude profile therefrom. For a large number of zero crossings of the detected signal, a high pool pair number of the electrical machine is also favorable, so that the reconstruction of the signal becomes more accurate. For measuring data acquisition, for example, 16 mechanical revolutions of the electric machine in the outlet are sufficient. With this number of revolutions, a sufficiently good accuracy of the reconstruction is achieved. With less than 8 mechanical revolutions involved, the method described becomes inaccurate. Preferably, therefore, the measurement duration for the voltage values u = f (t) should be between 8 and 16 mechanical revolutions of the electrical machine. However, if too much accuracy is not required, 4 mechanical revolutions may be enough to determine the measurements.

In the case of polyphase motors, the method according to the invention can be applied individually for each motor phase, for example, so that the motor characteristics can be determined more accurately with the aid of statistical methods. FIG. 8 shows by way of example the voltage induced in the phases U, V, W of a three-phase motor and the sequence of the positive zero crossings Uo, Vo and Wo. Also in this example, a period of the induced voltages corresponds to a mechanical revolution of the rotor of a motor having a pole pair. The induced voltage BEMF in the respective phase is in each case shown normalized above its maximum value BEMFmax. In the event that the direction of rotation of the motor is known, for example by being driven by external phase windings, the sequence of zero crossings Uo, V0 and W0 can also be used to check the interconnection of the

Phase windings are used. For this purpose, the measured sequence of zero crossings is compared with the expected sequence. Since the direction of rotation is known, it can be concluded that there is a fault in the connection of the motor in the event of a deviation of the measured from the expected sequence of zero crossings. If the zero crossings are expected, for example, in the case of a clockwise-rotating motor in the order Uo, V0, W0, then FIG. 8 shows an idealized measurement of a correctly interconnected motor. In contrast, in Figure 9, the order of the signals of the phases Vo and Wo reversed, and thus assume an error in the interconnection.

claims:

Claims (10)

claims A method of detecting the induced voltage in the windings of an electric machine at a constant speed by measuring the induced voltage during the coasting of the electric machine, comprising the steps of: driving the electric machine at a speed n corresponding to a period T, turning off the drive and operating the electric machine in the spout, detecting the induced voltage u (t) on at least one phase during the run of the electric machine, wherein the measured voltage values u (t) have an increasing period T and a falling amplitude, and converting the measured voltage values u (t) in corrected voltage values Ukorr (a) with a constant period and constant amplitude with respect to a rotation angle α of the electric machine. 2. The method according to claim 1, characterized in that the conversion of the measured voltage values u (t) into the corrected voltage values Ukorr (a) comprises a step for converting the measured voltage values u (t) into periodic voltage values u i by means of a correction function f (Ti) (a) with a constant period T with respect to the rotation angle α of the electric machine, and a step for converting the periodic voltage values u (a) by means of a correction function aA (a) into the corrected voltage values Ukorr (a). 3. The method according to claim 1 or 2, characterized in that the correction function aA (a) describes a decay function, which is determined from values with an angular distance of in each case multiples of 2π of the periodic voltage values u (a). 4. The method according to any one of claims 1 to 3, characterized in that the rotational speed n from the slope of the correction function f (Ti) is determined at zero point. 5. The method according to any one of claims 1 to 4, characterized in that the correction function f (Ti) from the voltage-time values of the zero crossings of the measured voltage signal u (t) is determined, in which the electric machine has completed a multiple of a mechanical revolution , 6. The method according to any one of claims 1 to 5, characterized in that for calculating the correction function f (Ti) additionally voltage-time values of further zero crossings of the measured voltage signal u (t) are used. 7. The method according to any one of claims 1 to 6, characterized in that the voltage values u (t) are measured by digital sampling with a constant sampling frequency. 8. The method according to any one of claims 1 to 7, characterized in that the electric machine is brought to self-propelled speed. 9. The method according to any one of claims 1 to 7, characterized in that the electric machine is brought to speed in the third-party drive. 10. The method according to any one of claims 1 to 9, characterized in that the electric machine is operated in a load-free generator mode. Ls. Gibler & Poth Patent Attorneys OG (Dr. F. Gibler or Dr. W. Poth)