JPH0634623B2 - Inverter control method - Google Patents

Description

The present invention relates to a control method for a three-phase inverter connected to a three-phase AC load.

[Conventional technology]

It is known to connect variable frequency and voltage inverters to AC motors. It is also known to perform PWM (pulse width modulation) control of this inverter. By the way, 3
When the phase inverter is driven by PWM control and an approximate sine wave is obtained, if each phase is modulated, it is likely to be affected by other phases, and an optimum output cannot be obtained. That is, since the line voltage of the inverter output is determined by the difference between the phase potentials, it is practically impossible to approximate each line voltage to a sine wave at the same time.

As a method for solving the above-mentioned drawbacks, the rotating magnetic field of the motor is detected, the difference between the detected rotating magnetic field and the reference rotating magnetic field is obtained, and the inverter is controlled so that the detected rotating magnetic field matches the reference rotating magnetic field. The method of doing is, for example, JP-A-59.
No. 25592.

According to the above method, it is possible to generate a magnetic field close to the reference rotating magnetic field in the motor, but since the rotating magnetic field of the motor has to be detected, the configuration becomes complicated. Therefore, the applicant of the present application has proposed in Japanese Patent Application No. 60-66259 a control method for an inverter that can easily obtain a desired rotating magnetic field. In addition, the applicant of the present application filed Japanese Patent Application No. 60-15
No. 3885, a method of obtaining a desired rotating magnetic field with high accuracy and easily was proposed.

[Problems to be solved by the invention]

By the way, when the switch of the inverter is on / off controlled, in order to prevent the pair of switches connected between the pair of DC power lines from being simultaneously turned on, a non-control period, that is, a unit vector generation period, There must be a rest period (no voltage period). For this reason, the output voltage of the inverter is reduced by the non-control period.

Then, the objective of this invention is providing the control method of the inverter which can obtain an output voltage higher than before.

[Means for solving problems]

In order to solve the above problems and achieve the above object, the present invention provides 6 to obtain an approximate rotating magnetic field vector φ that approximates to the reference rotating magnetic field vector φ _S in a three-phase AC load connected to a three-phase inverter. One unit vector and zero vector are determined, the angular position of the reference rotating magnetic field vector φ _S is divided into six equal parts, and two unit vectors and zero vector selected from the six unit vectors are used in each section. The switch of the three-phase inverter is turned on / off to obtain the approximate rotating magnetic field vector, and the inverter output voltage value is changed by changing the ratio of the unit vector generation time to the unit vector generation period. In an inverter control method for on / off control of an inverter, by increasing the output voltage of the inverter, the When the mutual intervals of the generation periods of the two unit vectors reach a predetermined minimum interval in a single or a plurality of locations, the first of the two unit vectors adjacent to each other in at least a part of the single or a plurality of locations The present invention relates to a control method for an inverter, which is characterized in that the generation period of the unit vector is extended to a later generation time of the unit vector.

[Work]

As described above, if the two adjacent unit vectors are connected continuously by the preceding unit vector, the output voltage of the inverter can be increased by this amount.

〔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.

(Structure of Inverter Device) FIG. 1 shows a three-phase inverter device. In FIG. 1, (1) is a DC power source, A ₁ , A ₂ , B ₁ , B ₂ ,
C ₁ and C ₂ are control switches composed of transistors,
Bridge connected. (2) is a three-phase AC motor as a load, which is connected to the A, B, and C phase output lines of the inverter. (3) is a microcomputer (hereinafter referred to as a microcomputer), which functions as a control signal generation circuit for the inverter. A read-only memory (ROM) is provided in the microcomputer (3), and the ROM includes the first to fourth memories M ₁ , M ₂ ,
It has M ₃ and M ₄ . The first memory M ₁ is written with a vector arrangement of unit sections (60 ° section) described later, and the second memory M ₂ is used for the first to sixth sections (0 ° to 360 ° section). The vector is written and the third memory M
₃ data for designating the ON time width of the vector is written to, the fourth memory M ₄ are written as shown in FIG. 8 is data that indicates where the buried between vectors other.

The microcomputer (3) has a CPU and RA in addition to the ROM.
Of course, M (not shown) and the like are also included. This microcomputer (3)
A voltage command signal and a frequency command signal are input to the control circuit, and a control signal for obtaining an inverter output of the commanded voltage and frequency is generated from the output terminals a _{1 to} c ₂ . As control signals from the microcomputer (3), the first, second and third control signals A, B and C corresponding to the three phases A, B and C are terminals a ₁ , b ₁ and
It originates from c ₁ and is fed to the control switches A ₁ , B ₁ and C ₁ . Further, the inverted signals of A, B and C are output to terminals a ₂ , b ₂ and c _2.
Supply to the control switches A ₂ , B ₂ and C ₂ . The lower control switches A ₂ , B ₂ , C ₂ are the upper control switches A ₁ ,
Since the operation is the reverse of B ₁ and C ₁ , the control switches A ₁ , B ₁ and
If the operation of C ₁ is specified, the operation of the entire inverter is specified. Therefore, in the following, the control state of the inverter is specified by the first, second and third signals A, B and C. When the control signals A, B, and C are at high level, that is, at logic 1, the control switches A ₁ , B ₁ , and C ₁ are turned on, and when they are at low level, that is, at logic 0, the control switch A ₁ ,
B ₁ and C ₁ are controlled to be off. Further, both the upper and lower switches are kept off during the non-control period.

(Explanation of Principle) FIG. 2 shows the relationship between the reference rotating magnetic field vector φ _S and the rotating magnetic field vector φ approximated to this by coordinates.
In this figure, a shows the locus of the end point (tip) of the reference rotating magnetic field vector φ _S corresponding to the ideal rotating magnetic field required by the motor (2), and φ is the approximation (virtual) introduced according to the present invention.
The rotating magnetic field vector is shown. The approximate rotating magnetic field vector φ is
It can be changed following the change in the angular position of the reference rotating magnetic field vector φ _S. That is, the unit vector V ′ to be added next is determined based on the error vector between the reference rotating magnetic field vector φ _S and the approximate rotating magnetic field vector φ, and this unit vector V ′ and the approximate rotating magnetic field vector φ are combined. , A new approximate rotating magnetic field vector φ shown by a dotted line is determined. The sequentially determined approximate rotating magnetic field vector φ is determined along the circular locus a. A previously specified vector is used as the unit vector V '. That is, the unit vector V 'is a vector that can occur in the on-off operation control switches A ₁ -C ₂ inverter.

The approximate rotating magnetic field vector φ is the inverter switch A ₁
Since those obtained by on-off control of -C _2, it is not possible to this match perfectly with the reference rotating magnetic field vector phi _S. In this embodiment, the 60 ° section (unit section) is divided into 12 at 5 ° intervals, and one significant unit vector is generated in the unit angle section (5 °).

(Explanation of Voltage Vector) FIG. 3 shows the relationship between the states of the control switches A ₁ , B ₁ , C ₁ of the inverter and the voltage vector. The possible states of the control switches (A ₁ , B ₁ , C ₁ ) of the three-phase inverter are (100) (110) (010) (011) (0
01) (101) (111) (000), the voltage vector in each state is, as shown in FIG. 3, six space vectors V _{1 to} V ₆ at 60 degree intervals and two zero vectors V. It can be represented by ₇ and V ₈ . By combining the above eight kinds of switching modes (eight voltage vectors), an arbitrary rotating magnetic field can be generated in the motor (2). If only six switching modes are arranged in one cycle (2π) as shown in FIG. 3, there are many harmonic components, which is not preferable in practice. Therefore, in the present invention, the inverter is controlled so as to obtain the approximate rotating magnetic field vector φ by using the unit vectors corresponding to the voltage vectors V _{1 to} V ₈ .

(Selection of Unit Vector) The determination of the approximate rotating magnetic field vector φ is performed as shown in FIG.
One cycle (2π) of the circular locus a of the reference rotating magnetic field vector φ _S
Is divided into six, first to sixth sections are set, and the processing is performed for each section.
In FIG. 4, an angle range of 60 ° to 120 ° determined based on the B-C phase voltage V _BC is defined as the first section,
° -180 ° is the second section, 180-240 ° is the third section, 240-300 ° is the fourth section, and 300-360
(0 °) is the fifth section, and 0-60 ° is the sixth section. The voltage vectors V ₂ and V ₃ in FIG. 4 are the same as the voltage vectors V ₂ and V ₃ in FIG. When determining the approximate rotation vector in the first section, the voltage vector V ₂
Unit vector V ′ ₂ (110) parallel to the voltage vector V ₃ or unit vector V ′ ₃ (010) parallel to the voltage vector V ₃ is added to the approximate rotating magnetic field vector φ. That is, in the first section of FIG. 4, the two voltage vectors V ₂ and V ₃ that allow the approximate rotating magnetic field vector φ to best follow the circular locus a.
The first and second unit vectors V ′ ₂ and V ′ _{3 corresponding} to In order to obtain the approximate rotating magnetic field vector, which will be described later, two unit vectors V ′ ₂ and V ′ ₃
Besides, using the zero vector V _'7 or V' ₈ of. Adding a zero vector to the approximate rotating field vector means
This means stopping the approximate rotating magnetic field vector at that angular position.

The approximate rotating magnetic field vector in the second section is formed by using the unit vectors V ′ ₃ and V ′ ₄ corresponding to the voltage vectors V ₃ (010) and V ₄ (011) shown in FIG. ₃ , and so on. In the third section, voltage vectors V ₄ (011) and V _4
The unit vectors V ′ ₄ and V ′ ₅ corresponding to ₅ (001) are used to form the voltage vector V ₅ (00
1) and V ₆ (101) corresponding to unit vectors V ′ ₅ ,
It formed using V _'6, the voltage vector V ₆ in fifth section
Unit vectors V ′ ₆ and V ′ ₁ corresponding to (101) and V ₁ (100), and unit vectors corresponding to voltage vectors V ₁ (100) and V ₂ (110) in the sixth section. V formed using _'1, V' _2.

(Determination of Unit Vector) FIG. 5 illustrates the determination of the unit vector in the first section. If the angular positions of the reference rotating magnetic field φ _S and the approximate rotating magnetic field φ are known, the unit vector to be used is determined by the principle explained in FIG. That is, in the case of the first section,
Unit vector V _'2, V' ₃ are used. In FIG. 5, assuming that the coordinates of the end point (tip) position A of the approximate rotating magnetic field vector φ are (x _A , y _A ), the point B obtained by adding one unit vector V ′ ₂ to this vector φ Coordinates (x _B , y
_B ), and the vector φ has the other unit vector V ′ ₃
The coordinate of the point C obtained by adding is (x _C , y _C ). Point B that may be the tip position of the new approximate rotating magnetic field vector
Alternatively, C is determined by calculation. After determining the point B and C coordinates, it determines whether to select one of the one unit vector V _'2 and other unit vector V' ₃ on the basis of this in operation.

Next, this determination method will be described in detail. The vector end point locus a of the reference rotating magnetic field changes for each angle section (clock). Therefore, the distance between the end point D (x _D , y _D ) of the reference rotating magnetic field vector φ _S and the point A, the distance between the point D and the point B, and the distance between the point D and the point C are obtained for each angle segment (clock). . In this example, the angular position where the A-D distance and the B-D distance are equal is ahead of the angular position where the A-D distance and the C-D distance are equal, so one unit 'selects only _2, the other unit vector V' vector V does not select _3.
Such a manner to determine one or generating the V _'2 and V' _3.

(Arrangement of Unit Vectors) FIG. 6 shows the arrangement of unit vectors in the first section determined by the principle of FIG. 60 to 120 in the first section _{(V BC}
As described above, the second and third unit vectors V ′ ₂ and V ′ ₃ are used to generate an approximate rotating magnetic field vector. The approximate rotational magnetic field vector φ that is as close as possible to the reference rotational magnetic field vector φ _S, that is, the ideal unit vector arrangement for obtaining an approximate sine wave with few harmonic components is
It is as shown in FIG. In this example, one section (60
(°) is divided into 12 and unit vectors are respectively arranged therein. That is, two types of significant unit vectors V ′ ₂ (110), V ′ ₂ and V ′ are included in the _{first to twelfth} angle sections T _{1 to} T _{12.
2, V '3 (010)} , V' 2, V '3, V' 2, V '3,
V ′ ₂ , V ′ ₃ , V ′ ₃ and V ′ ₃ are arranged in this order, and zero vectors V ′ ₇ (111) and V ′ ₈ (0
00) is arranged. Only the arrangement of the unit vectors of FIG. 6 (A) is written in the first memory M ₁ of FIG. Now, if an upward vector like the unit vector V ′ ₂ in FIG. 6 (A) is called a positive vector and a downward vector like the unit vector V ′ ₃ is called a negative vector, The memory M ₁ has positive, positive, positive,
Data indicating negative, positive, negative, positive, negative, positive, negative, negative, negative is written.

Since the unit vectors generated in the first to sixth sections shown in FIG. 4 are previously determined and written in the second memory M ₂ , they are shown in FIGS. 6 (A) and 9 (B). First shown in
In a section V _'2 is inserted, V at the negative vector' at the positive vector ₃ is inserted. In addition, the second shown in FIG. 9 (C)
In a section V _'3 (010) is inserted, V at the negative vector' at the positive vector ₄ (011) is inserted.
Further, in the sixth section shown in FIG. 9 (A) V at the positive vector _'1 (100) is inserted, V at the negative vector'
₂ (110) is inserted. Similarly, unit vectors are inserted in the third, fourth, and fifth sections.

The switches A _{1 to} C ₂ are moved to the sixth position in the 60 ° to 120 ° section of V _BC .
When the on / off control is performed so as to correspond to the unit vector of FIG. _{6A, the} line voltage V _BC shown in FIG. 6B is obtained. Further, in other angle ranges, the waveform shown by V _BC in FIGS. 9 (A) and 9 (C) is obtained. The distribution of such a pulse is
The harmonic component of the inverter output voltage is reduced, and a rotating magnetic field vector approximate to the reference rotating magnetic field vector can be obtained.

(Frequency and voltage control) By the way, it is required to change the output frequency and voltage of the inverter. In this example, the number of output pulses and the pulse width are kept constant up to a frequency of 55 Hz, and only the output pulse generation cycle is changed. That is, when a frequency command signal lower than 55 Hz is input as the f command signal to the microcomputer (3) in FIG. ₁ , the vector of the first memory M ₁ is based on the clock signal shown in FIG. 6 (C). The arrangement is read and the unit vector V ′ ₁ corresponding to the first to sixth sections
~ V ' ₆ is selectively output from the second memory M ₂ and
The switches A _{1 to} C ₂ of the inverter are on / off controlled so as to correspond to the vectors shown in FIGS. The generation time of each unit vector is set by the timer in the microcomputer (3). When the unit vector ends generated by the time tau ₁ which is set by the timer, to generate a zero vector V _'7 or V' _8. In order to prevent the upper and lower switches in the same arm of the bridge inverter from being turned on at the same time, the non-control period z (no-voltage period) shown by hatching in FIGS. 6, 9, and 10 is shown. ) Is provided. The non-control period z is about 15 μs. The control for keeping the output voltage of the inverter constant within the same frequency is performed by controlling the pulse width (vector generation time) based on the voltage command signal.

When the command to increase the frequency is input, the cycle of the clock signal shown in FIG. 6 (C) becomes shorter. As a result, the zero vector generation time (OFF period) becomes short, the duty ratio becomes high, and the output voltage rises with the relationship shown in FIG. Even if the output frequency and the voltage are increased, the generation arrangement of the unit vector does not change, so that the approximation to the reference rotating magnetic field vector is high and the harmonic components are small.

In this example, when the frequency is increased to 55 Hz, the unit vectors are in the closest state to each other as shown in FIGS. 9 and 10, and even if the clock frequency is simply raised, it becomes impossible to further approach. If the uncontrolled period z exists between the unit vectors in this way, it is impossible to make the line voltage V _BC continuous in the 60 ° to 120 ° section shown in FIG. 9 (B). The respective phase voltages V _A , V _B , and V _C at 55 Hz are shown in FIG.
FIG. 10 (B), 120 ° to 180 ° in the 0 ° to 120 ° section
It becomes as shown in Fig. 10 (C) in the ° section.

In the present embodiment, the control system is changed so that the rotating magnetic field of the circular locus and the approximate sine wave with less harmonic components can be obtained even at frequencies higher than 55 Hz. That is, a preceding unit vector is generated continuously in at least a part of a large number of unit vectors arranged according to the present invention to a subsequent unit vector. 11th
Figures (A) (B) (C) show 0 ° -60 °, 60 ° -1 when increasing the frequency to 56Hz and extending the unit vector generation period.
V _A , V _B , V _C in the interval of 20 °, 120 ° to 180 °,
_Shows V _BC . Now, looking at the first section of 60 ° to 120 ° in FIG. 11 (B), it is indicated by V _BC in FIG. 10 (B).
The number of pulses has changed to six.

The change as described above is performed using the on-time width data of the vector of the third memory M ₁ of FIG. ₁ and the data indicating the place for filling the space between the vectors of the fourth memory M ₄ .
FIG. 8 shows the contents of the fourth memory M ₄ . Y ₁ to Y _{12 in} FIG. 8 indicate positions between unit vectors in the vector arrangement shown in FIG. “1” at each address of Y _{1 to} Y ₁₂ is data for instructing to fill. That is, in the case of 56 Hz, Y ₂ , Y ₄ , Y ₆ , Y ₇ , Y ₉ , Y ₁₁
Is filled with the preceding unit vector. Frequency is 64,6
As it goes up to 8, 72, and 76 Hz, there are more places to be filled, and at 76 Hz, the control is filled with unit vectors that precede each other.

This filling operation will be described in more detail. Now, for example, when 56 Hz is input as the f command signal to the microcomputer (3) in FIG. 1, data indicating the place to be filled is output from the address corresponding to 56 Hz in FIG. First divided every 60 degrees
Since the arrangement of the unit vectors is the same in all of the 6th to 6th sections, the operation of filling with the data of FIG. 8 is performed in the same manner. The control for filling the space between the unit vectors is performed so that the preceding unit vector is extended to the rising time of the following unit vector. For example, Figure 10
When the Y ₂ between the second unit vector V ′ ₁ (100) and the third unit vector V ′ ₁ (100) of (A) is filled, it corresponds to the unit vector V ′ ₁ (100). Switches A _{1 to} C ₂ state (A ₁ on, B ₁ off, C ₁ off,
(A ₂ , B ₂ on, C ₂ on) are maintained between Y _{2 as} well. When the preceding unit vector and the following unit vector are the same, the switches A _{1 to} C ₂ are not switched at all.
However, the fourth unit vector (11
Y) between 0) and the fifth unit vector (100)
_{When 4} is filled, it is necessary to switch the switches B ₁ and B ₂ of the B phase. Therefore, the microcomputer (3) needs to switch the switch between the preceding vector and the next vector. It is determined whether or not there is a switch, and the switch of the phase that needs to be switched is controlled so that a non-control period (no-voltage period) is provided as before. That is, in the mutual Y ₄ , the A-phase and C-phase switches A ₁ , A ₂ , C ₁ , and C ₂ are not switched, but the B-phase switches B ₁ and B ₂ are not controlled. Switch.

Y ₂ in the vector arrangement in FIGS. 10 (A) (B) (C),
_{Y 4, Y 6, Y 7} , Y 9, Y 11 represents a position to fill, Y _2, Y ₄ phase voltages V _A, V _B, in _{V C, Y 6, Y 7} , Y 9, Y 11
Indicates the position where the waveform is continuous. If control is performed so as to fill the positions indicated by Y ₂ , Y ₄ , Y ₆ , Y ₇ , Y ₉ , and Y ₁₁ in FIGS. 10 (A), (B), and (C), the respective phase voltages V _A , V _B , V _C and the line voltage V _BC are shown in FIGS. 11 (A) (B) (C). Line voltage V
_{In BC} , comparing the number of pulses before and after filling,
6 to 5 in the 0 ° to 60 ° section, 60 ° to 120
The number changes from 12 to 6 in the ° section, and from 6 to 5 in the 120 to 180 ° section. Note that P _{1 to} P ₂₄ are also written in FIG. 11 so that the relationship between the pulses P _{1 to} P _{24 in} FIG. 10 and the pulse in FIG. 11 can be understood.

If control is performed so that the vectors are filled with each other, the voltage rises by this amount. Therefore, the pulse width is corrected so that the voltage does not rise suddenly even if the vectors are filled with each other. The correction data of the pulse width is written in the third memory M ₃ of the ROM. The microcomputer (3) is 56, 64,
In response to frequency signals of 68, 72, 76 Hz, pulse width data τ ₁ , τ ₂ , τ ₃ , τ ₄ , τ ₅ corresponding to the frequency is given as a third value.
From the memory M ₃ and the generation period of the unit vector is determined so as to correspond to this pulse width data. Considering now the 24 pulses P _{1 to} P ₂₄ showing the line voltage V _{BC, as} is apparent from the comparison between FIG. 10 and FIG. 11, the pulses P ₃ , P ₄ , P ₆ , P ₇ , P ₉ , P ₁₁ , P ₁₄ , P ₁₆ ,
The trailing edges of P ₁₈ , P ₁₉ , P ₂₁ , and P ₂₂ are adjusted so that they fall quickly. Since the adjustment width of each pulse is the same, 45
The mutual intervals of the 10 pulses of the line voltage V _BC , which are arranged in the section of ° to 135 degrees, are equal to each other. Therefore, if the frequency of the clock signal is increased so as to increase the frequency from 56 Hz to 63 Hz, the pulse width is kept constant and the off period between the pulses is narrowed, which increases the duty ratio and increases the voltage. As shown in FIG. 7, it increases linearly. Even when the frequencies become 64, 67, 72, and 76 Hz, the embedding as shown in FIG. 8 is performed, and the pulse width (ON width) is adjusted at the same time. As a result, finally, as shown in FIG. 12, one pulse in which the 45 ° to 135 ° section of the line voltage V _BC is continuous becomes one pulse. However, three pulses are respectively arranged before and after this, and basically match the pulse arrangement of the line voltage V _BC shown in FIGS. 9 and 10, so that an approximate sine wave and a circular locus with less harmonic components are generated. A rotating magnetic field is obtained.

[Modification]

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

(a) In the embodiment, the 60 degree section (unit section) is evenly divided into 12
Although the unit vectors are divided and the unit vectors are respectively arranged therein, for example, the unit vectors may be divided into 12 and eight unit vectors may be arranged, and the remaining four angular ranges may be zero vectors. in this case,
Positive vector in angle sections 1, 3, 6, 9 and angle section 4,
Negative vectors are placed at 7, 10, 12 and zero vectors are placed at angle sections 2, 5, 11. In addition, 10 division, 10 pulse format, 20 division, 20 pulse format, 26 division, 20
It may be a pulse format or the like.

(b) Instead of writing the vector arrangement in the first memory M ₁ , it may be obtained by calculation.

(c) If the capacity of the ROM is large, the vector arrangement in the 180-degree section or the 360-degree section may be written.

〔The invention's effect〕

As is apparent from the above, according to the present invention, it is possible to suppress the waveform distortion of the line voltage and increase the output voltage.

[Brief description of drawings]

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 present invention, and FIG. 3 shows a switch state of a three-phase inverter and a rotating magnetic field vector. Fig. 4 is a vector diagram showing the relationship, Fig. 4 is a vector diagram showing the arrangement of sections and changes in the approximate rotating magnetic field vector in the first section, Fig. 5 is a vector diagram for explaining the determination of the unit vector, and Fig. 6 is frequency. waveform diagram showing a first view of each part of the state of the case lower, FIG. 7 is a characteristic diagram showing a relationship between a frequency and voltage of the inverter, the FIG. 8 is the contents of the fourth memory M ₄ of FIG. 1 FIG. 9 is a waveform diagram showing the arrangement of vectors and pulses of the line voltage V _BC in the 0 ° to 180 ° section when the frequency is 55 Hz, and FIGS. 10 (A) (B) (C) are Vector when the frequency is 55Hz, each phase voltage, line voltage V Waveform diagram showing the relationship with _BC , Fig. 11 (A) (B) (C) shows each phase voltage when the frequency is 56Hz,
FIG. 12 is a waveform diagram showing the line voltage V _BC, and FIG. 12 is a waveform diagram showing the line voltage V _BC when the frequency is 76 Hz. (1) ... power, (2) ... motor, (3) .... microcomputer, A ₁ -C ₂ ... switch, M ₁ ~M ₄ ... memory.

Claims (1)

[Claims] 1. Six types of units determined by the switching state of the inverter in order to obtain an approximate rotating magnetic field vector (φ) that approximates a reference rotating magnetic field vector (φ _S ) in a three-phase AC load connected to a three-phase inverter. A vector and a zero vector are determined, the angular position of the reference rotating magnetic field vector (φ _S ) is divided into six, and two types of unit vectors and zero vectors selected from the six types of unit vectors are used in each section. Then, the switch of the three-phase inverter is turned on / off to obtain the approximate rotating magnetic field vector, and the inverter output voltage value is changed by changing the ratio of the unit vector generation time to the generation cycle of the unit vector. A method of controlling an inverter, which controls on / off of the switch, comprising increasing an output voltage of the inverter, Two unit vectors of the same type that are adjacent to each other in at least a part of the single or plural locations when the mutual intervals of the occurrence periods of two unit vectors of the same type that are in contact with each other become a predetermined minimum interval in the single or plural locations. A method of controlling an inverter, characterized in that the generation period of the previous unit vector in (1) is extended to the generation time of the subsequent unit vector.