JPH0452080B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method of controlling a polyphase inverter connected to a polyphase AC load.

[Conventional technology]

It is known to connect an alternating current motor with a variable frequency and voltage inverter. It is also known to control this inverter using PWM (pulse width modulation). By the way, when driving a polyphase inverter using PWM control to obtain an approximate sine wave, if each phase is modulated, it will be susceptible to the influence of other phases, making it impossible to obtain the optimum output. That is, since the line voltage of the output of the inverter is determined by the difference between the phase potentials, it is virtually impossible to make each line voltage approximate a sine wave at the same time.

As a method to solve the above-mentioned drawbacks,
A method of detecting the rotating magnetic field of a motor, determining the difference between the detected rotating magnetic field and a reference rotating magnetic field, and controlling an inverter so that the detected rotating magnetic field matches the reference rotating magnetic field is disclosed in, for example, Japanese Patent Laid-Open No. 59-25592. has been disclosed.

[Problem that the invention seeks to solve]

According to the above method, it is possible to generate a magnetic field in the motor that approximates the reference rotating magnetic field, but since the rotating magnetic field of the motor must be detected,
The configuration becomes complicated. SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an inverter control method that can easily obtain a desired rotating magnetic field.

[Means for solving problems]

To achieve the above object, the present invention provides a plurality of polyphase inverters so that a desired rotating magnetic field can be obtained with a polyphase AC load connected to the polyphase inverter, and the output voltage of the polyphase inverter has a desired value. A control method for an inverter that controls a control switch of a multiphase inverter includes the step of generating data indicating a reference rotating magnetic field vector φ _s corresponding to the desired rotating magnetic field for each clock pulse; The step of obtaining data indicating a virtual rotating magnetic field vector φ having the same value as the data of φ _s , and the step of obtaining data indicating a virtual rotating magnetic field vector φ having the same value as the data of φ A plurality of unit vectors V corresponding to voltage vectors V1 to V8
1' to V8' and adding this to the virtual rotating magnetic field vector φ based on the previous clock pulse, the reference rotating magnetic field vector φ _s
Obtain data indicating one unit vector selected to obtain a virtual rotating magnetic field vector φ close to , and add the data of this unit vector to the data indicating the virtual rotating magnetic field vector φ based on the previous clock pulse. The new virtual rotating magnetic field vector φ
the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ based on the data indicating the reference rotating magnetic field vector φ _s and the data indicating the virtual rotating magnetic field vector φ for each clock pulse; The error vector (φ _s −φ=
φ _e ), and for each clock pulse, the error vector φ _e changes in the directionality of the plurality of voltage vectors V1 to
A switch control signal A that can determine which voltage vector is the same as or similar to V8 and obtain the voltage vector that is the same as or similar to the error vector φ _e ;
A step of determining B and C based on the data of the error vector φ _e and supplying this switch control signal to each control switch of the inverter, and data indicating the unit level when changing the output voltage of the multiphase inverter. or a step of changing data indicating the reference rotating magnetic field vector φ _s .

[Effect]

When the reference rotating magnetic field (magnetic flux) vector φ _s and the virtual rotating magnetic field (magnetic flux) vector φ are given, the error vector φ _e between the two can be determined. Once the error vector φ _e is determined, an inverter control signal that reduces the error vector φ _e is determined.
The virtual rotating magnetic field vector φ is the voltage vector V ₁ ~
This vector is generated based on the unit vectors V' ₁ to V' ₈ corresponding to V ₈ and is approximate to the reference rotating magnetic field vector φ _s , so the difference between this virtual rotating magnetic field vector φ and the reference rotating magnetic field vector φ _s is If the error vector φ _e is controlled to be small, a desired rotating magnetic field can be naturally obtained in the inverter load. Moreover, since the control is not performed for each phase but for all phases at once, a desired inverter output can be obtained satisfactorily.

〔Example〕

Next, an inverter device and a control method thereof according to an embodiment of the present invention will be described with reference to the drawings.

(Configuration of Inverter Device) FIG. 1 shows a three-phase inverter device. In this Figure 1, 1 is a DC power supply, A ₁ ,
A ₂ , B ₁ , B ₂ , C ₁ , and C ₂ are control switches composed of transistors, and are bridge-connected. 2
is a three-phase AC motor as a load, and is connected to the A, B, and C phase output lines of the inverter.
3 is a microcomputer (hereinafter referred to as microcomputer), which functions as a control signal generation circuit for the inverter. This microcomputer 3 is provided with a read-only memory (ROM), in which data V ₁ to V ₈ for controlling the control switches A ₁ to C ₂ are written in advance. Memory M ₁ , memory M ₂ in which sin θ data is written in advance, memory M ₃ in which cos θ data is written in advance, reference rotating magnetic field vector φ _s
Error vector φ _s − between and virtual rotating magnetic field vector φ
a memory _M4 in which data f (φ _eQ ) necessary for determining the angular position of φ=φ _e is written in advance;
It has a memory _M5 in which arithmetic expressions used to determine unit vectors V' ₁ to V' ₈ are written in advance.

In addition to the above ROM, the microcomputer 3 has a CPU and
Of course, it also includes RAM (not shown) and the like. A voltage command signal and a frequency command signal are input to the microcomputer 3, and a control signal for obtaining an inverter output of the commanded voltage and frequency is generated. The microcomputer 3 outputs first, second, and third control signals A, B, and C as control signals corresponding to the three phases A, B, and C.
is generated and supplied to control switches A ₁ , B ₁ , and C ₁ . Also, by NOT circuits 4, 5, 6, A, B,
An inverted signal of C is formed and the control switches A ₂ , B ₂ ,
Supplied to C ₂ . Lower control switch A ₂ , B ₂ ,
Since C ₂ operates in the opposite manner to the upper control switches A ₁ , B ₁ , C ₁ , if the operations of the control switches A ₁ , B ₁ , C ₁ are specified, the operation of the entire inverter can be specified.
Therefore, in the following, the first, second, and third
The control state of the inverter is specified by the signals A, B, and C. Note that when the control signals A, B, and C are at high level, that is, logic 1, the control switches A ₁ , B ₁ , and C ₁
is controlled to be on, and control switches A ₁ , B ₁ , and C ₁ are controlled to be off when at a low level, i.e., logic 0.

(Explanation of Principle) FIG. 2 shows the basic idea of this system using D-Q orthogonal coordinates. In this figure, a shows the locus of the end point of the reference rotating magnetic field vector φ _s corresponding to the ideal rotating magnetic field required by the motor 2, and φ indicates the virtual rotating magnetic field vector introduced according to the present invention. The virtual rotating magnetic field vector φ is changed following the change in the angular position of the reference rotating magnetic field vector φ _s . That is, based on the error vector φ _e between the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ, a unit vector V to be added is determined, and this unit vector V is synthesized with the virtual rotating magnetic field vector φ, and is expressed as a dotted line. A new virtual rotating magnetic field vector φ is determined. Virtual rotating magnetic field vector φ determined sequentially
is determined along the circular locus a. Further, as the unit vector V, a vector specified in advance is used. That is, this unit vector V is a vector that can be generated by the on/off operation of the control switches _A1 to _C2 of the inverter.

Reference rotating magnetic field vector φ _s to obtain circular locus a
The rectangular coordinate data indicating
It is sequentially generated by sin θ of M ₂ and cos θ of memory M ₃ . Determination of the virtual rotating magnetic field vector φ is carried out using the memory M ₄ where this is located in the rectangular coordinates.
This is done by selecting a unit vector V that fits _this position, and adding it to the current virtual rotating magnetic field vector φ.
Once the coordinate data of the reference rotating magnetic field vector φ _s and the coordinate data of the virtual rotating magnetic field vector φ are obtained,
Data of the error vector φ _e is obtained through arithmetic processing based on this data, and the next virtual rotating magnetic field vector φ and control signals A, B, and C are determined based on this data.

(Explanation of voltage vector) Figure 3 shows the inverter control switches A ₁ , B ₁ ,
This shows the relationship between the state of C ₁ and the voltage vector. Three-phase inverter control switch A ₁ , B ₁ , C ₁
Possible states are (100) (110) (010) (011)
There are eight types: (001) (101) (111) (000), and the voltage vectors in each state are six space vectors V ₁ to V ₆ at 60 degree intervals and two zero vectors as shown in Figure 3. It can be expressed as V ₇ and V ₈ . By combining the above eight types of switching modes (eight voltage vectors), it is possible to generate any rotating magnetic field in the motor 2. Merely arranging six switching modes as shown in FIG. 3 within one cycle (2π) causes a large number of harmonic components, which is not practical. Therefore, in the present invention, a virtual rotating magnetic field vector φ is obtained by using unit vectors corresponding to the voltage vectors V ₁ to V ₈ , and this virtual rotating magnetic field vector φ and the reference rotating magnetic field vector φ _s are set to 1 of the rotating magnetic field. A control signal is determined by comparing each clock pulse (hereinafter simply referred to as a clock) that occurs many times during a cycle.

(Principle of virtual rotating magnetic field vector) The virtual rotating magnetic field vector φ is determined by determining the circular locus a of the reference rotating magnetic field vector φ _s as shown in FIG.
Divide the period (2π) into 6, set sections (1), (2), (3), (4), (5) and (6), and perform the process for each section. Now, set the center of section (1) to θ=0°
Then, section (1) corresponds to the range from -30° to +30°. The voltage vector V ₂ in this Fig. 4,
V ₃ is the same as voltage vectors V ₂ and V ₃ in FIG. When determining the virtual rotating magnetic vector in section (1), the unit vector parallel to the voltage vector V ₂ is
V′ ₂ , and a unit vector parallel to the voltage vector V ₃
Add V′ ₃ to the virtual rotating magnetic field vector φ. Since the virtual rotating magnetic field vector φ is determined by calculation, it can be generated to match the reference rotating magnetic field. However, since this virtual rotating magnetic field vector φ is used to determine the control signal of the control switch of the inverter, the voltage vector V _{1 to V} in FIG. Decide using _V8 . For this reason, φ
and φ _s do not match completely.

In section (1) of Fig. 4, unit vectors V' ₂ and V' 3 corresponding to the two voltage vectors V ₂ and V ₃ that can best follow the virtual rotating magnetic field vector to the circular locus _a are used. has been done. In addition, in order to obtain the virtual rotating magnetic field vector, in addition to the two vectors V ₂ and V ₃ , the zero vector
V ₇ or V ₈ is used to improve the followability of the virtual rotating magnetic field vector to the circular locus a. Adding the zero vector to the virtual rotating magnetic field vector means stopping the virtual rotating magnetic field vector at that angular position. The zero vector used is V ₇
Although there is no problem in principle with either the V8 or _V8 , in this embodiment, they are used alternately at 60° intervals as shown in FIG. For example, −60° to 0°, 60° to 120°
V ₇ is used at 0° to 60° and V ₈ is used at 0° to 60°.
The reason for specifying the use of _V7 and _V8 in this way is to reduce the number of switching operations in the inverter. For example, in the interval 0° to 60°, the voltage vector
The zero vector of V ₈ (000) is selected because the unit vector V′ ₃ parallel to V ₃ (010) is generated many times. As a result, when the V ₃ (010) switching mode and the V ₈ (000) switching mode are adjacent to each other, switches B ₁ and B ₂ are turned on and off.
All you have to do is turn it off, which reduces the number of times you have to switch. On the other hand, for example, in the −60° to 0° interval, there are many switching modes of V ₂ (110), so the zero vector V ₇ (111) is selected, and similarly
In the 60° to 120° section, the switching mode of V ₄ (011) increases, so the zero vector V ₇ (111) is selected.

(Description of Error Vector) FIG. 5 shows an error vector (φ _s −φ=φ _e ) between the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ from which the circular locus a is obtained. In microcomputer 3,
When the coordinate data (φ _sQ , φ _sD ) of the end point of the reference rotating magnetic field vector φ _s and the coordinate data (φ _Q , φ _D ) of the end point of the virtual rotating magnetic field vector φ are given, the coordinates indicating the error vector φ _e are given. Data (φ _eD , φ _eQ ) is obtained. Note that the vertical axis component of the error vector φ _eD is φ _sD
−φ _D = φ _eQ , and the horizontal axis component φ _eD is φ _sQ −φ _Q
It is obtained by the formula =φ _eQ . Once the data of the error vector φ _e shown in FIG. 5 is obtained, the control switch of the inverter to be controlled can be determined based on this data. As the control switch to turn on the inverter, one that can generate a vector that approximates the error vector φ _e is selected.

FIG. 6 explains the determination of a unit vector approximated to the error vector φ _e of FIG. 5 and the determination of a switch control signal. The error vector φ _e in FIG. 6 is obtained by moving the error vector φ _e in FIG. 5 into the interval (2). V ₂ in section (2), section (3) in Figure 6
V ₃ is the same as V ₂ and V ₃ in FIGS. 3 and 4. In FIG. 4, as is clear from the fact that the virtual rotating magnetic field vector φ in section (1) is formed using V ₂ and V ₃ , most of the error vector φ _e obtained in section (1) is in section ( 2) It can be moved in parallel within (3). If the direction of the error vector φ _e is included in the interval (2) shown in FIG.
The control switches A ₁ to C ₂ are controlled so that a voltage vector V ₂ intermediate between the two is obtained. That is, microcomputer 3
Generates (110) as control signal ABC from , turns on control switches A ₁ and B ₁ and turns off C ₁ ,
Control switches A ₂ , B ₂ , C ₂ are operated in the opposite direction to A ₁ , B ₁ , C ₁ .

The error vector φ _e at a certain clock is an interval
The microcomputer 3 determines whether the section belongs to (2) or not.
(2), then V ₂ of memory M ₁
The address where (110) is written is specified,
V ₂ (110) is read out, corresponding to ABC (110)
is output.

If the direction of the error vector φ _e obtained by calculating φ _s −φ = φ _e in interval (1) falls in interval (3), then
Output V ₃ (010) from microcomputer 3. If the direction of error vector φ _e in interval (1) does not fall within either interval (2) or (3), zero vector V ₇ or V ₈
is output from microcomputer 3.

In interval (1), data V ₂ (110) from which a voltage vector matching or approximating the error vector φ _e obtained by calculating φ _s −φ=φ _e as described above can be obtained; V ₃ (010) and zero data
By outputting V ₈ (000) and V ₇ (111) and controlling the inverter control switch accordingly,
It is possible to generate a rotating magnetic field in the motor 2 that approximates the reference rotating magnetic field vector φ _s .

We have just described section (1), but if the reference rotating magnetic field vector φ _s is located in section (2), it is difficult to determine whether the error vector φ _e belongs to section (3), (4), or (3). )
(4), and if it belongs to section (3), data V ₃ (010) corresponding to voltage vector V ₃ is determined.
is output from the microcomputer 3, and when it belongs to section (4), data V ₄ (011) corresponding to the voltage vector V ₄ is output from the microcomputer 3. In addition, when it does not belong to sections (3) and (4), the data corresponding to the zero vector V ₇ (111)
Or output V ₈ (000).

When calculating φ _s −φ=φ _e in interval (3), voltage vectors V ₄ and V ₅ are used instead of voltage vectors V ₂ and V ₃ in interval (1).

When calculating φ _s −φ=φ _e in interval (4), the interval
In place of the voltage vectors V ₂ and V ₃ in case (1), voltage vectors V ₅ and V ₆ are used.

When calculating φ _s −φ=φ _e in interval (5), the interval
The voltage vectors V ₆ and V ₁ are used instead of the voltage vectors V ₂ and V ₃ in case (1).

When calculating φ _s −φ=φ _e in interval (6), voltage vectors V ₁ and V ₂ are used instead of voltage vectors V ₂ and V ₃ in interval (1).

In addition, when calculating φ _s −φ = φ _e in each interval (1) to (6), φ _e = φ e at the start of each interval.
Make it 0. That is, φ _s =φ is set.

(Specific Description) Next, the control method according to the present invention will be specifically described.

First, the data of the reference rotating magnetic field vector φ _s is obtained according to the clock (angle θ data) according to the frequency ( ) command signal given to the microcomputer 3 . That is, sin θ written in memories M ₂ and M ₃ ,
Give angle data θ corresponding to the clock to cosθ,
The horizontal axis and vertical axis data φ _sQ and φ _sD of the Q-D coordinate on the circular locus a in FIG. 2 are obtained by the following equation.

φ _sQ = Acos θ φ _sD = Asin θ However, A is the radius of the circular locus a. As a result, each coordinate data on the circular locus a in Figure 2 is obtained sequentially, and now, if we show the interval (1) from θ = -30° to 0° in an analog way, it is the black dot in Figure 8. This will be the trajectory shown.

The horizontal axis and vertical axis data φ _Q and φ _D of the Q-D coordinate indicating the virtual rotating magnetic field vector φ are determined as follows. First, at the start of each section, the data of the reference rotating magnetic field vector is read and used as an initial value. Now, to explain using interval (1) as an example, the initial value of the rotating magnetic field vector is the voltage vector V ₂ of interval (2).
Add the unit vector V′ ₂ corresponding to . next,
Add unit vector V′ ₃ . This unit vector
When V' ₂ and V' ₃ are shown in analog form, they are shown as arrows connecting the x marks in FIG. The x mark in FIG. 8 indicates analog data of the virtual rotating magnetic field vector φ.

The unit vectors V' ₁ to V' ₈ necessary to obtain the virtual rotating magnetic field vector φ are determined as follows. The unit vectors V' ₁ to V' ₈ used correspond to the three-phase voltage vectors V ₁ to _{V 8} shown in FIG. Control data A, B, and C of the control switches _A1 to _C1 for the voltage vectors _V1 to _V8 of the three phases A, B, and C are written in the memory _M1 of the microcomputer 3. That is,
Corresponding to V ₁ (100), Corresponding to V ₂ (110), V ₃
(010), V ₄ (011), V ₅ (001), V ₆ (101), V ₇ (111), V ₈ (000) is written correspondingly. Note that the data contents (100) etc. correspond to the output ABC of the microcomputer 3. Therefore, V ₁ to _{V 8}
can be represented by V, A, B, and C. It would be advantageous if this data could be used to determine the unit vectors V' ₁ -V' ₆ . Data V ₁ (100) ~
Q of unit vector V′ ₁ ~V′ ₆ based on V ₆ (101)
-D coordinate data V _Q and V _D can be obtained using the following formula.

V _Q =√3/2(B-C)...(1) V _D =A-1/2B-1/2C...(2) However, A, B, and C are data V ₁ (100) ~ _V6
The contents of (101) correspond to (100) to (101), that is, A, B, and C, and the contents of data V ₁ (100) to _{V 6} (101) are 1
In the case of , it is substituted as it is into the equations of V _Q and V _D , and if the content is zero, it is substituted as -1 into the equations of V ₁ to V ₆ . As a result, each coordinate data of a unit vector in the case of a circular locus is as follows.

V′ ₁ (0,2) V′ ₂ (√3,1) V′ ₃ (√3,−1) V′ ₄ (0,−2) V′ ₅ (−√3,−1) V′ ₆ (−√3,1) FIG. 11 shows the unit vectors V ₁ to V ₆ and coordinate data. The above equations (1) and (2) are the memory of microcomputer 3.
Since it is written in M ₅ , the data read from memory M ₁ V ₁ (100) to _{V 6} (101) and the above equations (1) (2)
The unit vectors V' ₁ to V' ₆ can be easily determined by the combination with Note that this approximate value is used for √3/2 in equation (1).

Next, the procedure of arithmetic processing will be explained with reference to FIGS. 7 and 8. At time _t0 in FIG. 8, the processing of blocks 11 and 12 in FIG. 7 is performed. That is, the virtual rotating magnetic field vector φ is made to match the reference rotating magnetic field vector φ _s .

Next, in block 13, the section shown in FIG.
Determine the end of (1) to (6). Since the 60 degree section has not ended at t ₀ , NO (hereinafter simply referred to as N)
The output is obtained and the address of φ _s is incremented in the next block 14.

Next, in block 15, the calculation φ _s −φ=φ _e is performed.

That is, an error vector φ _e between the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ is determined. In addition,
The data of this error vector φ _e is obtained from the coordinate data of the Q axis and the D axis (φ _eQ =φ _sQ −φ _Q )(φ _eD =φ _sD −φ _D ).

Next, in block 16, it is determined whether φ _eQ >0. t ₀
Since φ _eQ = 0 at this point, the output is N,
A zero vector is selected. Zero vector data is
Since there are two types, V ₇ (111) and V ₈ (000), the angular position of the reference rotating magnetic field vector φ _s is determined in block 19, and V ₇ or V ₈ is selected based on this. This selection of V ₇ and V ₈ is performed in the manner already explained with reference to FIG. Figure 8 is an example of section (1), so V ₇ is selected at t ₀ , V ₇ of memory M ₁ of microcomputer 3 is addressed, and (111) is output to microcomputer outputs A, B, and C. Output.
As a result, control switches A ₁ , B ₁ , and C ₁ are turned on.
A ₂ , B ₂ , C ₂ are turned off, and each line voltage A-B,
B-C and C-A become 0V.

Next, when the timer in block 28 generates a clock pulse at time _t1 in FIG. 8 according to the timing set by the frequency command signal, block 13 again determines that the 60° section has ended, and the output is N. Then, in block 15, the calculation φ _s −φ=φ _e is performed again. At this time t ₁ , the error vector φ s1 −φ ₀ =φ _e between the reference rotating magnetic field vector φ _s1 at the time t ₁ and the virtual rotating magnetic field vector _{φ 0} set at the time t ₀
is required.

Next, in block 16, it is determined whether φ _eQ >0. Since φ _s1 >φ ₀ at time t ₁ , an output of YES (hereinafter simply referred to as Y) is obtained. Therefore, in block 17, it is determined that φ _eD >0. φ _eD is the vertical axis component of φ _e . Since the time t ₁ is a time a little further clockwise than −30° in FIG. 4, the direction of the error vector φ _e is almost the same as the voltage vector V ₂ in FIG. 4. Therefore, φ _eD of block 17 >
The output of 0 becomes Y. When the output Y is obtained from block 17, memory M ₄ is stored according to block 20.
The function f(φ _eQ ) is read. function f
(φ _eQ ) is the vertical axis (D axis) corresponding to the horizontal axis (Q axis) component of the vector corresponding to the boundary line of each section (1) to (6)
This is to find the ingredients. Therefore, when the error vector φ _e shown in FIG. 6 is obtained, f(φ _eQ ) in the horizontal axis component φ _sQ of this error vector φ _e is immediately obtained. The value of f(φ _eQ ) is the vertical axis component at the point where the line vertically rising from φ _eQ intersects the straight line D=f(Q) in FIG. As is clear from Figure 6,
At time t ₁ , f(φ _eQ ) > φ _eD , so block 2
An output of 1 to Y is generated, V ₂ of memory M ₁ is addressed, and V ₂ (110) is read.
In block 22, in order to obtain the next virtual rotating magnetic field vector, the following calculations are performed: φ _Qo = φ _Q(o-1) +√3 φ _Do = φ _D(o-1) +1.

φ _Qo is the horizontal axis component of the new virtual rotating magnetic field vector φ, φ _Q(o-1) is the horizontal axis component of the virtual rotating magnetic field vector φ currently written in RAM, φ _Do is the vertical axis component of the new virtual rotating magnetic field vector φ The axis component φ _D(o-1) is the vertical axis component of the virtual rotating magnetic field vector φ currently written in the RAM.

At time t ₁ in FIG. 8, φ _Qo = φ _0Q +√3=φ _1Q φ _Do = φ _0D +1=φ _1D are obtained. Note that φ _0Q and φ _0D are the Q and D components of φ ₀ , and φ _1Q and φ _1D are the Q and D components of φ ₀ . In short, a new virtual rotating magnetic field vector φ ₁ is obtained by adding _the unit vector V to the virtual rotating magnetic field vector φ 0 at time t ₀ in FIG. 8. This vector φ ₁ is then written to RAM or a register for use as φ=φ ₁ in the operation in block 15.

In block 23, it is determined whether the reference rotating magnetic field φ _s belongs to any of the sections (1) to (6), and it is determined whether it is necessary to replace the V ₂ vector with another vector. . Since the time point t ₁ is in interval (1), the address indicating V ₂ of the memory M ₁ is specified, and V ₂ (110) is output. This results in V ₂ (110)
In response to this, control switches A ₁ and B ₁ are turned on, C ₁ is turned off, A ₂ and B ₂ are turned off, and C ₂ is turned on.
The output shown in FIG. 8 is obtained between lines B-C and C-A.

When the processing at time _t1 is completed, a clock at time _t2 is generated from the timer of block 28, and the process moves on to the next operation. That is, even at this time _t2 , block 1
In step 5, the calculation φ _s −φ=φ _e is performed. At this time t ₂ , the new virtual rotating magnetic field vector φ ₁ has already been determined, so this is read from the RAM and φ _s −
Perform the calculation φ=φ _s2 −φ ₁ =φ _e . In Figure 4, for the sake of explanation, the unit vectors V' ₂ and V' ₃ are drawn extremely large, so the unit vector V' ₃ in a different direction is drawn next to the first unit vector V' ₂ . However, in reality, processing is performed at fine angular intervals, so the error vector at time t ₂ is φ _s2 −φ ₁ =
φ _e is in almost the same direction as the vector V ₂ , an output of Y is obtained from block 16, and block 17
An output of Y is also obtained from block 21, and an output of Y is also obtained from block 21, and vector _V2 is selected.
As a result, in block 22 a new virtual rotating magnetic field vector φ ₂ is determined. That is, the vector φ ₁ +V′ ₂ =φ ₂ is determined. On the other hand, in response to the selection of vector _V2, (110) is obtained as outputs A, B, and C from microcomputer 3, and at the next time _t3 , block 1
When calculating φ _s −φ=φ _s3 −φ ₂ =φ _e in block 16, the virtual rotating magnetic field vector φ ₂ is ahead of the reference rotating magnetic field vector φ _s3 , so an output of N is generated in block 16. A zero vector is selected. Then, in block 19, as the zero vector
A decision is made whether to output V ₇ (111) or V ₈ (000), and at time t ₃ , vector V ₇ (111) is selected, and the microcomputer outputs A, B, and C become (111). Become. As a result, control switches A ₁ , B ₁ , C ₁ are turned on, and control switches A ₂ , B ₂ , C ₂ are turned off. Therefore, the new virtual rotating magnetic field vector is not changed at time _t3 .

At time t ₄ , according to block 15, φ _s4 −φ ₂
=φ _e is calculated. In this case, the output of block 16 becomes Y, and the output of blocks 17 and 21 becomes Y.
The output of will also be Y, the V ₂ vector will be selected, and φ ₂
A new virtual rotating magnetic field vector φ ₃ is determined by +V′ ₂ =φ ₃ . Furthermore, based on the selection of vector _V2 , the microcomputer outputs A, B, and C become (110).

At time t ₅ , according to block 15, φ _s5 −φ ₃ =
The calculation of φ _e is performed, and in this case, since φ ₃ is ahead of φ _s5 , the output of block 16 becomes N,
Zero vector V ₇ is selected again, and the microcomputer output A,
B and C become (111). A new virtual rotating magnetic field vector is not determined at this time t ₄ .

At time t ₆ , according to block 15, φ _s6 −φ ₃ =
When calculating the error vector φ _e by calculating φ _e , we get
A Y output indicating φ _eQ >0 is obtained from block 16,
The next block 17 provides N outputs. That is, at the level shown analogously in FIG.
Since φ _s6 is below the level of φ ₃ , φ ₃ to φ _s6
The error vector directed toward becomes downward, the vertical axis component φ _eD of φ _e becomes negative, and the output of block 17 becomes N. This means that the error vector φ _e is in the interval shown in Figure 6.
(3) or indicates that it belongs to a further advanced section.

In block 18, it is determined whether the error vector is included in the interval (3) using f(φ _eQ ) in the memory _M4 .
In preparation for making a decision using , polarity conversion _{φ eD =} −φ _eD of the vertical axis component φ eD of φ _e is performed.

Next, in blocks 24 and 25, it is determined in the same manner as in blocks 20 and 21 whether or not the error vector φ _e is within the interval (3). As a result, when a Y output is obtained, vector V ₃ is selected.

In block 26, calculations are performed to obtain a new virtual rotating magnetic field vector φ ₄ . That is, φ ₃ +V′ ₃ =φ ₄ is determined. This coordinate component is determined by the equation in block 26 as follows.

φ _eo = φ _Q(o-1) +√3 = φ _3Q +√3=φ _4Q φ _Qo = φ _D(o-1) −1 = φ _3D −1=φ _4D In block 27, at time t ₆ In this case, since it is the interval (1), the microcomputer output A, corresponding to the selected vector _V3 is determined.
(010) is sent out as B and C, control switch A ₁ is turned off, control switch B ₁ is turned on, and control switch C ₁ is turned off.

At time t ₇ , φ _s − φ = φ _s7 − φ ₄ in block 15.
=φ _e operation is performed. As is clear from the positional relationship between φ _s7 and φ ₄ in FIG. 8, the direction of the error vector φ _e is downward. As a result, the output of block 16 becomes Y and the output of block 17 becomes N. Furthermore, since this error vector φ _e has a direction that belongs to a position further advanced than the interval (3), the output of block 25 becomes N, and the zero vector V ₇ (111) is selected.

The above-mentioned operation is repeated every clock, and when section (1) ends, an output of Y is obtained from block 13, .phi.=. _phi.s is set again, and the operation of section (2) begins. In interval (2), voltage vectors V ₃ and V ₄ are used instead of voltage vectors V ₂ and V ₃ in interval (1).

Further, by operating in the order of sections (3), (4), (5), and (6), a rotating magnetic field for one period can be obtained.

As is clear from the above explanation, in the method of this embodiment, the virtual rotating magnetic field vector φ is set and it is controlled to follow the reference rotation vector φ _s , so that the reference rotation vector is applied to the motor 2 of the inverter output stage. A rotating magnetic field approximately corresponding to φ _s can be obtained.

In this method, when you want to change the output voltage value of the inverter, use equations (1) and (2) to calculate the unit vectors V' ₁ to V' ₆ using the voltage command signal to the microcomputer 3.
Multiply by a constant. When lowering the voltage, for example, V _Q =(√3/2(B-C))×2 V _D =(A-1/2B-1/2C)×2 is determined. As a result, V′ ₁ to V′ ₆ in FIG. 11 are
V′ ₁ (0,4), V′ ₂ (2√3,2), V′ ₃ (2√3,
−
1), V' ₄ (0, -4), V' ₅ (-2√3, -2), V
' ₆
(−2√3,2), respectively.

FIG. 9 shows the state in the θ=−30° to 0° interval in this case in the same manner as FIG. 8.

Another method for adjusting the inverter output voltage is to change the magnitude of the reference rotating magnetic field vector φ _s . In this method, when determining the reference rotating magnetic field vector φ _s based on the voltage command signal to the microcomputer 3, a constant is multiplied. That is, the radius A in FIG. 2 is changed. In order to lower the inverter output voltage, if the magnitude of the reference rotation vector φ _s in FIG. 8 is reduced to 1/2, the operation will be as shown in FIG. 10.
(Explanation of FIGS. 12 to 15) In FIGS. 12 to 15, in order to simplify calculations, the end point locus of the reference rotating magnetic field vector φ _s is an elliptical locus, and the unit vector V' ₁ The operation when ˜V′ ₆ is determined to match this will be explained. In the case of the circular locus shown in FIG. 11, the unit vectors V' ₂ , V' ₃ , V' ₅ , and V' ₆ contain √3. Therefore, unit vectors V' ₁ to V' ₆ that follow an elliptical locus that does not include this √3 are set as shown in FIG. 12. To obtain this unit vector V′ ₁ ~V′ ₆ ,
Write 2/√3sinθ in memory _M2 of microcomputer 3 in advance. As a result, the horizontal axis component V _Q and vertical axis component V _D of the unit vectors V' ₁ to V' ₆ are determined based on V _Q = B - C V _D = A-1/2B-1/2C. I can do it. In this case as well, the values of data A, B, and C in memory _M1 are 1.
In the case of , it is substituted as is, and in the case of 0, -1 is substituted. As a result, the unit vectors V' ₁ (0, 2), V' ₂ (2, 1), V' ₃ (2, -1) shown in Fig. 12 are obtained.
,
V' ₄ (0, -2), V' ₅ (-2, -1), V' ₆ (-2,
1)
is obtained.

It is exactly the same as the case of a circular locus, except that the unit vector V ₁ for determining the virtual rotating magnetic field vector φ is set as described above, and the locus of the reference rotating magnetic field vector φ _s is an ellipse. control is carried out.

Figure 13 shows the state of each part in the case of an elliptical locus.
Shown in correspondence with the figure.

Figure 14 shows that in order to change the inverter output voltage, the unit vectors are V' ₁ (0, 4), V' ₂ (4, 2),
V' ₃ (4, -2), V' ₄ (0, -4), V' ₅ (-4, -
2),
The case where V' ₆ (-4, 2) is set is shown in correspondence with FIG. 9.

Figure 15 shows the reference rotating magnetic field vector φ _s of an elliptical locus.
The case where the value of is reduced to 1/2 is shown corresponding to FIG.

The inverter output frequency is changed by changing the clock frequency using a frequency command signal.

[Modified example]

The present invention is not limited to the above-described embodiments, and, for example, the following modifications are possible.

(b) In each section (1) to (6) in Figure 4, the first 30° region of the virtual rotating magnetic field vector and the second 30° region of the section are symmetrical with respect to the central angle of the section. , after the first half of the calculation process of 30 degrees is completed, the process may be looped back. Figure 16 shows the principle of this control.
Unit vectors V′ ₂ , V′ ₃ ,
If V′ ₂ , V′ ₃ , and V′ ₂ are used in this order, then V′ ₅ , V′ ₆ , V′ ₅ , V′ ₆ , and V′ ₅ are used when turning back. Note that when turning back, a vector in the opposite direction is used.

(b) If the memory capacity of microcomputer 3 is large,
The virtual rotating magnetic field vector may be determined by writing various unit vectors V' ₁ to V' ₆ in advance and reading out the unit vectors corresponding to the output voltages.

(c) Instead of calculating the coordinate data of the reference rotating magnetic field vector φ _s , the coordinate data may be sequentially written into a memory and read out and used sequentially according to the clock.

(d) Instead of sequentially calculating the virtual rotating magnetic field vector φ, the coordinate data of the virtual rotating magnetic field vector φ used at each angular position (clock) is sequentially written in memory, and this is read out every clock. Alternatively, the error vector φ _e may be determined.

〔Effect of the invention〕

As is clear from the above, according to the present invention, a reference rotating magnetic field vector is generated, a virtual rotating magnetic field vector that is the same as the reference rotating magnetic field vector is determined at the time of starting the inverter, and thereafter, for each clock pulse, the vector of the previous rotating magnetic field is determined. A new virtual rotating magnetic field vector is obtained by adding one selected from a plurality of predetermined unit vectors to the virtual rotating magnetic field vector based on the clock pulse, and this new virtual rotating magnetic field vector is combined with the reference. Since the error vector with respect to the rotating magnetic field vector is determined and the switch control signal is determined based on this error vector, the desired rotating magnetic field can be easily obtained. Further, since the output voltage of the inverter is changed by changing the unit vector or the reference rotating magnetic field vector, the voltage can be adjusted while maintaining a desired rotating magnetic field.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an inverter device according to an embodiment of the present invention, Fig. 2 is a vector diagram for explaining the principle of the invention, and Fig. 3 is a diagram showing the switch state of a three-phase inverter and the rotating magnetic field vector. FIG. 4 is a vector diagram showing the change in the virtual rotating magnetic field vector in section (1) and the selection range of the zero vector, FIG. 5 is a vector diagram showing the error vector, and FIG. 6 is a vector diagram showing the relationship. A vector diagram showing the method of determining the vector to be selected based on the error vector, Figure 7 is a diagram showing the flow of the control procedure by the device in Figure 1, and Figure 8 is a diagram showing the state of each part in Figure 1 in principle. Figure 9 is a waveform diagram corresponding to Figure 8 when the magnitude of the unit vector is changed, Figure 10 is a waveform diagram corresponding to Figure 8 when the magnitude of the reference rotating magnetic field vector is changed. 11 is a vector diagram showing unit vectors in the case of a circular locus, FIG. 12 is a vector diagram showing unit vectors in the case of an elliptical locus, FIGS. 13 and 14.
and FIG. 15 are waveform diagrams showing the case of an elliptical locus corresponding to FIGS. 8, 9, and 10.
FIG. 6 is a vector diagram showing a method for obtaining a virtual rotating magnetic field vector in a modified example. 1...Power supply, 2...Motor, 3...Microcomputer,
φ...Virtual rotating magnetic field vector, _φs ...Reference rotating magnetic field vector, _V'1 to _V'6 ...Unit vector.

Claims (1)

[Claims] 1. A plurality of control switches of the polyphase inverter are controlled so that a desired rotating magnetic field is obtained in the polyphase AC load connected to the polyphase inverter, and the output voltage of the polyphase inverter becomes a desired value. A method of controlling an inverter includes the steps of: generating data indicating a reference rotating magnetic field vector φ _s corresponding to the desired rotating magnetic field for each clock pulse; and generating data of the reference rotating magnetic field vector φ _s when starting the multiphase inverter. a step of obtaining data indicating a virtual rotating magnetic field vector φ having the same value as , and a plurality of voltage vectors V1 to V1, which are predetermined to correspond to the on/off states of the plurality of control switches of the multiphase inverter, for each clock pulse. V
8 and is added to the virtual rotating magnetic field vector φ based on the previous clock pulse, thereby creating a virtual rotating magnetic field close to the reference rotating magnetic field vector φ _s. 1 selected so that the vector φ is obtained
a step of obtaining data indicating a new virtual rotating magnetic field vector φ by adding the data of this unit vector to data indicating a virtual rotating magnetic field vector φ based on the previous clock pulse to obtain data indicating a new virtual rotating magnetic field vector φ; data indicating the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ
a step of obtaining data indicating an error vector (φ _s −φ=φ _e ) between the reference rotating magnetic field vector φ _s and the virtual rotating magnetic field vector φ based on data indicating the error vector φ for each clock pulse; A switch control signal A that can determine whether _e is the same as or approximate to any of the plurality of voltage vectors V1 to V8 in vector direction, and obtain the voltage vector that is the same as or approximate to the error vector φ _e . , B, and C based on the data of the error vector φ _e and supplying this switch control signal to each control switch of the inverter, and indicating the unit level when changing the output voltage of the multiphase inverter A method for controlling an inverter, comprising the step of changing data or data indicating the reference rotating magnetic field vector _φs .