WO2008146996A1

WO2008146996A1 - Compensation method of zero-current-clamping effect in pulsating carrier-signal injection-based sensorless drives

Info

Publication number: WO2008146996A1
Application number: PCT/KR2007/005632
Authority: WO
Inventors: Jul-Ki Seok; Chan-Hee Choi
Original assignee: Industry-Academic Cooperation Foundation, Yeungnam University
Priority date: 2007-05-30
Filing date: 2007-11-09
Publication date: 2008-12-04
Also published as: KR20080105364A; KR100877599B1

Abstract

The present invention relates to a method of compensating a zero- current-c lamping effect in an alternating current (AC) motor without position and speed sensors. According to this method, a high frequency current data is collected from an off-line test for a high frequency injection sensor less drive AC motor so as to obtain a d-q axis inductance, and then a distortion coefficient caused by a zero-current-c lamping effect is found based on the obtained d-q axis inductance. A resultant coefficient value obtained by multiplying the found distortion coefficient by an actual high frequency current allows the zero-current-clamping effect to be compensated through an on-line compensator in an actual operation situation, and the compensation voltage calculated by the on-line compensator is out put ted while being limited to the maximum distortion voltage by a switching element of an inverter.

Description

[DESCRIPTION]

[Invention Title]

COMPENSATION METHOD OF ZERO-CURRENT-CLAMPING EFFECT IN PULSATING CARRIER-SIGNAL INJECTED-BASED SENSORLESS DRIVES

[Technical Field]

<i> The present invention relates to a method of compensating a zero- current-clamping effect in an alternating current (AC) motor without position and speed sensors, and more particularly, to a method of compensating a zero- current-clamping effect in an alternating current (AC) motor without a position and speed sensor, which can improve an error of an estimated position angle or a speed acceleration/deceleration performance by compensating a zero-current-clamping effect in a method of injecting a sinusoidal high frequency signal so as to estimate the position and speed.

<2>

[Background Art]

<3> The greatest merit of an alternating current (AC) drive system is that there is no problem in repair and maintenance due to mechanical abrasion of a brush and a commutator bar in an existing direct current (DC) motor drive system. Owing to such rigid structure and economic advantage, the drives of the AC motor using an inverter is gradually increasing recently in various industrial application fields requiring a high-performance control.

<4> The basic requirements for the high-performance control are that torque current and flux current are controlled independently and the two components are spatially orthogonal to each other. To this end, the position of a magnetic flux should be grasped. The position of the magnetic flux can be directly measured by a flux position detector or can be grasped through estimation of the flux position based on the measurement of the rotational speed of a rotor. However, such position and speed detectors entail the following problems.

<5> First, a rotor position detector and the cost spent for its attachment causes an increase of the price of a motor drive system. Also, signals received from the rotor position detector are converted into signals applicable for the control through various methods. An electronic circuit for this purpose contributes to complexity of a control system. Since a signal generated from the rotor position detector is generally susceptible to an electromagnetic noise, erroneous information may be provided, which becomes a factor degrading stability of a drive system. An environment and the like in which the mechanical structure or the drive system is installed makes it to attach the rotor position detector to a rotary shaft of the motor.

<6> Likewise, since there occur several problems in grasping the speed of the rotor or the position of the flux for high-performance control of the AC motor, a research has been recently in progress to obtain the position of the flux without the position and speed detector or the flux detector. This control method is referred to as a sensorless control method.

<7> Examples of such a sensorless control method include U.S. Patent Nos. 5,886,498 and 6,069,467. The U.S. Patent No. 5,886,498 discloses a pulsating injection method in which a high frequency fluctuating signal is injected to the estimated flux axis, and the U.S. Patent No. 6,069,467 discloses a synchronous injection method in which a flux axis stator command signal is injected in a synchronous rotor-flux frame, and the drive system uses the injected stator command signal to provide injected power to the induction machine.

<8> Such a synchronous injection method is one in which a high frequency voltage is injected in a still coordinate system as shown in FIG. 1. This method employs a negative-sequence component of a high frequency current. The negative sequence component is very small in size, and is problematic in that a significantly serious distortion occurs when there is a zero-current- clamping phenomenon.

<9> In addition, the pulsating injection method is one in which a high frequency voltage of a given level is injected in an estimated synchronous coordinate system as shown in FIG. 2. In such a method, the high frequency voltage is injected to extract an angular error from a high frequency current inside a motor to thereby estimate an angle. This method is advantageous in that it is very resistant to the zero-current-clamping, enables a zero-speed full-load operation which was impossible in an existing voltage integration type sensorless control method, and does not require a motor constant.

<io> In the above AC motor, precision control is directly proportional to the accuracy of position and speed information. In case of the sensorless control method, since an actual position and speed controller is removed, an accurately estimated position and speed which are comparable to the position and speed measured by an actual position and speed sensor must be obtained using a high frequency.

<π> However, as shown in FIG. 3, the zero-current-clamping effect occurs at every time point when a phase current passes the zero point due to the dead time of a Pulse-Width Modulation inverter for driving the motor and the parasite capacitance of a switching element.

<i2> This effect has a nonlinear property that a distortion voltage value (nonlinearIy inverse-proportional to the equivalent transient time) varies depending on the amount of the current, as shown in FIG. 4, and the equivalent transient time has a maximum value near the zero current and is gradually reduced as a current value increases.

<i3> This zero-current-clamping effect distorts the high frequency current including rotor position information to thereby cause distortion of a position signal that is synchronized at every zero point of the current as shown in FIG. 5. Such a position error causes degradation of the performance of the motor as well as induces system diffusion. In addition, an estimated speed necessary for speed control is obtained through differentiation of an estimated position. The position error, shown in FIG. 5 causes a ripple of the estimated speed to thereby bring about a trip of the drive system due to an overcurrent .

<i4> In order to temporarily address and solve this problem, it is required that the bandwidth of an estimated angle measurement unit and a speed controller should be decreased. This contributes to a remarkable deterioration of the acceleration/deceleration performance of the motor drive system, resulting in degraded productivity and accuracy of the motor.

<i5> As an alternative method, there is proposed a look-up table method in which data, which are obtained through long-term tests, are arranged according to the experiences of an expert. This method encounters a demerit that it requires a long development time and an additional memory device for storing data, and hence has many limitations in the direct application to actual industrial fields. [Disclosure] [Technical Problem]

<i7> Accordingly, an object of the present invention has been made to solve the above-mentioned problems occurring in the prior art, and it is an object of the present invention to provide a method of compensating a zero-current- clamping effect in an alternating current (AC) motor without position and speed sensors, in which a voltage is injected in an estimated synchronous coordinate system, an off-line test for compensation and an on-line compensation are performed in a still coordinate system in such a fashion that after a simple off-line test has been carried out by using a load condition replication technique which replicates a load condition due to a difficulty in finding a distortion coefficient in case of a non-load in the off-line test, a zero-current-clamping effect is compensated in real-time to reduce a ripple of an estimated angle or speed for sensorless control, thereby improving stability and performance of a sensorless system. [Technical Solution]

<i8> To accomplish the above object, according to the present invention, there is provided a method of compensating a zero-current-clamping effect in an alternating current (AC) motor without position and speed sensors. According to this method, a high frequency current data is collected from an off-line test for a high frequency injection sensorless drive AC motor so as to obtain a d-q axis inductance, and then a distortion coefficient caused by a zero-current-clamping effect is found based on the obtained d-q axis inductance. A resultant coefficient value obtained by multiplying the found distortion coefficient by an actual high frequency current allows the zero- current-clamping effect to be compensated through an on-line compensator in an actual operation situation, and the compensation voltage calculated by the on-line compensator is outputted while being limited to the maximum distortion voltage by a switching element of an inverter. [Advantageous Effects]

<2i> As described above, the present invention has an advantageous effect in that after a simple off-line test without a separate device, a zero-current- clamping effect is compensated in real-time to reduce a ripple of an estimated angle or speed for sensorless control, thereby improving stability and performance of the sensorless system. [Description of Drawings]

<22> FIG. 1 is a block diagram illustrating a synchronous injection method of a senseless AC motor;

<23> FIG. 2 is a block diagram illustrating a pulsating injection method of a senseless AC motor;

<24> FIG. 3 is a graph showing the relationship between the position of the basic wave current according to the zero-current-clamping effect and the current in a still coordinate system;

<25> FIG. 4 is a graph showing distorted voltage values according to the amounts of phase current;

<26> FIG. 5 is a graph showing the relationship between time and current/rotor position according to the zero-current-clamping effect;

<27> FIG. 6 is a graph showing the relationship between rotor position and phase current/high frequency current under no-load condition;

<28> FIG. 7 is a graph showing the zero-current-clamping effect under a load condition;

<29> FIG. 8 is a block diagram showing a zero-current~clamping compensating unit according to the present invention; <30> FIG. 9 is a view showing the construction of a compensator of FIG. 8; <3i> FIG. 10 is a graph showing a no-load test waveform in case where the zero-current-clamping effect is not compensated; <32> FIG. 11 is a graph showing a no-load test waveform in case where the zero-current-clamping effect is compensated; <34> FIG. 12 is a graph showing a waveform before the zero-current-clamping effect is compensated upon the application of a load; <35> FIG. 13 is a graph showing a waveform after the zero-current-clamping effect is compensated upon the application of a load; and <36> FIG. 14 is a graph showing the relationship between time and compensation voltage/phase current.

[Best Mode] <38> Reference will now be made in detail to a preferred embodiment of the present invention with reference to the attached drawings. <39> In the following embodiment, superscript/subscript 's' denote a still coordinate system, and superscripts/subscripts 'r' and 'e' denote synchronous coordinate systems. Both 'r'and 'e'are used herein.

<40>

<4i> 1. Modeling in still coordinate system

<42> When a phase current passes the zero point, the output voltage of an inverter is distorted by a parasite capacitance of an insulated gate biopolar transistor (IGBT). This phenomenon is called a zero-current clamping.

<43> A voltage error (Vd) caused by the zero-current clamping is expressed by the following Math Figure 1: [Math Figure 1]

<44>

<45> where Ten designates an effective dead time, Ttr designates an equivalent trailing time by a parasite capacitance near the zero point of the current, Ts designates a sampling time, and Vdc designates a DC link voltage. <46> FIG. 4 shows an example of a typical Ttr curve. This non-linear curve is approximated by the following Math Figure 2. In this example, Math Figure 2 can be approximated in such a manner that a = Ten and b = 1. When a higher order term in Math Figure 2 is ignored due to a very small amount of current

and Math Figures 1 and 2 are combined together, a distortion voltage ^c ' by

a high frequency current is approximated by the following Math Figure 3:

[Math Figure 2]

T » JL± ⁴ ° L* -4bi³ + 6b²i² - 4b³i + b⁴

[Math Figure 3]

<49> where α denotes a distortion coefficient cased by the zero-current clamping and has different values depending on a pulse width modulation (PWM) inverter.

<50> It can be seen from Math Figure 3 that is proportional to the high frequency current. <-^*>•> If a frequency ω_c of a high frequency voltage being injected is sufficiently high, the relationship between the high frequency current and the high frequency voltage can be expressed by the following Math Figure 4: [Math Figure 4]

<52>

<53> where denotes an injection voltage command which is expressed by the following Math Figure 5, ω_c denotes a frequency of an injection voltage

and denotes an inductance matrix of a target motor. In the meantime, the above Math Figure 4 is a mathematical expression which is satisfied each time the phase current passes the zero point. [Math Figure 5]

<55> For example, each time an A-phase current passes the zero point, a high frequency voltage matrix in the still coordinate system can be expressed by the following Math Figure 6 using Math Figure 4: [Math Figure 6]

<57>

<58> In the above Math Figure 6, s denotes a Laplace operator, θ_r denotes a rotor angle, L_s and ΔLs denote an average inductance and a difference inductance in the still coordinate system, respectively.

<59> <60> 2. Off-line test in still coordinate system <6I> As elucidated in the modeling in the still coordinate system, an accurate estimation of the distortion coefficientC α ) is a key point of compensation of the zero-current-clamping. But, since there occurs no distortion at the zero point of current as shown in FIG. 6 upon no application of a load, it is very difficult to find a distortion coefficient.

<62> For this reason, as shown in FIG. 7, a position error ( π /6 in this embodiment) is forcibly added to an estimated angle upon no application of a load so as shift the distribution of the high frequency current and the zero point of the phase current.

<63> In this case, since the estimated angle and an actual angle are shown as third and forth waveforms of FIG. 7 and the actual angle coincides with an angle of the phase current, the preset invention has an advantage in that it is possible to know an actual angle without an encoder (position sensor).

<64> In FIG. 6, i_as denotes an a-phase current in a still coordinate system,

S id_h denotes a high frequency current of a d-axis in the still coordinate

system, denotes a high frequency current of a q-axis in the rotor coordinate system. Also, FIG. 7, θ_rτr denotes an actual angle of a rotor and

θ. an estimated angle of the rotor.

<65> In order to perform an off-line test, an average inductance Ls and a difference inductance ΔLs must be first obtained in the still coordinate system. To this end, the injection voltage command is expressed as in the following Math Figure T- [Math Figure 7]

<67>

<68> Also, since there occurs no distortion at a peak portion of the phase current, a high frequency current at this peak portion is expressed by the following Math Figure 8: [Math Figure 8]

L_ss

<69>

<70>

<7 I > I Inn tthhii :s s ccaassee,, iif θ_r=0° peak region), the high frequency current of a d-q axis in the still coordinate system in Math Figure 8 is obtained by the following Math Figures 9 and 10: [Math Figure 9]

V_f1 cos ω_ct cos θ ldh 0 s(L_s - AL_x )

<72>

[Math Figure 10]

Ϊ S V_h cos ω_c/^t sin θ qh O s(L_s + AL_x )

<73> <74>

<75> In the above Math Figures 9 and 10, the maximum value of the high frequency current is generated when ω_ct=0° In this case, Math Figures 9 and

10 are changed to the following Math Figures 11 and 12: [Math Figure 11]

<76> [Math Figure 12]

V_h sin θ lqh max o' co_a (L _s H- A/,_y )

<77> <78>

<79> Further, when the above Math Figures 11 and 12 are combined together, two inductances can be obtained by the following Math Figures 13 and 14: [Math Figure 13]

<80>

[Math Figure 14]

dh max

o- J

<82>

<83> A denominator of Math Figure 5 is expressed by the following Math Figure 15:

[Math Figure 15]

\L_ss + ex = S(L _S — AL _S cos 2 θ_r ) + OC — i'ΔI_y sin 2θ _r

— sALj, sin 2θ_r ^S(L_$ + ^^-.y cos 20,. ) -+- oc

<84> <85>

<86> Thus, the high frequency current in still coordinate system expressed in Math Figure 4 can be re-expressed by the following Math Figures 16 and 17: [Math Figure 16]

<87>

[Math Figure 17]

<88>

<89> Also, when θ_r=90° zero point), a measured high frequency current of a d-axis expressed in Math Figure 16 is expressed by the following Math Figure 18:

[Math Figure 18]

<90> <91>

<92> Since the maximum value of the high frequency current of the d-axis is generated when ω_ct =0° Math Figure 18 is re-expressed by the following Math

Figure 19:

[Math Figure 19]

<94>

<95> In addition, a distortion coefficient from Math Figure 19 is uniquely determined by the following Math Figure 20: [Math Figure 20]

<97>

<98> 3. Real-t ime compensat ion method

<99> A compensat ion vol tage from Math Figure 3 must be determined by the fol lowing Math Figure 21 : [Math Figure 21]

^Y h _com ^{== 0}^h -

<I00>

<ιoi> In the present invention, voltage compensation is performed in a still coordinate system. FIG. 8 is a block diagram showing a zero-current-clamping compensating unit for implementing a compensating method according to the present invention. In FIG. 8, of coursed, an angle controller may employ a bang-bang type estimator, a PI estimator or a tracking observer.

<iO2> A zero-current-clamping compensating unit shown in FIG. 8 permits a subtracter 16 to subtract a basic current command i_dq being inputted and a

current i_dq obtained by allowing a current * of a motor in a still

coordinate system to pass a band stop filter 12 and a coordinate converter 14. The subtracted current command is inputted to a current controller 18 r_* which in turn outputs a voltage command V_dq so as to allow the current

command to be controlled to a predetermined current. Then, the outputted r* r_* voltage command V_dq and a high frequency voltage V_dh are added by an adder

com

22 and then a compensation voltage generated from a compensator 30 is added by an adder 42 after conversion into a still coordinate system through

V₁ , * a coordinate converter 24. The added voltage command ( ) is applied to a motor 46 through a PWM inverter 44. <iO3> In the meantime, in the above construction, the compensation voltage

rcont generated from the compensator 30 becomes a value obtained by

rcont allowing an adder 32 to add a compensation voltage ( ) for an injected high frequency voltage determined by a value obtained by multiplying, a distortion coefficient (α) caused by a zero-current-clamping effect, which is found based on a high frequency current data collected through an off-line test to obtain a d-q axis inductance, by an actually measured high frequency

current ^A of a motor 46 after allowing the actually measured high frequency current to pass a bandpass filter 52, and a compensation

voltage — for a basic wave voltage determined by a current value

4/ obtained by allowing an output current of an effective motor to pass a band stop filter 12, as shown in FIG. 9.

<iO4> Finally, the compensation voltage has a feature that it is outputted while being limited to the maximum distortion voltage by a switching element of an inverter through a limiter 34.

<105>

<i06> 4. Test result

<iO7> The method of the present invention was performed on a 600-W permanent magnet synchronous motor (PMSM) for test. A PWM inverter is switched at

1OkHz and has a dead time of 2μs . The magnitude and the frequency of an injected voltage were set as 10V and 850Hz. <i08> First, a high frequency current is measured to obtain an inductance in an off-line test. In this case, the speed of a motor is 30r/min, and the _

magnitude of the d~axis phase current is 2A. Thereafter, a distortion coefficient is found using Math Figure 18.

<i09> FIG. 10 is a graph showing a no-load test waveform in case where the zero-current-clamping effect is not compensated.

<iio> In FIG. 10, an A-phase current, a q-axis high frequency current of an estimated synchronous coordinate system, an actual position and an input signal (εsin)in of an angular estimator are shown in the order of the upper waveform to the lower waveform. It can be seen from this test that although the magnitude of the high frequency current becomes nearly '0' at the zero point of the phase current, a small distortion occurs still. Of course, when a load is applied, larger distortion may occur.

<iii> FIG. 11 is a graph showing a no-load test waveform in case where the zero-current-clamping effect is compensated.

<ii2> In this case, the test of the same condition was carried out under a compensation condition. The compensation voltage is added to a final output voltage as shown in FIG. 8. It can be seen from the graph of FIG. 11 that a compensation effect appears more absolutely as compared to that of FIG. 10.

<ii3> FIG. 12 is a graph showing a waveform before the zero-current-clamping effect is compensated upon the application of a load. FIG. 14 is a graph showing a waveform after the zero-current-clamping effect is compensated upon the application of a load. It can be seen that a ripple of a distortion is remarkably reduced after compensation of the zero-current-clamping effect as a result of comparison of the two waveforms of FIGs. 12 and 13. In both cases, a load was applied for about two seconds.

<ii4> FIG. 14 is a graph showing the relationship between time and compensation voltage/phase current.

<ιi5> When a phase current passes beyond a zero-current-clamping region, there appears a waveform which maintains a constant value. When the phase current passes a region near the zero point, there appears a waveform which has the same frequency and phase as those of the high frequency current due to an influence of a high frequency component. <ii6> Although the specific embodiments of the present invention have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims.

<117>

Claims

[CLAIMS] [Claim 1]

<ii9> A method of compensating a zero-current-clamping effect in an alternating current (AC) motor without position and speed sensors, the method comprising:

ream

<i20> a step of outputting a compensation voltage ( ) as a value obtained by allowing an adder (32) to add a compensation voltage

( _com ) f_{or an} j_nj_ecte(} J₁Jg]₁ frequency voltage determined by a value obtained by multiplying, a distortion coefficient (α) caused by a zero- current-clamping effect, which is found using Math Figure 20 based on a high frequency current data collected from an off-line test for an alternating current (AC) motor without position and speed sensors in a still coordinate system to obtain a d-q axis inductance using Math Figures 13 and 14 through a

_* compensator (30), by an actually measured high frequency current ( ) of a motor (46) after allowing the actually measured high frequency current to pass a bandpass filter (52), and a compensation voltage for a basic wave

voltage determined by a current value ( ) obtained by allowing an output current of a motor to pass a band stop filter (12), the compensation voltage being outputted while being limited to the maximum distortion voltage by a switching element of an inverter through a limiter (34);

<i2i> a step of adding a pulsating high frequency voltage (V_dh ) to a voltage r* command (V_dq ) outputted from a current controller (18) in a synchronous

coordinate system through an adder (22); and <i22> a step of adding the voltage command (V_dq ) which has passed the adder

(22) through a coordinate converter (24) and the compensation voltage _

ream

( ) outputted from the compensator (30) while being limited to the maximum distortion voltage by a switching element of an inverter through an adder (42) after conversion into a still coordinate system through a coordinate converter (24), and outputting the added compensation voltage to the motor (46) through an inverter (44).

[Claim 2] The method according to claim 1, wherein before the d^~q axis inductance is obtained through the compensator (30), a position error is forcibly added to an estimated angle of a rotor of the motor under no-load condition so as to shift the distribution of the high frequency current and the zero point of the phase current.