JP6326832B2 - Inverter control method and voltage type inverter - Google Patents

Description

  The present invention relates to PWM modulation control for generating a PWM pattern by a triangular wave carrier signal and a voltage command.

  In a variable speed drive device or the like, PWM modulation control is applied to an inverter circuit to output an AC voltage corresponding to a voltage command. PWM modulation control includes a method of generating a PWM pulse pattern by using a triangular wave carrier signal and a voltage command.

  A method for changing the carrier frequency in accordance with the slope of the reference sine wave is disclosed in Patent Document 3.

  Further, Patent Document 4 discloses that within the control cycle, the carrier frequency is set high in a region where the voltage command value is large, and the carrier frequency is set low in a region where the voltage command value is small.

  Further, the ripple current width Ipr of the motor current controlled according to the PWM control is detected, and a control signal that indicates the carrier frequency of the PWM control is set based on the ripple current width deviation ΔIpr from the reference value Ipr # of the ripple current width. Thus, Patent Document 5 discloses that feedback control of the carrier frequency for maintaining the ripple current width at an appropriate level is realized.

  Patent Document 6 discloses that the carrier frequency of the triangular wave generating circuit is changed depending on the vehicle speed or the rotational speed of the AC machine and the load condition.

JP 2009-100653 A JP 2009-194950 A JP 2010-35260 A JP 2009-291019 A JP 2009-11028 A JP-A-5-184182

  The shape drawn by the current ripple locus as shown in FIGS. 14 and 15 decreases as the carrier frequency fc is increased, and increases as the carrier frequency fc is decreased. This current ripple component causes copper loss and iron loss in the electrical equipment of the load. Therefore, the current ripple is small, that is, the carrier frequency fc is desired to be high, but on the other hand, the number of times of switching per unit time is large. Then, the switching loss of the main circuit increases and the temperature rise of the semiconductor element becomes a problem. In view of this, there is a demand for a method that satisfies both conflicting problems of suppressing current ripple and reducing switching frequency.

  Therefore, it is necessary to select an appropriate carrier frequency fc in consideration of these relationships.

  The points and problems of the conventional example will be described.

  In the voltage type PWM inverter, the current ripple changes depending on the carrier frequency. For the load machine, an increase in copper loss and a harmonic iron loss in the iron core due to the current ripple component increase, and therefore it is preferable that the current ripple component is smaller, that is, the carrier frequency is higher. On the other hand, for the inverter, if the carrier frequency is increased, the switching frequency (occurrence frequency of ON / OFF operation) of the switching element is increased, and the switching loss causes an increase in the temperature of the element and a decrease in efficiency.

  Therefore, the carrier frequency is usually selected in consideration of the total loss of the inverter and the load machine and the respective cooling capacity.

  It has been considered to change the carrier frequency in accordance with the operating conditions. Conventionally, there are many methods in which the carrier frequency is variable based on the magnitude of the load current and the rotation speed (frequency).

  In a certain method, when the load current is large, the switching loss increases. Therefore, the carrier frequency is lowered in accordance with the increase in the load to reduce the number of times of switching, thereby suppressing an increase in the heat generation amount of the element.

  Other than that, in the variable speed drive of an AC motor, the voltage also changes almost proportionally to the rotation speed (frequency) as in V / F control, so that the total loss and There is also a method of changing the carrier frequency in consideration of the temperature rise of each part.

  However, in the conventional method, although the carrier frequency is switched, the frequency is changed suddenly at the time of switching, or the time change is moderated to take a time sufficiently longer than the cycle of the rated frequency.

  On the contrary, a method for controlling the carrier frequency in consideration of the ripple of the current ripple that changes every 60 ° of the fundamental period has not been disclosed yet. Therefore, although the maximum value of the current ripple varies if the carrier frequency is constant, conversely, a method of calculating the carrier frequency so that the maximum value of the current ripple is fixed to a constant value can be considered. If this method is illustrated as a current ripple, the maximum value of the current ripple can be made constant by making the envelope, which was elliptical in FIG. 28, circular as shown in FIG.

  If the radius of the circular envelope in FIG. 30 coincides with the long axis of the ellipse in FIG. 28, the PWM carrier period of the short axis portion is lengthened, which is a limited period of the short axis portion. The carrier frequency can be reduced, and thus the average switching frequency can also be reduced.

  As described above, in the inverter device, it is a problem to select an appropriate carrier frequency, to reduce switching loss and to suppress current ripple.

  The present invention has been devised in view of the above-described conventional problems, and one aspect thereof is an inverter control method for controlling an inverter that outputs a three-phase AC voltage by a PWM modulation method at a variable voltage variable frequency. The voltage command is updated in synchronization with the upper and lower vertices of the triangular wave carrier signal, or the period that is an integral multiple of the voltage command is used as the update cycle, and the voltage command is input and three-phase modulation is performed based on the reference carrier frequency. The voltage command is corrected by applying zero-phase modulation by the method or the two-phase modulation method, and the vector locus of the current ripple of the output current during the half cycle of the triangular wave carrier signal is estimated using the corrected voltage command. , Calculate the norm of the vector locus of those current ripples, select the maximum norm that is the maximum component of the norm, and the value of the maximum norm of the current ripple is the allowable current ripple. The carrier frequency correction coefficient is calculated so that it matches, the value obtained by multiplying the reference carrier frequency by the carrier frequency correction coefficient is set in the carrier frequency command, the corrected voltage command, and the triangular wave carrier based on the carrier frequency command And a PWM pattern for controlling on / off of the switching element using the signal.

  When estimating the current ripple norm, the rotational coordinate system is taken, and the phase angle of the vector of velocity electromotive force due to the field magnetic flux is regarded as equal to the phase angle obtained by converting the voltage command into polar coordinates, and is used for rotational coordinate conversion. The value obtained by adding the reference phase and the phase angle obtained by converting the voltage command to the polar coordinate is used as the reference axis of the rotation coordinate, and the corrected voltage command vector is selected as the reference axis of the rotation coordinate. Calculate the current ripple trajectory from the voltage component of the difference between the scalar amount of the output voltage vector of the reference axis and the velocity electromotive force scalar amount due to the field magnetic flux and the time interval at which the voltage vector is output. It is characterized by estimating.

  When calculating the norm of the vector locus of the current ripple, the current norm of the reference axis and its orthogonal axis are multiplied by different weighting factors, and the norm is calculated from each axis component after the weighting factor multiplication. The maximum norm is selected from.

  According to the present invention, it is possible to select an appropriate carrier frequency in the inverter device. Thereby, reduction of switching loss and suppression of current ripple can be achieved.

The whole block diagram of the PWM inverter in V / F control. Explanatory drawing which shows the definition of a three-phase coordinate and a rectangular two-phase coordinate. Explanatory drawing which shows the definition of a voltage command and a PWM output voltage pattern. The time chart which shows a triangular wave carrier signal. Schematic which shows the structure of a PWM comparator. The time chart which shows the zero phase modulation example of three phase modulation. The time chart which shows the zero phase modulation example of two phase modulation. The block diagram which shows the inverter control apparatus by zero phase modulation (three phase modulation). The block diagram which shows the inverter control apparatus by zero phase modulation (two phase modulation). A time chart (three-phase modulation) showing a PWM carrier signal, a voltage command, and a PWM pattern. The time chart (two phase modulation, P side shift) which shows a PWM carrier signal, a voltage command, and a PWM pattern. The time chart (two phase modulation, N side shift) which shows a PWM carrier signal, a voltage command, and a PWM pattern. Explanatory drawing which shows a voltage command vector, an induced electromotive force, and a PWM pattern (fixed coordinate system). Explanatory drawing (PWM period) which shows the current vector locus by zero phase modulation (three phase modulation). Explanatory drawing (PWM period) which shows the current vector locus by zero phase modulation (two phase modulation). Explanatory drawing (PWM period, fixed coordinate) which shows the norm component of the current ripple by zero phase modulation (three phase modulation). Explanatory drawing (PWM period, fixed coordinate) which shows the norm component of the current ripple by zero phase modulation (two phase modulation). Explanatory drawing which shows the relationship between a voltage command vector, an induced electromotive force, and a PWM pattern (induced electromotive force reference coordinate system). Explanatory drawing which shows the norm component of the current ripple by zero phase modulation (three phase modulation) (PWM period, induced electromotive force coordinate system). Explanatory drawing which shows the norm component of the current ripple by zero phase modulation (two phase modulation) (PWM period, fixed coordinate system). Explanatory drawing (fixed coordinate system) which shows the relationship between a voltage command vector and the conditions with equal induced electromotive force, and a PWM pattern. Explanatory drawing which shows the relationship between a voltage command vector and an induced electromotive force being equal, and a PWM output pattern (induced electromotive force reference coordinate). Explanatory drawing which shows the norm component of the current ripple by the zero phase modulation (three-phase modulation) on condition that a voltage command vector and an induced electromotive force are equal (PWM period, fixed coordinate system). Explanatory drawing which shows the current ripple norm component by zero phase modulation (three-phase modulation) on the conditions where a voltage command vector and induced electromotive force are equal (PWM period, induced electromotive force reference coordinate system). Explanatory drawing which shows the norm component of the current ripple by zero phase modulation (two phase modulation) (PWM period, fixed coordinate system). Explanatory drawing which shows the norm component of the current ripple by zero phase modulation (two phase modulation) (PWM period, induced electromotive force reference coordinate system). Explanatory drawing which shows the phase command group from which the phase angle changed. Explanatory drawing which shows the vector locus of a current ripple. Explanatory drawing which shows the relationship between the current vector of a fundamental wave component, and a current ripple locus. Explanatory drawing which shows the vector locus | trajectory of a current ripple when the maximum norm of a current ripple is made equal. The block diagram which shows the inverter control apparatus of the variable carrier frequency system in a three-phase modulation system. Explanatory drawing which shows the locus | trajectory of the current vector between the carrier periods in two-phase modulation. Explanatory drawing which shows the locus | trajectory group which extracted only the current ripple component between the carrier periods in two-phase modulation. The block diagram which shows the inverter control apparatus of the variable carrier frequency system in a two-phase modulation system. The time chart which shows the simulation result of a voltage command, a PWM modulation waveform, and a current waveform. The time chart which shows the simulation result of a voltage command, a PWM modulation waveform, and a current waveform. 28 is an explanatory diagram showing the current ripple component of FIG. 28 obtained by subtracting the fundamental wave component (fixed coordinate system, e ₀ vector is around ₀ to 70 deg). Explanatory view showing the current ripple component of Figure 30 obtained by subtracting the fundamental wave component (around fixed coordinate system, e ₀ vector 0~80deg). The figure which shows the relationship between a modulation factor (amplitude of a voltage command) and an average carrier frequency. Explanatory drawing which shows the current ripple locus | trajectory at the time of fixed carrier frequency (two phase modulation). Explanatory drawing which shows the current ripple locus | trajectory at the time of carrying out variable control of the carrier frequency (two-phase modulation). Explanatory drawing which shows the current ripple locus | trajectory at the time of carrying out variable control of the carrier frequency (two-phase modulation, Embodiment 5). Time chart (two-phase modulation) showing simulation results when carrier frequency is fixed. The time chart which shows the simulation result at the time of carrying out variable control of the carrier frequency (two phase modulation). The time chart which shows the simulation result at the time of carrying out variable control of the carrier frequency (two-phase modulation, Embodiment 5).

[1. overall structure]
Before describing the contents of the present invention, first, a V / F control method for a variable speed drive device will be described as a basic prior art example.

Various coordinate systems are defined as shown in FIG. Each phase in the three-phase AC output of the main circuit is the U phase, V phase, and W phase, the biaxial coordinates of the fixed coordinate system based on the U phase are αβ coordinates, and the phase angle that integrates the angular frequency command ω ^* The biaxial coordinate of the rotating coordinate system synchronized with θ is defined as the dq coordinate. Here, it is assumed that the initial phase of the rotational coordinates coincides with the α axis and the d axis when time t = 0.

Due to the PWM control, the phase voltage, which is the output of the main circuit, is output in a time-division _{manner as} binary levels of V _p = + V _dc / 2 and V _n = −V _dc / 2, and there are three phases. There are a total of eight types of PWM pattern states as combinations of phase voltages. In the following, eight types of voltage vectors represented on the αβ coordinates as shown in FIG. 3 are defined and expressed as equivalent voltage vectors instead of the PWM pattern.

  FIG. 1 is a control block diagram of a basic V / F control method. Here, in order to make the principle easy to understand, an example of a continuous type (analog type) is shown, and each component is listed below.

(1) Input command and biaxial voltage command calculation unit 1
Input command is an angular frequency command corresponding to the rotation speed omega ^*, in this angular frequency command omega ^* biaxial voltage command calculation unit 1, the angular frequency command omega ^* and approximately proportional to the biaxial voltage command V _d ^*, V _q ^* is generated. This is the voltage of the orthogonal biaxial component of the rotating coordinate system (dq coordinate) defined in FIG. A control method in which the angular frequency and the voltage are substantially proportional is called V / F control.

(2) Carrier generator 2
A triangular wave carrier signal Cry necessary for generating a PWM pattern is generated. This is a triangular wave signal having a carrier frequency command f ^* c as shown in FIG. 4 as a frequency and an amplitude of ± 1. A timing signal (sample signal) S _smpl for synchronizing the triangular wave carrier signal Cry and other control signals is generated at the time of the upper and lower vertices of the triangular wave.

(3) Phase voltage command generator 3
With respect to the biaxial voltage commands V _d ^* and V _q ^{* in the} rotational coordinate system, the rotational coordinate conversion (P ⁻¹ ) is performed to convert the voltage component of the αβ coordinate, and the two-phase / three-phase conversion (C ^−1). ) And converted into a three-phase sine wave AC voltage command (hereinafter referred to as a three-phase voltage command) V ′ _u , V ′ _v , V ′ _w . The three-phase voltage commands V ′ _u , V ′ _v , and V ′ _w further add “zero-phase components including third-order harmonics” to the three phases by the same amount “zero-phase voltage correction (hereinafter referred to as zero-phase modulation). ) ”Is applied to convert the voltage commands into new voltage commands V _u , V _v , and V _w . It is known that by applying this zero phase modulation, the ratio of the maximum output voltage to the DC power supply (voltage utilization factor) can be increased by about 15%. There are many methods for this zero-phase modulation, and in this specification, “two-phase modulation”, which will be described later, is used, but here, a basic “three-phase modulation” zero-phase modulation method is also included. For the sake of clarity, it is represented by the name “zero phase modulation”.

(4) PWM comparator 4
Here, the triangular wave carrier signal Cry and the three-phase voltage commands v _{u_pwm} , v _{v_pwm} , and v _{w_pwm} are compared to generate a PWM pattern. FIG. 5 shows an example in which the three-phase PWM comparator 4 is composed of a large and small comparator. Since the DC link voltage V _dc of the main circuit 6 fluctuates due to voltage fluctuations of the power supply system, the voltage commands V _u , V _v , and V _w are 1/2 times the DC link voltage V _dc (= V _dc / 2). By dividing, each voltage command of V _p = + V _dc / 2 and V _n = −V _dc / 2 is converted into a command value of unit method corresponding to the amplitude of the triangular wave carrier signal Cry of +1 and −1. And the triangular wave carrier signal Cry are sequentially compared in magnitude.

If the triangular wave carrier signal Cry is smaller than the voltage commands V _{u_pwm} , V _{v_pwm} , V _{w_pwm of} each phase, the upper arm side is turned on (+ Vdc / 2 output, H mode), and the triangular wave carrier signal Cry is larger. For example, the PWM pattern pwm (u, v, w) is output so that the lower arm side is turned on (-V _dc / 2 output, L mode).

  Here, the comparison condition does not include an equal sign as in (a> b), but actually, one of the rising (UP) and falling (DOWN) of the triangular wave carrier signal Cry is changed from “>”. There is an improvement method such as maintaining symmetry by replacing “≧”, but it has a simple notation because it is not directly related to the present invention.

Further, here, since it is handled as a discrete system, a sample hold circuit S / H is inserted in front of the PWM comparator 4 so that the voltage commands V _u , V _v , V _w are the apexes of the triangular wave carrier signal Cry (sample signal S _smpl ). It is updated (sampled) only at this time, and is held (held) during other periods. As a result, the voltage command input to the PWM comparator 4 is converted into a stepwise discrete value, as in _{va_pwm} , _{vb_pwm} , and _{vc_pwm} in FIG.

(5) Dead time generation / gate drive unit 5
Here, PWM patterns pwm of the three-phase generated by the PWM comparator 4 (u, v, w) of the gate signal g _{_u} for on-off control the switching elements of the sixth _{arm, g _v, g _w, g} _x , to convert g _{_y,} to g _{_z.} Actually, there is a delay time of the switching operation in the switching element. Therefore, when the ON / OFF state of the upper arm and the lower arm are alternately switched, a short-circuit prevention period (dead time) in which both arms are turned OFF is inserted. the gate signal _{g _u, g _v, g _w} , g _x, g _y, are generating g _{_z.}

(6) Main circuit 6
This is an inverter circuit section that generates a PWM voltage to be actually output. Here, a three-phase six-arm bridge circuit as shown in FIG. 1 is used. The DC link voltage is V _dc as described above.

The above is the component of FIG. 1, and the power switching element of the main circuit 6 is PWM controlled to amplify and output the angular frequency command ω ^* and the three-phase AC output voltage having an amplitude substantially proportional to the angular frequency command ω ^* . To do. In the present invention, new functions are added to the carrier generation unit 2 and the phase voltage command generation unit 3 which are the parts (2) and (3) in this configuration example.

[2. Three-phase voltage command generation method (classification and definition of zero-phase modulation method)]
In this section, in order to facilitate the explanation of the content of the present invention, first, mathematical expressions are defined for the conventional example of the phase voltage command generation unit 3.

As a method for generating the three-phase voltage commands V ′ _u , V ′ _v , and V ′ _w , in addition to the above method, a “three-phase sine wave vibrator with 120 ° phase difference” can be used, or “polar coordinates → three There are various realization methods such as using a phase converter, but in recent years, voltage commands V _d ^* and V _q ^* are given as two-axis coordinate components of a rotating coordinate system as shown in FIG. A method of converting coordinates to three-phase voltage commands V′u, V′v, and V′w is common.

The voltage command is given as two-axis voltage commands V ^* _d and V ^* _q which are two-axis coordinate components (dq components) of the rotating coordinate system as shown in the following equation (1). The biaxial voltage commands V ^* _d and V ^* _q can be regarded as voltage vectors. On the rotating coordinate (dq coordinate) that is moved by the phase angle θ obtained by time-integrating the angular frequency command ω ^* (electrical angle), When the voltage command vector is observed, the voltage amplitude is | V ^* | and the phase is a function of the phase angle ψ _V from the d-axis.

Here, the reference phase θ of the rotation coordinate conversion is obtained by integrating the rotational angular velocity number (angular frequency command) ω ^* as shown in the following equation (2).

  The coordinate transformation matrix used in FIG. 1 is defined below.

The transformation matrix C from the three-phase component to the biaxial coordinate system and its inverse transformation matrix C ⁻¹ are expressed by the following equation (3).

In the biaxial coordinate system, a transformation matrix P for transforming from a fixed coordinate system to a rotating coordinate system and its inverse matrix P ⁻¹ are expressed by the following equation (4).

If the rotating coordinate conversion equation (5) and the two-phase three-phase conversion equation (6) are applied to the two-axis voltage commands V ^* _d and V ^* _q (1) for the dq-axis component, the three-phase AC component can be obtained. Can be converted.

  Next, zero phase modulation for improving voltage utilization will be described. Here, two types of “three-phase modulation” and “two-phase modulation” will be described.

  First, a method of “constant maximum / minimum amplitude” as “three-phase modulation” is shown. Although this method is not used in the embodiment of the present specification, it is described because it is a basic principle.

In this method, first, the maximum voltage v_max and the minimum voltage V_min at each time are selected according to the equation (7) from the three-phase voltage commands V ′ _u , V ′ _v and V ′ _{w in the} equation (6).

Next, as shown in equation (8), an intermediate value between the maximum value and the minimum value of equation (7) is calculated to calculate a zero-phase modulation correction value V _{ofs — 3ph} .

Then, a correction value V _{ofs — 3ph} common to all phases is added to the three-phase sine wave of equation (6) as in the following equation (9).

FIG. 6 is a waveform chart of an example in which zero-phase compensation called “three-phase modulation” is applied, from a maximum value and a minimum value of a three-phase voltage command V ′ _u , V ′ _v , V ′ _w of a sine wave to a triangular wave A correction value V _{ofs — 3ph} of a zero-phase voltage component having a similar shape is calculated, and this is added to all phases of the three-phase voltage commands V ′ _u , V ′ _v , and V ′ _w to be voltage commands V _u , V _v , V It converts to a waveform like _w . It can be seen that the waveform after this conversion has a shape in which the vicinity of the maximum value of the sine wave is dented, and even when the maximum amplitude is reduced, that is, the same DC link voltage V _dc can be output. In addition, the maximum and minimum values of the corrected voltage commands V _u , V _v , and V _w are characterized by positive and negative symmetric waveforms with respect to the 0 level at any time. The symmetry of the trajectory shape can be realized.

  Next, zero phase modulation called “two-phase modulation” will be described. Instead of (7) and (8), the zero-phase voltage is calculated by the following (10) and (11), and for all phases in (12) as in the case of three-phase modulation. To correct. There are many other calculation methods, but here, a general example in which the waveform after modulation is positive-negative symmetric and has periodicity every 120 degrees is shown.

The correction value V _{ofs — 2ph} of the zero-phase modulation component of the two-phase modulation is compensated by adding the in-phase component for the three phases according to the following equation (12), similarly to the zero-phase modulation of the three-phase modulation.

FIG. 7 shows an example of correction of the zero-phase voltage called “two-phase modulation”, and the correction value _{ofs_2ph} in the equation (11) is _added to the three-phase voltage commands V ′ _u , V ′ _v and V ′ _w . As a result, the difference voltage between the maximum value of the original three-phase voltage command V ′ _u , V ′ _v , V ′ _w and the P-side potential + V _dc / 2 of the DC link voltage V _dc , and the original three-phase voltage command The difference voltage between the minimum value of V ′ _u , V ′ _v , V ′ _w and the N side potential −V _dc / 2 of the DC link voltage V _dc is compared, and the difference voltage on the maximum voltage side (P side) is small. In such a case, the minimum value of the three-phase alternating current is corrected so as to be in contact with the N-side potential when the difference voltage on the minimum voltage side (N side) is small.

  The period in contact with the P side and the N side alternately occurs every 60 degrees, and only one phase in the section is at a level that coincides with the apex of the triangular wave carrier signal Cry. Switching stops without occurring. As a result, when the following average switching frequency is defined, when “two-phase modulation” is applied, the average frequency is reduced to about 2/3 as compared with the case where “three-phase modulation” is applied. be able to.

(Definition of average switching frequency)
The number of times of switching is defined as "one set of switching from the change from L to H mode and the opposite change from H to L mode", and the number of times of switching in a time interval that is an integer multiple of the fundamental wave period. The value divided by the time interval is defined as the “average switching frequency” of the phase. If the phase is not specified, the average of the three phases is indicated as “average switching frequency”.

  The above is an example of the configuration of a V / F control type inverter system using PWM control in consideration of general zero-phase modulation. In the selection of three-phase modulation and two-phase modulation, the restriction that a pulse with a narrower time width than the dead time is not often applied, and when the output voltage (PWM modulation rate) is small, three-phase modulation On the contrary, when the zero phase modulation is large, the zero phase modulation of the two phase modulation is often selected.

[3. Three-phase PWM pattern generation method]
In this specification, in the PWM converter having the phase voltage command generation unit 3 to which the above-described zero-phase modulation is applied, in order to reduce the average switching frequency, the carrier frequency is calculated using the current ripple component flowing through the load as an evaluation function. This is a variable control method. In order to explain the contents, it is necessary to define a current ripple and to express a mathematical expression representing it. Therefore, the current ripple component when the voltage generated by the PWM generation method as described above is output will be described and defined.

  FIG. 10 shows an example in which the zero-phase component of the three-phase modulation of equation (8) is corrected with respect to the voltage command and then converted into a discrete system by the sample hold circuit S / H of FIG. A triangular wave carrier signal Cry is also shown.

In the two-phase modulation, the case where the original three-phase alternating current is in contact with the P potential by applying the zero-phase component correction value V _{ofs — 2ph} in the expression (1) of the expression (11) is shown in FIG. FIG. 12 shows the case where the original three-phase alternating current is in contact with the N potential by applying the zero-phase component correction value V _{ofs — 2ph} in FIG.

  The variables and symbols in the three types (FIGS. 10, 11, and 12) basically use the variables in FIG. Further, additional information is added so that the time of the sample timing and the hold period between them can be distinguished. The explanation of these elements and symbols is listed below.

Cry: triangular wave carrier signal Down period: period in which the triangular wave carrier signal changes in the negative direction UP period: period in which the triangular wave carrier signal changes in the positive direction t _(n) , t (m), t (n + 1), t (m + 1) : The time at the apex of the triangular wave carrier signal Cry, which is a sample (update) timing of control configured in a discrete system. Since it occurs twice in the carrier period, the upper vertex is n, the lower vertex is m, and v ^* _n , v ^* _m , v ^* _{(n + 1)} , v are expressed as (n + 1) and (m + 1) in the next period. ^* _{(m + 1)} : A voltage command vector obtained by sampling the times t (n), t (m), t (n + 1), and t (m + 1). This is not used in the figure or explanation, but is defined for the moment θ _vn , θ _vm , θ _{v (n + 1)} , θ _{v (m + 1)} : Voltage command vector v ^* _n , v at each sample time The phase angle of ^* _m , v ^* _{(n + 1)} , v ^* _{(m + 1)} is expressed with a fixed coordinate reference. In the case of time t (n), the phase angle θ _vn is a combination of the reference phase θ (t (n)) of the rotating coordinate system and the phase angle φ _v (t (n)) of the voltage vector viewed from the rotating coordinate reference. = Θ (t (n)) + φ _v (t (n)). Then, t (n), t ( m), t (n + 1), t (m + 1) v a_pwm that defines the phase angle at each time of _{(n), v b_pwm (n} ), v c_pwm (n): the time t Among the three-phase voltage commands v _{u_pwm} (n), v _{v_pwm} (n), and v _{w_pwm} (n) _{sampled in} (n), the maximum voltage phase component is _{va_pwm} (n) and the intermediate voltage phase is v _{b_pwm} (n ), _{And let the} minimum voltage phase be v _{c_pwm} (n). In addition, subscripts (n, m, n + 1, m + 1) of the sample time are added in “()”, and voltage commands that change with time are classified and specified.

The reason for using this is that, in three-phase alternating current, the magnitude relationship of the phase voltage commands is switched depending on the phase angle, and therefore the order of the generation times of the PWM patterns is switched and becomes complicated. Therefore, in order to simplify the explanation, a parametric technique is used in which the u, v, and w phases are not used directly, but are replaced with the a, b, and c phases in order of the phase voltage as shown in Table 1. Decided to do. The relationship between the u, v, w phase and the a, b, c phase is determined by the phase angle θ _v = (θ + ψ _v ) on the fixed coordinates of the three-phase voltage command as shown in Table 1.

t _a (n), t _b (n), t _c (n): time of intersection of the three-phase voltage command of _{va_pwm} (n), v _{b_pwm} (n), v _{c_pwm} (n) and the triangular wave carrier signal Cry It is. n, m, (n + 1), and (m + 1) are the voltage command sample timings, and (a, b, c) are the phases of the voltage command and indicate the phases in which switching occurs.

pwm (a), pwm (b), pwm (c): Output of the magnitude comparison result between the voltage commands v _{a_pwm} (n), v _{b_pwm} (n), v _{c_pwm} (n) of each phase and the triangular wave carrier signal Cry Signal. The PWM state of each phase changes to PWM patterns pwm (a), pwm (b), and pwm (c) at times t _a (n), t _b (n), and t _c (n). It is a signal for each phase. “V” is output during the period of V a — _pwm (n)> Cry, and “L” is output during the period of V a — _pwm (n) <Cry. Here, in the case of “H”, the upper arm element is turned on, that is, the P potential + V _dc / 2 of the DC power source is output, and in the case of “L”, the lower arm element is turned on and the N potential of the DC power source is output. -V _dc / 2 is output.

t ₀ (n), t _λ (n), t _μ (n), t ₇ (n): Considering the sample period corresponding to the carrier half-cycle in the DOWN state from t _{( n)} to t _(m) Three switching times (time t _a (n), t _b (n), t _c (n)) occur in the period. Using this time as a delimiter, t _{(n) to} t _(m) can be divided into four periods. The time width of this section is t ₀ (n), t _λ (n), t _μ (n), t ₇ (n) in the order of occurrence. The relationship between the voltage commands v _{a_pwm} (n), v _{b_pwm} (n), v _{c_pwm} (n) and the triangular wave carrier signal Cry in each section, and the PWM patterns pwm (a), pwm (b) of each phase generated thereby. , Pwm (c) have the relationship shown in Table 2.

  This symbol (0, λ, μ, 7) indicates the period of the carrier half cycle, but when the triangular wave carrier signal Cry is in the UP state, the generation order is reversed to 7, 7, x, 0. However, the PWM pattern with this symbol is common to both DOWN / UP.

Since the magnitude relationship is set as v _{a_pwm} (n), v _{b_pwm} (n), and v _{c_pwm} (n), _among the eight types of three-phase PWM patterns (voltage vectors), between the carrier half-cycles as shown in Table 2 There are four types of PWM states generated at the same time, and the generation order is also determined.

FIGS. 10, 11 and 12 illustrate the principle. The voltage commands V a — _pwm , V _{b — pwm} , V _{c — pwm} and the triangular wave carrier signal Cry which changes with time are compared in magnitude to generate PWM. ₀ (n), t _λ (n), t _μ (n), and t ₇ (n) are numerical calculations at that time if the voltage commands V a — _pwm , V _{b — pwm} , V _{c — pwm} and the carrier frequency are determined as shown in Table 3. Can also be obtained. That is, the PWM pattern can be predicted when the command value is determined without actually generating the triangular wave carrier signal Cry, and this will be used in the following description.

In the two-phase modulation of Table 3, there are columns of T ₀ = 0 and T ₇ = 0. T ₀ = 0 means that in FIG. 11, the condition is V a — _pwm (n) = + V _dc / 2 (P potential). As a result, the switching time when changing to H and changing to L becomes T _a (n) = t (n), t _a (m) = t (n + 1), and different changes occur simultaneously in the same phase. Switching does not occur in the phase where the voltage phase is P potential. Similarly, since T ₇ = 0 has a relationship of V _c — _pwm (n) = − V _dc / 2 (N potential) in FIG. 12, this time t _c (n) = t _c (m) = t ( Two types of switching changes occur at the same time of m). Eventually, no switching occurs in the phase where this voltage phase becomes N potential.

When the time at which this switching occurs and the phase order thereof are determined, the relationship between the u, v, w phase and the a, b, c phase is determined by the range of the phase angle θ _v as shown in Table 1, and therefore each PWM period t _{0 (n), t λ (n} ), t μ (n), output at t ₇ (n) voltage _{V 0 (n), V λ} (n), V μ (n), V 7 (n) is 3 Are replaced with the following eight types of voltage vectors.

  Table 8 shows the eight types of voltage vectors in Table 4.

As an explanation of Table 5, consider an example of voltage vector V ₄ = [v _u , v _v , v _w ] ^T. Since the three-phase switching state and the DC link voltage V _dc to the output voltage of each phase V ₄ are determined as shown in the following equation (13), this is expressed by the inverse rotation coordinate transformation matrix C ^{−1 in} equation (5). Biaxial components as shown in Table 5 can be obtained by converting to biaxial coordinates and applying the two-phase three-phase transformation of equation (6).

Here, in order to combine the coefficient √2 / 3 of the two-phase / three-phase conversion and the coefficient 3/2 generated by the three-phase synthesis, a three-phase voltage that outputs only two values of ± V _dc / 2 is used as a biaxial component. In the amplitude component of the voltage vector converted to, a coefficient of √3 / 2 (V _dc / 2) appears as shown in equation (13).

[4. Current vector locus by PWM pattern and current ripple component]
Up to Section 3, the PWM pattern generation method and its voltage vector have been described. The current ripple of the output current (current flowing through the terminals U, V, and W in FIG. 1) generated by the PWM output voltage component which is the core will be described in terms 4 and 5. First, in this section, the explanation will be developed in the basic fixed coordinate system, and then, it will be developed in the rotational coordinate system and the newly defined rotational coordinate system based on the speed electromotive force.

(Current ripple component in fixed coordinate system)
When a synchronous motor is selected as an example of the load, the voltage / current equation expressed in the orthogonal biaxial coordinate system (αβ coordinate) of the fixed coordinate system is expressed by equation (14). A plurality of output voltage V ₀ in the PWM control, V _lambda, V _mu, divided at the V _7, by outputting a voltage command [v ^* α v ^* _β] equivalent voltage, the fundamental wave component of the current A ripple of high frequency components is superimposed.

  Here, since a variable that changes over time and a variable that approximates to be constant during the sample period coexist, the variable that changes over time will be indicated by adding “(t)”.

Next, by applying the approximation [e _α (t) e _β (t)] ^T = [e _α0 e _β0 ] (constant) and further moving the differential term to the left side, the state equation of equation (15) Can be converted to a format.

Carrier frequency f _c is sufficiently higher than the angular frequency command omega ^*, displacement of the phase angle to rotate the sample period (θ _vn → θ _vm) assuming negligible for small, _[e α (t) e _β (t)] ^T = [e _α0 e _β0 ] (constant). If the voltage drop component of the winding resistance R is also small and ignored, the current component can be approximated to an initial value and a simple integral component as shown in equation (17).

Then, [i _α (t) i _β (t)] ^T expressed by the equation (18) obtained by removing the initial current vector from the equation (17) is hereinafter referred to as “current ripple vector locus” or “current ripple”. I will call it.

  Although the synchronous motor is described here as an example, it is clear that induction machines and system linkage devices can be similarly handled as long as the electromotive force component can be approximated to be constant.

As shown in Table 4, since the switching order of the three phases differs every 60 degrees depending on the phase angle of the voltage command, here also the a, b, and c phases, which are not the u, v, and w phases but the phase voltages are in order of magnitude. Will be handled in Further, in the following description, in order to provide a specificity, an example of a voltage command in which the phase is 0 deg <(θ _v ) <60 deg as shown in FIG. 13 is used.

In fact, such a voltage drop of the delay time and the element of the switching element is generated, where the assumption of an ideal PWM control, voltage command ^{_{[v * α v * β]}} T equivalent to the voltage vector V _{0, Assume} that V _λ , V _μ , V ₇ and output periods T ₀ , T _λ , T _μ , T ₇ can be realized.

The velocity electromotive force due to the field magnetic flux is expressed as [e _α0 e _β0 ] ^T, and it can be assumed that the change between the sample periods is small and can be approximated to be constant.

  Using the voltage component and the time component defined as described above, the change component of the current vector of the carrier cycle is obtained.

The time change (current ripple) [i _α (t) i _β (t)] ^T of the current vector in equation (18) is the output voltage [v _α (t) v _β (t)] ^T output from the inverter and the speed The difference vector of electromotive force [e _α0 e _β0 ] ^T is obtained by time integration. Output voltage [v _α (t) v _β (t)] ^T is PWM, and the output voltages of V ₀ , V _λ , V _μ , and V ₇ are at time intervals of T ₀ , T _λ , T _μ , and T _7. Therefore, if the change amount of the current vector changing between the switching times is approximated by “product of voltage and time”, the current change amount as shown in Table 6 is obtained. The upper four rows are the DOWN period of the triangular wave carrier signal Cry, and the lower four rows are the UP period, and the generation order of these is reversed.

Although the locus of the current vector is to that by integrating the current change amount in Table 6, the prediction of the current ripple of the sample period (n) is a current vector i _n the start time as the initial value, by calculating the cumulative thereto That's fine. Actually, A / D conversion for current detection, PWM setting calculation, and current ripple prediction calculation, etc. cannot be performed simultaneously at the instant of sample time t _n. It is necessary to take measures such as adding the predicted change in the fundamental current between sample periods, but here it is handled in an ideal state.

  If the current change amount of each period in Table 6 is integrated from the initial value, the current vector at each switching time can be predicted as the current i in Table 7. Here, both “(n)” indicating the voltage update timing and “a, b, c” of the switching phase for indicating the switching time are appended to the current i in Table 7.

Since the half cycle is the basic unit this time, in Tables 6 and 7, “(n)” is added in the DOWN period of the triangular wave carrier signal Cry, and “(m)” is added in the UP period. Regarding the evaluation period of the current ripple, not only the case of the half cycle unit of the triangular wave carrier signal Cry but also the case of the unit of one cycle can be considered. In the case of a half cycle unit, the initial value i (n) of the DOWN period and the initial value i (m) of the UP period may be reset to zero, and only the amount of change in the current trajectory thereafter may be considered. If it is sufficient to integrated continuously regarded as the initial value i _m of the (n) the final estimated current vector for the next carrier half cycle of the period (m) is when evaluating one cycle.

  Since the current locus was calculated as shown in Table 7, the definition of the current ripple component will be considered next. For the current ripple, the difference vector between this reference point and the current vector at each switching time, with the carrier half-cycle as the evaluation target and the initial value i (n) or i (m) of each section of DOWN / UP as the reference point The component is regarded as a current ripple. Then, the distance between the locus of the current ripple and the reference point is obtained, and this is called the “norm of the current ripple”. The maximum value of the current ripple norm is defined as “the maximum norm of the current ripple”. In the case of PWM control, one of the current ripples becomes the condition of this maximum norm at the switching time that occurs three times in the carrier half cycle.

  However, in practice, if the starting point between carrier half-cycles is used as a reference point, there is a problem that this does not coincide with the end point of the current ripple. When an example of the locus of the current vector in Table 6 is drawn on the fixed coordinate system, as shown in FIGS. 14 and 15, i (m) after a half cycle of the carrier with respect to the initial value i (n) of the current vector. ) Will move. FIG. 14 shows the case of three-phase modulation, but the same applies to the case of two-phase modulation shown in FIG.

Therefore, as a new reference point for the DOWN period, an intermediate point i _nm between the start point i (n) and the end point i (m) of the carrier half cycle is employed as shown in the equation (19), and the current ripple component is also the new reference point. I decided to make corrections to comply with the points. Actually, the difference between the start point and the end point is the amount of movement of the fundamental wave component, and it changes at a constant speed as time passes.However, the exact calculation considering it is too complicated, so it approximates the intermediate point. I decided to simplify the calculation.

Similarly, for a half cycle of the carrier UP period also, the midpoint i _m of the starting point i _m and end points i _{(n + 1)} of the _{(n + 1)} as a new reference point for the period (m).

  Since the reference point for calculating the current ripple is updated in this way, the current ripple component at each switching time is also corrected as shown in Table 8. This shows an example of three-phase modulation, and when this is represented by a vector diagram, the components are as shown in FIG. In the case of two-phase modulation, the components are as shown in FIG.

Thus, since the current ripple at each switching time can be defined, for example, the current ripple norm of Δi _a (n) is decomposed into biaxial components as shown in equation (20), and further squared as shown in equation (21). If the square root of the sum is taken, a norm component | Δi _a (n) | is obtained. Then, the norm component of the current ripple vector at each switching time in FIG. 16 is calculated, and the maximum component during the carrier half cycle is selected to obtain the “maximum norm of current ripple”.

In the above, the norm component of the current ripple in the three-phase modulation is calculated. However, even in the case of the two-phase modulation, if the output time width is changed as shown in Table 3, Tables 6 to 8 can be applied as they are. In the case of two-phase modulation and shifting to the P-side potential, Δi _a (n) = 0 and Δi _a (m) = 0 with respect to the current change amount of the three-phase modulation, and Δi _C (n) and Δi _C ( Since the component m) is doubled, the result is a current ripple component as shown in FIG. Similarly, when shifting to the N-side potential, Δi _C (n) = 0, Δi _C (m) = 0, and Δi _a (n) and Δi _a (m) components are doubled. The current ripple component is as shown in b). After that, after calculating the norm of each current ripple, the maximum norm is selected.

  The method for calculating the maximum norm of the current ripple at the switching time of each phase has been described above.

[5. Method of handling current ripple component as active current and reactive current]
In the previous section, the current ripple component in the fixed coordinate system was calculated. Eventually, we would like to expand this current ripple component so that it can be handled separately as an active power component and a reactive power component. First, a method for handling current ripple as an orthogonal biaxial component of a rotating coordinate system will be described.

(1) Load model of rotating coordinate system Expressions (14), (15), (16), (17), and (18) are shown as equations of the load model of the fixed coordinate system. When converted into the rotating coordinate system (dq coordinate) defined in FIG. 2, the conversion can be performed as shown in the following equations (22), (23), (24), (25), and (26).

When the differential term is converted into rotational coordinates, it is necessary to replace it with di _αβ / dt → di _dq / dt + J · (ωL · i _dq ), so that the (ωL · i _dq ) term appears in the speed electromotive force component. Therefore, the combined speed electromotive force of this component and the field magnetic flux is represented as [e _d1 (t) e _q1 (t)].

After the differential term is moved to the left side and converted into the form of the state equation, it is approximated that the change in electromotive force during the carrier period and the amount of current ripple are small. Specifically, since the carrier frequency f _C is sufficiently higher than the angular frequency command ω ^* , the rotation amount of the voltage command phase θ _v = (θ (t) + ψ _v (t)) during the carrier period is very small. Assuming that it can be ignored, it is regarded as [ed ₁ e _q1 ] ^T (constant) as shown in equation (23). By applying the approximation of the equation (23) and the approximation that also ignores the voltage drop component of the resistor R, the equation (22) can be simplified to the equation (24).

Note Here, the speed electromotive force of the rotating coordinate system is expressed as [e _d1 e _q1] ^T, simply different since those obtained by converting a rotating coordinate the speed electromotive force _[e α0 e _β0] ^T of the fixed coordinate system is necessary.

As for the short-term current vector change called the PWM carrier cycle, the current vector at the start time (carrier vertex) of the sample period is set as the initial value, and the change component of each PWM state is integrated (25), as in the equation (17). ). Here again, the component excluding the initial value component [i _d0 i _d0 ] ^T is referred to as “current ripple vector” in the rotating coordinate system.

If the integral calculation is approximated to the time product of the voltage components, the change in the current vector due to the output voltage component [v _d v _q ] and the time width ΔT in the PWM pattern of four sections (0, λ, μ, 7). The quantity [Δi _d Δi _q ] is treated as an approximate expression of the expression (26).

  For the generation of the PWM pattern, refer to [3. Since the procedure described in the section “3-phase PWM pattern generation method” can be used, [4. The calculation part of the current change amount in the term of “current vector locus and current ripple component by the PWM pattern” is expressed by the equation (14) to (18) to (22) to (26). If it replaces with a formula, the formula of the rotation coordinate which makes d-axis the reference axis will be obtained.

Now consider the difference between rotating and fixed coordinates. The PWM output voltage conforms to a fixed coordinate system, the speed electromotive force conforms to a rotating coordinate system, and originally includes components of different coordinate systems. When approximation is not possible as much as possible, an error occurs because the coordinates are different in the direction of [e _α0 e _β0 ] ^T on the fixed coordinate system, and the PWM voltage [v _d (t) v _q (t) on the rotating coordinate system. ] An error occurs because the coordinates are different in ^T.

Accordingly, it is considered that the same level of error occurs in the current ripple regardless of which is selected. However, [6. The section of the current ripple component on the case where the carrier frequency can be approximated with sufficiently high ^{_{[v * d v * q]}} T ≒ [e d1 e q1] While applying a bold approximation that ^T, fixed coordinate in this case Since the [e _d1 e _q1 ] ^T component of the rotational coordinates has less error than the [e _α0 e _β0 ] ^T component of the above, the velocity electromotive force and current ripple calculation of the rotating coordinate system is adopted.

(2) Rotational coordinate system based on velocity electromotive force Finally, in order to handle current ripple as a component corresponding to active power and reactive power, the phase of e ₁ = [ed ₁ e _q1 ] ^T vector in FIG. A new rotation coordinate (XY coordinate) is defined.

By converting equation (23) into polar coordinates, a phase angle φ _e1 from the d axis to the e ₁ vector is obtained as in equation (27).

Then, in the same manner as in Table 8, with respect to the current ripple vector component [Δi _d Δi _q ] calculated by the equation (26) on the rotation coordinate, only the phase angle φ _e1 from the dq coordinate as in the equation (28). If rotation coordinates are converted to XY coordinates by rotation, it can be finally converted into components of rotation coordinates (XY coordinates). As shown in FIG. 19 for FIG. 16 and FIG. 20 for FIG. 17, the e ₁ vector is rotated so as to coincide with the X axis [| e ₁ | 0]. It is characterized by the ^T vector. As a result, the X-axis component of the current ripple indicates the effective component and the Y-axis component indicates the ineffective component.

In the above, the method of converting the coordinates after calculating the current ripple in the dq coordinate system has been shown. Instead, the voltage component of each period of the PWM pattern is first converted into the rotation coordinates (XY coordinates). Therefore, it is easier to calculate the current change amount and the ripple component on the XY coordinates. Therefore, in the following, as a practical method, a procedure for calculating after converting this voltage component into a new rotation coordinate (XY coordinate) based on the [| e ₁ | 0] ^T vector will be described.

The phase angle θ _v of the voltage command, the phase angle θ from the α axis to the d axis, the phase angle φ _e1 from the d axis to the e ₁ vector, and the combined phase θ _e1 = θ + φ _e1 . What is necessary is just to calculate a PWM pattern on a fixed coordinate as usual. However, in order to handle each voltage vector component of the PWM pattern on the rotation coordinate, it is easier to handle it on the polar coordinate than the biaxial component called αβ component.

Therefore, instead of Table 4 based on the voltage command V ^* of the phase angle theta _v as shown in Table 9 were classified into 6 sections, and the phase theta _lambda of the output voltage vector V _lambda of T _lambda period as polar phase components T set the phase theta _mu output voltage vector V _mu of _mu period. Here, since the amplitudes of V ₀ and V ₇ are zero, phase information is not used at the time of calculation and is omitted.

By doing so, the voltage vectors V ′ ₀ , V ′ ₇ , V ′ _λ , and V ′ _μ expressed in the XY coordinate system can be calculated by the equation (29), similarly to the voltage components in Table 4. Since the speed electromotive force e ′ _{1 in the} XY coordinate system coincides with the reference axis (X axis), it becomes a scalar component as shown in equation (30). Similarly to Table 7, the amount of change in the current vector and the locus of the current vector expressed in the XY coordinate system can be calculated as Table 10.

If it is desired to consider that the current ripple trajectory at the carrier vertex moves at the angular velocity of the fundamental wave component during the half cycle of the carrier even on this coordinate, the start point i of the carrier half cycle is obtained as in the equation (20). An intermediate point i ′ _nm between “(n) and end point i ′ (m)” is calculated by the equation (30), and this may be corrected as a new reference point.

If the current ripple component is corrected by regarding this intermediate point i ′ _nm as a new reference point, it can be expressed as shown in Table 11.

  From the current ripple viewed from the reference point at each switching time in Table 11, the norm of each component can be calculated by equation (32).

Here, correction coefficients k _X and k _Y are newly added to the equation (32). This is for use in the third embodiment. At this time, it is assumed that k _X = 1 and k _Y = 1, and the explanation is ignored.

Finally, the norm is calculated for each current ripple component in Table 11, and the maximum norm is selected from among them as shown in Equation (33), and is set as | Δi ′ _max (n) |.

  The above is an example of calculating the current ripple variation and locus for three-phase modulation, but in the case of two-phase modulation as well as in Table 3, if the output time of each voltage vector is replaced, Tables 9 to Since 11 can be used as it is, description thereof is omitted.

  As described above, the norm component of the current ripple at the switching time of each phase can be calculated even in the rotating coordinate system based on the speed electromotive force.

  Finally, Table 12 shows voltage components in the XY coordinate system corresponding to Table 5.

[6. Current ripple component when carrier frequency is sufficiently high]
[4. Current vector locus and current ripple component by PWM pattern] and [5. In the section “Method of handling current ripple component as active current and reactive current”], a general form is considered in which a difference occurs between the start point and end point of the current ripple locus of the carrier half cycle. However, if the following two types of approximation are applied, it is not necessary to calculate the intermediate point between the start point and end point of the current ripple, and the calculation can be simplified. The content of this specification is “estimating the current ripple component by PWM and variably controlling the carrier frequency based on it”, and if the current and voltage control accuracy is not affected, the estimated current ripple is somewhat Even if an error is included, there is often no problem in practical use. Although the switching frequency is affected by this error, if the temperature design margin is large, the actual harm is small, and measures such as monitoring the average switching frequency are possible. Therefore, in this section, a method for simplifying the calculation by applying a large approximation with an emphasis on practicality will be described.

(Approximate condition)
(A) It is assumed that the carrier frequency fc is sufficiently high.
(B) Further, as shown in FIG. 21, it is approximated that the voltage command [v ^* _d v ^* _q ] ^T and the speed electromotive force component [e _1d e _1q ] ^T are substantially equal.

  In the case of three-phase modulation, as shown in FIG. 23, it moves repeatedly on the same current ripple locus. The difference between the start point and the end point of this current ripple locus becomes small.

When the above conditions are set, the voltage command [v ^* _d v ^* _q ] and the speed electromotive force [e _1d e _1q ] in the vector diagram of FIG. 13 are approximated to (v ^* ≈e ₁ ) as shown in FIG. Thus, the locus of the current ripple in FIG. 16 changes as shown in FIG. 23, and the start point and end point of the sample period coincide. Further, in the XY coordinates, the current ripple component of FIG. 19 becomes as shown in FIG. 24 by approximating the voltage vector of FIG. 18 as shown in FIG.

In FIG. 23, the current vectors i _n = i _m = i _{(n + 1)} = i _mn = i _{m (n + 1)} at the respective sampling times coincide with each other, and from t _{n to} t _{(n + 1)} The trajectory between carrier periods is a parallelogram in which two triangles are arranged point-symmetrically with respect to the middle point of the side. Therefore, if the approximation to the triangle and the symmetry are used, not only the correction calculation of the reference point becomes unnecessary, but also the calculation amount of the current ripple at each switching time required for estimating the maximum norm component is 3 It is possible to reduce the current ripple calculation from two times to only two current times. This can greatly simplify the calculation.

In the case of the XY coordinates of FIG. 24, the current change amount (Δi ₀ (n), Δi ₇ (n)) component, that is, the parallelogram diagonal line during the period of outputting the zero vector coincides with the X axis. The Y-axis component becomes zero, and operations such as square and square root are not necessary for norm calculation of this component. These are examples of three-phase modulation, but even in the case of two-phase modulation, the current vectors i _n = i _m = i _{(n + 1) in} FIG. 17 are matched, and two triangles as shown in FIG. Becomes a shape arranged symmetrically with respect to the vertex. Similarly, the XY coordinate component also has a shape in which two triangles are arranged symmetrically with respect to the vertex as shown in FIG. Therefore, similarly in FIG. 26, the Y-axis component of the current ripple component in the output period of the zero vector voltage is zero, and operations such as the square and the square root for norm calculation are not required.

  By applying the approximation provided in this section in this way, it is possible to greatly reduce the amount of computation by selecting coordinates that can handle the current ripple due to zero vector voltage as an approximation that matches the start point and the end point. become.

Even if this approximation is applied, there still remains a factor that the shape of the current ripple locus in FIG. 25 changes when the phase of the voltage command in FIG. 21 changes. In order to investigate this, FIG. 32 shows the current vector trajectory set by setting several voltage command phase conditions as shown in FIG. As the voltage command of FIG. 21, the d-axis to 60 deg, i.e., when the voltage command between the phase of V ₆ from V ₄ is present, PWM output voltage vector _{(V 4, V 6, V} 0, V ₇ ), but the output time ratio changes. Therefore, the current locus also changes depending on the phase of the voltage command as shown in FIG.

  The center point (initial value) of the current ripple locus rotates in synchronization with the voltage command, and the shape of the triangle drawn by the current ripple also changes. In order to clearly show how the maximum norm of the current ripple changes, only the current ripple component for the carrier period is extracted from FIG. 32 and translated so that all the reference points coincide with the origin. 33. It can be seen that the vertex group constituted by the maximum norm of the current ripple trajectory group is distributed in an elliptical shape and has a major axis and a minor axis. The present invention focuses on the fact that the maximum norm of the current ripple varies depending on the phase of the voltage command.

  If the vertices of the maximum norm distributed in an elliptical shape are arranged in a circle having a radius of the current ripple component set as shown by the broken line in FIG. 33, the maximum norm of the current ripple can be controlled to be constant, Conversely, consider setting a short axis and a long axis component of an ellipse to obtain a new effect.

In the present invention, attention is paid to the shape change of the vector locus of the current ripple as shown in FIG. FIG. 28 shows current ripple trajectories at a plurality of different voltage command phases θ _{V. When} the current ripple is Δi (θ _{v — b} ) and Δ i (θ _{v — c} ), the radius | Δi ^* | Exists on a circle. However, when the current ripple is Δi (θ _{v — a} ) and Δi (θ _{v — d} ), the maximum norm is smaller than the circle of radius | Δi ^* |. If this is the _Δi (θ v_a) Δi of (theta _{V_d)} it indicates that there is room to expand the current ripple can be thought of as its amount corresponding carrier frequency may be more reduced. Therefore, a method of predicting and calculating current ripple and correcting the carrier frequency so that the maximum norm component becomes constant | Δi ^* | can be considered. In this way, the maximum norm of the current ripple is controlled so as to match the allowable current error circle | Δi ^* |, which is limited to a partial period, but if the carrier frequency is reduced even a little, the average The carrier frequency can be reduced, and the switching loss can be reduced. This is the first idea regarding the present invention.

As a method for estimating the maximum norm of the current ripple during the PWM half cycle, [4. [5. Current vector locus by PWM pattern and current ripple component] and the like handled in a fixed coordinate system, [5. Three types are shown depending on the difference in the coordinate system to be handled, such as those handled in a rotating coordinate system and a special rotating coordinate system (XY coordinate system) such as the method of handling current ripple components as active current and reactive current]. Further, a method of calculating strictly without applying approximation, [6. There is also a method of simplifying the calculation of the current ripple component by making a bold approximation such as [current ripple component when approximation is possible when the carrier frequency is sufficiently high]. In addition, there are several types of control indices described below, and when these are combined, the number of embodiments becomes enormous. Therefore, as described below, by limiting to the combination of three basic methods and three-phase / two-phase modulation: a strict method and a practical method, and an example of extending to control of active / reactive power components The number of embodiments is limited.
Embodiment 1: When exact calculation is applied in a fixed coordinate system (three-phase modulation)
Embodiment 2: When applying exact calculation in a fixed coordinate system (two-phase modulation)
Embodiment 3: Case where simplification is applied by applying approximation in a special rotating coordinate system (XY coordinate, considering active power and reactive power) (three-phase modulation)
Embodiment 4: Case where simplification is applied by applying approximation in a special rotating coordinate system (XY coordinate, considering active power and reactive power) (two-phase modulation)
Embodiment 5: The method of evaluating in Embodiment 4 by further weighting different effective and invalid current ripples.

  FIG. 8 shows a case where the three-phase modulation is applied to the phase voltage command generation unit 3 and FIG. 9 shows a case where the two-phase modulation method is applied to the control configuration example of the conventional method of FIG. This is configured as a discrete system in which samples are updated at the vertex time of the carrier signal, and FIGS. 8 and 9 are conventional examples of the present invention. On the other hand, FIG. 31 is a control block diagram of the embodiment in the case of three-phase modulation, and FIG. 34 in the case of two-phase modulation.

In FIG. 31, the following five elements are added to FIG.
(1) The maximum norm | Δi (θ) | of the current ripple is estimated and calculated from the three-phase voltage command after applying the zero-phase modulation of the three-phase modulation.
(2) The carrier frequency correction coefficient k _fc is calculated from the ratio between the maximum norm | Δi (θ) | of the current ripple and the allowable current ripple | Δi ^* |.
(3) to the basic carrier frequency f ^* c, calculates a carrier frequency command f _c by multiplying the carrier frequency correction factor k _fc.
(4) for generating a triangular wave carrier signal Cry the frequency of the carrier frequency command f _c as expected.
(5) In accordance with the carrier frequency command f _c , the rotation coordinate conversion reference phase θ ₀ for converting the voltage command from the rotation coordinate to the fixed coordinate is calculated.

In FIG. 34, the following five elements are added to FIG.
(1) The maximum norm | Δi (θ) | of the current ripple is estimated and calculated from a three-phase voltage command obtained by applying two-phase modulation.
(2) The carrier frequency correction coefficient k _fc is calculated from the ratio between the maximum norm | Δi (θ) | of the current ripple and the allowable current ripple | Δi ^* |.
(3) to the basic carrier frequency f ^* c, calculates a carrier frequency command f _c by multiplying this carrier frequency correction factor k _fc.
(4) for generating a triangular wave carrier signal Cry the frequency of the carrier frequency command f _c as expected.
(5) In accordance with the carrier frequency command f _c , the rotation coordinate conversion reference phase θ ₀ for converting the voltage command from the rotation coordinate to the fixed coordinate is calculated.

  This is common to Embodiments 1, 3, 5 and Embodiments 2, 4, 5 respectively. The difference between the first embodiment, the second embodiment, the third embodiment, and the fourth embodiment is the content of the maximum norm of current ripples | Δi (θ) | arithmetic unit, and different calculation methods are divided into a plurality of embodiments. Describe.

[Embodiment 1]
The components of the block diagram of FIG. [1. Describe parts that overlap with [Overall configuration] in an easy-to-understand manner.

(A) Command update unit 7
Based on the sample signal S _smpl generated at the apex of the triangular wave carrier signal Cry of the carrier generating unit 2, the voltage command [v _d ^* v _q ^* ] corresponding to the expression (1) in the sample hold circuit S10 and the angle by the sample hold circuit S11 The frequency command ω ^* is latched and held. Further, the carrier frequency command f ^* c is also latched by the sample hold circuit S12.
(B) Reference phase calculation unit 8
The reference phase calculation unit 8 calculates the half period time 1 / (2f ^* c) from the carrier frequency command f ^* c, and discretizes the integral of equation (2) as in equation (34). calculate. Furthermore, since the PWM pattern calculated here is actually output as an actual voltage with a delay of one sample, it is necessary to correct the phase advance for one sample in the voltage command used for the PWM pattern calculation. Therefore, the phase angle component Δθ ^ that is advanced by the time when the next sample signal S _smpl is generated according to the angular frequency command ω ^* is calculated by the equation (34), and this advance is performed with the reference phase θ ₀ that has been calculated previously. By adding the phase angle Z ⁻¹ · Δθ ^ and then setting it as a new reference phase θ ^ of the rotational coordinate conversion unit P ⁻¹ of the voltage command, prediction correction of the phase advance of the voltage command is realized.

(C) Phase voltage command generator 3
With respect to the voltage command [v _d ^* v _q ^* ] ^T , a symmetric three-phase voltage command V ′ _u by the rotational coordinate transformation P ^{−1 in} equation (4) and the two-phase three-phase transformation C ^{−1 in} equation (3). , V ′ _v and V ′ _w . Here, the above-described reference phase θ ^ is used as the reference phase for coordinate transformation.
(D) Zero phase modulation (three phase modulation) section 9a
For the biaxial voltage commands V _d ^* and V _q ^* converted into symmetrical three-phase voltage commands V ′ _u , V ′ _v and V ′ _w , the equations (7), (8) and (9) Applying the zero-phase modulation correction as shown in the equation, the voltage commands V _u , V _v , and V _w are converted. Specifically, the zero-phase modulation calculation unit 9a calculates v _max and v _min in the equation (7), and further calculates the zero-phase modulation correction value V _{ofs — 3ph} from these average values as in the equation (8). Then, this is added to the three-phase voltage commands V ′ _u , V ′ _v , V ′ _w as a three-phase common offset correction as in equation (9), and new voltage commands V _u , V _v , V _w are added. Output.
(E) Current ripple norm (| Δi (θ) |) estimation unit 10
Here, calculation of the PWM pattern and prediction of the current ripple when it is assumed that the reference carrier frequency command f ^* c is applied from the voltage commands V _u , V _v , V _w and the DC link voltage V _dc information after the zero phase modulation. Perform the operation.

First, from the phase θ _v of the voltage command vector and Table 1, the u, v, and w phases, which are three-phase AC components, are associated with the a, b, and c phases indicating the order of voltage. As a result, as shown in Table 2, the phase sequence in which switching occurs in the carrier half cycle and the four types of voltage vectors are determined. In the following, the calculation proceeds using the a, b, and c phases, which are the order of occurrence of this switching.

The PWM pattern can be defined as four types of output voltage vectors and their output times. The time widths of T ₀ , T _λ , T _μ and T ₇ are calculated from Table 3, and the voltage vector V corresponding to the time width is calculated from Table 4. _0, V _λ, V _μ, to select a V _7. Here, three types of calculation methods are shown in Table 3 by the zero-phase modulation method, but the second column is used in the case of three-phase modulation.

If the PWM pattern can be calculated, the switching time (voltage vector change time) is determined as shown in the first column of Table 7, and during the DOWN period of the triangular wave carrier signal Cry, between the start time t (n) and the end time t (m). Three switching times t _a (n), t _b (n), and t _c (n) are t between the start time t (m) and the end time t (n + 1) in the UP period of the triangular wave carrier signal Cry. Three switching times _c (m), t _b (m), and t _a (m) are determined, and the voltage components generated between these times are determined as shown in Table 6. Next, the amount of change in the current vector during the period in which each voltage vector is output is calculated in Table 6, and the initial initial value current in the carrier half cycle is represented by i as shown in the fourth column of Table 7. The vector component of the current ripple at the switching time is calculated by adding _n = 0 and accumulating the current change amount in the order of occurrence. I _n in the carrier half cycle of the DOWN period also _{this, i na, i nb, i} nc, the switching order of the i _m, also, i _m in the carrier half-period of UP _{period, i mc, i mb, i} ba , I _{(n + 1)} switching order.

  The current vector at each switching time corresponds to the apex of the triangular current locus in FIG. Since it is necessary to set a reference point in order to extract a current ripple component from the current vectors at these vertices, offset correction is performed by calculation shown in the second column of Table 8 from information on the start point and end point during the carrier half cycle. The component is calculated and defined as the reference point for current ripple.

Then, by changing the current vector in the fourth column of Table 7 to this reference, the difference vectors Δi _a , Δi _b , and the current vectors of the respective switching times with respect to the new reference point as in the third column of Table 8 the .delta.i _c is defined as the current ripple component of the final switching time. This means that each component in FIG. 14 is converted into a reference value and a vector component therefrom in units of carrier half cycles as shown in FIG. Since this current ripple Δi _a, Δi _b, Δi _c is a vector quantity, after decomposed into (20) orthogonal axes component _Δi α _(n) _α _{as, Δi β (n) _β (} 21) The amplitude component (hereinafter referred to as the norm) | Δi _a (n) |

Three switching times occur during the half cycle of the triangular wave carrier signal Cry. Current ripple components Δi _a (n) and Δi _c (n) and Δi _a (m) and Δi _c (m) in FIG. Since the vectors have the same amplitude, the current ripple components Δi _a (n), Δi _b (n) or Δi _b (m) and Δi _a (m) are actually used if this symmetry (equal norm) condition is used. It is sufficient to calculate only two of these, and either contains the maximum norm component. Therefore, as an actual calculation, a large amplitude component of the two types is output as an estimated value | Δi | = max (| Δi _a |, | Δi _b |) of the maximum norm of the current ripple in this voltage command.
(F) Carrier correction coefficient calculation unit 11
Since the maximum norm estimated value | Δi | of the current ripple during the carrier half cycle when the reference carrier frequency command f ^* c is assumed, the carrier correction coefficient calculation of FIG. 31 is calculated from this and the allowable current ripple | Δi ^* |. The unit 11 calculates a carrier frequency correction coefficient k _fc . This is because the estimated current ripple norm | Δi (θ) |, which is the output of the “current ripple norm | Δi (θ) | estimator 10”, coincides with the current ripple amplitude command | Δi ^* |. The carrier frequency fc is corrected and is calculated by the equation (35).

And (36) as in equation a correction coefficient k _fc and the reference carrier frequency f ^* c of the carrier frequency by multiplying, obtain the set value f _c of the finally output to the PWM carrier frequency.

(G) Carrier frequency correction unit 12
Generating a triangular carrier signal Cry variable frequency corresponding to the carrier generating portion 2 carriers corrected carrier frequency command f _c. For example, since the amplitude of the triangular wave is related to the DC voltage, it can be fixed and the carrier frequency can be changed equivalently by changing the slope of the triangular wave.

Since the voltage command Vu, Vv, Vw is updated synchronously with the vertex of the triangular wave carrier signal Cry by the sample hold circuit S20 from the timing signal S _smpl at the vertex of the triangular wave carrier signal Cry, the sample hold circuit S22 similarly carrier frequency command f _c to be input to the carrier generating portion 2 also be updated simultaneously. In this way, since the slope of the triangular wave carrier signal Cry is fixed at a constant period during which the voltage command is held constant, accurate PWM modulation is possible. Here, since the carrier frequency command f _c is not yet determined only at the first start-up, measures such as substituting the reference carrier frequency command f ^* c as an initial value are applied.

In the calculation of the PWM pattern, a predicted reference phase θ ^ to which phase advance correction based on the reference carrier frequency command f ^* c is applied is used. However, changing the carrier frequency command f _c, a half cycle of the triangular wave carrier signal Cry, that is, to change the time interval until the next timing signal S _smpl, using again the carrier frequency command f _c (37) formula The final phase is calculated by calculating the final next reference phase θ, and this is held as θ ₀ by the sample hold circuit S21 until the next calculation time until the next calculation time. In this way, even if the carrier frequency command fc is changed, the phase lead angle of the voltage command, that is, the fundamental frequency can be output accurately.

  The above is the configuration example in the first embodiment.

[Embodiment 2]
Although the first embodiment is a case of three-phase modulation, the second embodiment will explain the case of the method using the two-phase modulation of FIG.

The configuration is almost the same as the components in the block diagram of FIG. 31, and there is a difference only in the zero-phase modulation method, the calculation of the PWM pattern, and the norm | Δi | Therefore, the parts that are the same as those in the first embodiment are omitted, and only items that are different are described.
(A) Command update unit 7
(B) Reference phase calculation unit 8
(C) Phase voltage command generator 3
The above three items have the same configuration as in the first embodiment. There is a difference in the next term, but it is difficult to understand because only the corrected part is broken.

(D) Zero phase modulation (two phase modulation) 9b
In contrast to the voltage command converted into a symmetric three-phase AC component, in the two-phase modulation, a new voltage command V is applied by applying zero-phase modulation correction as shown in equations (10) to (11) and (12). _u, V _v, converted to V _w.

Specifically, the P-side potential modulation amount calculation unit 13 calculates the correction value V _{ofs — 2phP} of (1) in the equation (10), and the N-side potential modulation amount calculation unit 14 calculates (2) in the equation (10). The correction value V _{ofs_2phN} is calculated. Then, the two kinds of components are compared, and the smaller absolute value is selected by the switch Sect based on the equation (11). Alternatively, the same operation is performed even if the phase angle θ _{v of the} voltage is classified into the modes shown in Table 9 and the Sect is controlled. Here, a method of using the phase of the voltage command is depicted. Then, the selected zero-phase modulation correction amount is added to the three-phase voltage commands V ′ _u , V ′ _v , and V ′ _w as a three-phase common offset correction as shown in the equation (12) to obtain a new voltage command V _u. , V _v , V _w are output.

(E) Current ripple norm (| Δi (θ) |) estimation unit 10
Here, assuming that the reference carrier frequency command f ^* c is applied, the PWM pattern calculation between the carrier half periods is calculated from the three-phase voltage commands V _u , V _v , V _w after zero-phase modulation and DC link voltage V _dc information. And predicting current ripple.

First, from the phase θ _v of the voltage command vector and Table 1, the u, v, and w phases, which are three-phase AC components, are associated with the a, b, and c phases indicating the order of voltage. As a result, as shown in Table 2, the phase order in which switching occurs in the carrier half cycle and four types of voltage vectors are determined. In the following, the calculation proceeds using a, b, and c phases indicating the phase order in which this switching occurs.

The PWM pattern can be defined as four types of output voltage vectors and their output times. The time widths of T ₀ , T _λ , T _μ and T ₇ are calculated from Table 3, and the voltage vector V corresponding to the time width is calculated from Table 4. _0, V _λ, V _μ, to select a V _7. Table 3 shows three types of calculation methods according to the zero-phase modulation method. In the case of two-phase modulation, the third column (zero-phase modulation in contact with the P side) in the DOWN period of the triangular wave carrier signal Cry. ) In the UP period, the fourth column (zero-phase modulation in contact with the N side) is selected. In addition, the time element includes an item including zero, which is set to be handled in a unified manner with three-phase modulation. For example, when T ₀ = 0, a V ₀ vector is generated. Without moving to the next voltage vector.

If the PWM pattern can be calculated, the switching time (voltage vector change time) is determined as shown in the first column of Table 7, and during the DOWN period of the triangular wave carrier signal Cry, between the start time t (n) and the end time t (m). Three switching times t _a (n), t _b (n), and t _c (n) are t between the start time t (m) and the end time t (n + 1) in the UP period of the triangular wave carrier signal Cry. Three switching times, _c (m), t _b (m), and t _a (m), are determined.

Further, the voltage components generated during this time are determined as shown in Table 6. Next, the amount of change in the current vector during the period in which each voltage vector is output is calculated in Table 6, and the initial initial value current in the carrier half cycle is represented by i as shown in the fourth column of Table 7. The vector component of the current ripple at the switching time is calculated by adding _n = 0 and accumulating the current change amount in the order of occurrence. I _n in the carrier half cycle of the DOWN period also _{this, i na, i nb, i} nc, the switching order of the i _m, also, i _m in the carrier half-period of UP _{period, i mc, i mb, i} ba , I _{(n + 1)} switching order.

The current vector at each switching time corresponds to the apex of the triangular current locus in FIG. Since it is necessary to set a reference point in order to extract a current ripple component from the current vectors at these vertices, the offset correction components i _nm , im _{(n + 1)} is calculated and defined as a reference point for current ripple. Then, as shown in the third column of Table 8, the difference vectors Δi _a , Δi _b , Δi _c between the reference point and the current vector at each switching time are defined as current ripple components at the switching time.

This means that each component in FIG. 15 is converted into a reference value and a vector component therefrom in units of carrier half cycles as shown in FIG. This current ripple component .DELTA.i _a, .DELTA.i _b, because .DELTA.i _c is a vector quantity (called later norm) amplitude component by (20) was decomposed into orthogonal biaxial components from (21) as equation Can be calculated.

In the case of two-phase modulation, there is a period with a time width of zero as shown in Table 3, and since the current ripple components at the start time and end time are small in the carrier half cycle as shown in FIG. In fact, Δi _a (n), Δi _b (n) in the case of N-side modulation, or Δi _b (m), Δi _a (in the case of P-side modulation. It is only necessary to calculate two of m), and either contains the maximum norm component. Therefore, a large amplitude component of these two types of interest is output as an estimated value | Δi | = max (| Δi _a |, | Δi _b |) of the maximum norm of the current ripple in this voltage command.

  Although the calculation formula is different from that of the first embodiment, the calculation amount is almost the same.

(F) Carrier correction coefficient calculation unit 11
(G) Carrier Frequency Correction Unit Since these two terms have the same configuration as that of the first embodiment, the description thereof is omitted. The above is the configuration example of the second embodiment.

[Embodiment 3]
In the first embodiment, for PWM to which three-phase modulation is applied, a current vector change is predicted using an arithmetic expression of a fixed coordinate system for calculating a current ripple norm | Δi | The accuracy of the norm | Δi | of the current ripple was ensured. On the other hand, in the third and fourth embodiments, a method that focuses on simplifying complicated calculation by applying approximation as much as possible will be described.

In order to realize the third embodiment, the main points changed from the first embodiment are the following three points.
(Condition a) Handling in the rotating coordinate system (Condition b) Applying the approximation of voltage command [v _d ^* v _q ^* ] ^T ≈ [ed ₁ e _q1 ] ^T (Condition c) Further, the reference phase angle of the rotating coordinate Speed electromotive force [e _d1 e _q1 ] Select in the vector direction of ^T.

The same control block diagram as that of FIG. 31 is used, and only the contents of the “current ripple norm estimation unit 10” are different. Therefore, the “(e) current ripple norm (| Δi (θ) | Only the changes in the portion corresponding to the estimation unit 10) will be described.
(E) Calculation of Norm (| Δi |) of Current Ripple in Embodiment 3 The calculation of the PWM pattern is the same as in Embodiment 1, and the reference phase used for rotational coordinate conversion assuming the reference carrier frequency command f ^* c The voltage command V _u , V _v , V _w converted to three-phase AC is converted by θ ^, and the PWM pattern and current change due thereto are predicted from information such as this and the DC link voltage V _dc .

From the phase θ _v of the voltage command vector and Table 1, the U, V, and W phases, which are three-phase AC components, are associated with the a, b, and c phases indicating the order of voltage. Since the PWM pattern is set as an output voltage vector and a time interval for outputting it, the time widths of T ₀ , T _λ , T _μ and T ₇ are shown in Table 3, and the voltage vector V corresponding to the time width is shown in Table 4. _0, V _λ, V _μ, to select a V _7. These V ₀ , V _λ , V _μ , and V ₇ actually correspond to those selected from the eight types of voltage vectors shown in Table 5. Here, three types of calculation methods are shown in Table 3 by the zero-phase modulation method, but the second column is used in the case of three-phase modulation. Up to this point, the configuration is the same as that of the first embodiment.

  Since the PWM pattern is calculated, the locus of the current vector is calculated next. Here, in the third embodiment, the above three settings ((a) to (c)) and approximation are applied.

First, an approximation of “substituting the voltage command [v _d ^* v _q ^* ] ^T by applying (condition b) instead of calculating the speed electromotive force [e _d1 e _q1 ] ^T of the equation (23)”. Apply. That is, the phase of the speed electromotive force φ _e1 shown in the equation (27) is assumed to be equal to the phase of | v ^* | ∠φ _{v *} obtained by polar-transforming the biaxial voltage command [V _d ^* V _q ^* ] ^T. Substitute with angle φ _{v *} . Then, a reference phase (X-axis) is used as a reference axis (X-axis) with the reference phase of the conventional rotational coordinates, that is, the phase angle θ ^ from the U-phase to the d-axis and the combined phase θ _e = θ + φ _{v *} of this phase angle φ _{v *} The current ripple component is calculated using the coordinates (XY coordinates). It is a feature of the third and fourth embodiments that a special rotation coordinate axis is used.

In order to convert the voltage vectors V _λ and V _μ on the XY coordinates, the conversion may be performed as shown in equation (29). Here, the values in Table 9 are used for the phases θ _λ and θ _μ . This corresponds to one selected from the components in Table 12 obtained by rotating coordinate transformation of eight types of voltage vectors in the αβ coordinate system in Table 5. That is, in order to convert the two types of voltage vectors V _λ and V _μ on the XY coordinates, instead of selecting from Table 5, the corresponding voltage vectors V ′ _λ and V ′ _μ are selected from Table 12. Therefore, this is re-expressed as the voltage component equation (29) and the phase in Table 9. Since V ₀ and V ₇ have zero amplitude components, they are still zero vectors even after rotational coordinate conversion, and are also replaced by zero vectors V ′ ₀ and V ′ ₇ .

When the voltage command [v _d ^* v _q ^* ] ^T is converted into XY coordinates based on its own phase as shown in FIG. 22, this vector is only the X-axis component as [| v ^* | 0] ^T. The Y-axis component becomes zero and a simple scalar quantity is obtained. As a result, a zero term appears, so that the subsequent calculation can be simplified.

(Supplement) In this way, when (Condition a, Condition b, Condition c) is applied, the voltage command becomes only the X-axis component, and as a result, the current change in the output period of V ₀ and V ₇ also occurs only in the X-axis direction. No longer. Then, in addition, in the current norm calculation, one of the vector components is zero, but it can be handled as a scalar quantity. By applying this approximation, the vector component can be replaced with the scalar quantity as much as possible, and as a result, an effect of reducing the calculation can be expected.

  If each voltage vector component of XY coordinates and the output time width are obtained, the current ripple component can be calculated in the same procedure as in the first embodiment.

  Here, since (Condition b) is applied, the starting point and the ending point of the current vector locus during the carrier half cycle coincide with each other as shown in FIG. Therefore, in the first embodiment, offset correction is applied as shown in Table 8, but this is unnecessary and can be further simplified.

Therefore, the current ripple component can use the voltage in the second column of Table 10 to calculate the amount of current change in the period in which the voltage vector is output in the third column, and further integrate it as in the fourth column. For example, the current vectors i ′ _na , i ′ _nb , i ′ _nc or i ′ _mc , i ′ _mb , i ′ _{ma at} each time can be calculated. Since offset correction is not required, the correction components in the second column of Table 11 can be set as i ′ _nm = 0 and i ′ _{m (n + 1)} = 0, so that the current ripple component can use the values in Table 10 as they are. , I ′ _a (n) = i ′ _na , i ′ _b (n) = i ′ _nb , i ′ _c (n) = i ′ _nc and i ′ _c (m) = i ′ _mc , i ′ _b (m ) = I ′ _mb , i ′ _a (m) = i ′ _ma, and calculation of the reference point as shown in Table 11 and correction operation using the reference point are not necessary.

Furthermore, since the voltage difference (V ′ ₇ −e ₁ ) includes only the X-axis component, the i ′ _c (n) and i ′ _c (m) components also generate only the X-axis component, and therefore can be handled as a scalar quantity. The i ′ _a (n) and i ′ _a (m) components can also be handled as scalar quantities. In this way, if the amount is a scalar, the square or square root calculation for obtaining the norm is not required, and the amount of calculation can be further reduced. That is, the scalar quantity i ′ _a (n) or i ′ _a (m) is regarded as a norm as it is, and only the vector quantity i ′ _b (n) or i ′ _b (m) is the norm after vector calculation. Calculate the components.

Although the voltage command ^{_{[v * d v * q]}} T is an instruction before the zero-phase modulation, the load of the inverter, a three-phase motor neutral point of the armature winding is insulated (i.e. the Zero-phase current does not flow when neutral point grounding, resistance grounding or capacitor grounding is not performed. Therefore, there is no problem in calculating the norm (| Δi |) of the current ripple by the above method using the voltage command after the zero phase modulation.

  Thereafter, the same configuration as that of the first embodiment is adopted. In principle, the same function as that of the first embodiment is to be realized, but the feature of the third embodiment is that the amount of calculation such as a current vector and a norm can be reduced.

Here, k _x = 1 and k _y = 1 are still set in equation (32) when calculating the norm component of the current ripple. These coefficients are used in the fifth embodiment.

[Embodiment 4]
As the first embodiment approximates the third embodiment, the second embodiment can apply the approximation in the same manner.

The main points changed to realize the fourth embodiment are the following three points as in the third embodiment.
(Condition a) Handled in the rotating coordinate system (Condition b) Approximation of voltage command [v _d ^* v _q ^* ] ^T ≈ [ed ₁ e _q1 ] ^T is applied.
(Conditions c) further selects a reference phase rotating coordinate vector direction of the velocity electromotive force [e _d1 e _q1] ^T.

  The control block diagram is the same as that shown in FIG. 34, and only the contents of the “current ripple norm estimation unit 10” are different. Therefore, here, the “(e) current ripple norm (| Δi (θ) | Only the changes in the portion corresponding to the estimation unit 10) will be described.

(E) Calculation of Norm (| Δi |) of Current Ripple in Embodiment 4 The calculation of the PWM pattern is the same as in Embodiment 1. Assuming a reference carrier frequency command f ^* c, it is converted into a three-phase voltage command V ′ _u , V ′ _v , V ′ _w converted into a three-phase alternating current by a reference phase angle θ ^ used for rotational coordinate conversion. Thus, a PWM pattern and a current change caused by the PWM pattern are predicted from information such as the DC link voltage v _dc .

Associating a phase theta _v and Table 4 of the voltage command vector, U is a three-phase alternating current component, V, a indicating the magnitude order of the W-phase voltage, b, and c-phase. Since the PWM pattern can be expressed as an output voltage vector and a time interval for outputting it, the time widths of T ₀ , T _λ , T _μ , and T ₇ are shown in Table 3, and the voltage vector V ₀ corresponding to the time width is shown in Table 4. , V _λ , V _μ , V ₇ are selected. These V ₀ , V _λ , V _μ , and V ₇ actually correspond to those selected from the eight types of voltage vectors shown in Table 5.

Here, there are three types of calculation methods in Table 3, but in the case of two-phase modulation, the triangular wave carrier signal Cry is in the second column (zero-phase modulation in contact with the P side) in the DOWN period, and in the UP period. Three rows (zero phase modulation in contact with the N side) are used. In addition, there is a period including zero in the time element, which is set to be handled in a unified manner with the three-phase modulation. For example, when T ₀ = 0, no V ₀ vector is generated. Immediately shifts to the next voltage vector.

  Since the PWM pattern can be calculated, the locus of the current vector is calculated next. Here, the above three settings and approximations are applied as in the third embodiment.

First, an approximation of “substituting the voltage command [V _d ^* V _q ^* ] ^T by applying (condition b) instead of calculating the speed electromotive force [e _d1 e _q1 ] ^T of equation (23)”. Apply. That is, the voltage command [V _d ^* V _q ^* ] ^T of the phase angle φ _e1 of the speed electromotive force expressed by the equation (27) is assumed to be equal to the phase of | V ^* | ∠φ _{v *} obtained by polar coordinate conversion. Use φ _{v *} instead. Then, the reference phase of the conventional rotational coordinate, that is, the phase angle θ from the U phase to the d axis and the combined phase θ _e = θ + φ _{v *} of the phase angle φ _{v *} are used as a reference axis (X axis). The feature is that the current ripple component is calculated using the axis coordinates (XY coordinates).

In order to convert the voltage vectors V _λ and V _μ on the XY coordinates, the conversion may be performed as shown in equation (29). Here, the values in Table 9 are used for the phase angles θ _λ and θ _μ . This corresponds to one selected from the components in Table 12 obtained by rotating coordinate transformation of eight types of voltage vectors in the αβ coordinate system in Table 5. That is, in order to convert the two types of voltage vectors V _λ and V _μ on the XY coordinates, instead of selecting from Table 5, the corresponding voltage vectors V ′ _λ and V ′ _μ are selected from Table 12. Therefore, this is re-expressed as the voltage component equation (29) and the phase in Table 9. Since V ₀ and V ₇ have zero amplitude components and are still zero vectors even after rotational coordinate conversion, they are also replaced by zero vectors V ′ ₀ and V ′ ₇ .

When the voltage command [V ^* _dV ^* _q ] ^T is converted to the XY coordinates as shown in FIG. 22, this vector becomes the reference axis, so only the X-axis component is obtained as [| V ^* | 0] ^T. The Y-axis component becomes zero and a simple scalar quantity is obtained.

(Supplement) As described above, when (conditions a, b, and c) are applied, the voltage command becomes only the X-axis component, and as a result, the current change during the output period of V ₀ and V ₇ also occurs only in the X-axis direction. . Then, in addition, in the current norm calculation, one of the vector components is zero, but it can be handled as a scalar quantity. By applying this approximation, the vector component can be replaced with the scalar quantity as much as possible, and as a result, an effect of reducing the calculation can be expected.

  If each voltage vector component of XY coordinates and the output time width are obtained, the current ripple component can be calculated in the same procedure as in the second embodiment.

  Here, since (Condition b) is applied, the starting point and the ending point of the current vector locus during the carrier half cycle coincide with each other as shown in FIG. Therefore, in the second embodiment, offset correction is applied as shown in Table 8, but this is unnecessary and can be further simplified.

Therefore, the current ripple component uses the voltage in the second column of Table 10, calculates the amount of current change during the period in which the voltage vector is output in the third column, and further integrates it as in the fourth column. For example, the current vector i ′ _na , i ′ _nb , i ′ _nc or i ′ _mc , i ′ _mb , i ′ _{ma at} each time can be calculated, and offset correction is unnecessary, so the component in the second column of Table 11 is i ′ _nm = 0 and i ′ _{m (n + 1)} = 0, the current ripple component can be used as it is, and i ′ _a (n) = i ′ _na , i ′ _b (n) = i ′ _nb , i ′ _c (n) = i ′ _nc and i ′ _c (m) = i ′ _mc , I ′ _b (m) = i ′ _mb , i ′ _a (m) = i ′ _ma The calculation of the reference point as shown in Table 11 and the correction operation using the reference point are not required.

Further, when shifting to the P side in the two-phase modulation of Table 3, since T ₀ (n) = T ₀ (m) = 0, I ′ _a (n) = 0 and i ′ _a (m) = The norm calculation becomes unnecessary. Similarly, in the case of shifting to the N side in the two-phase modulation of Table 3, since T ₇ (n) = T ₇ (m) = 0, i ′ _c (n) = 0 and i ′ _c (m) = 0, and this norm calculation is also unnecessary.

Further, when the P-side shift is performed in the two-phase modulation shown in Table 3, the voltage difference (V ₇ ′ −e ₁ ) includes only the X-axis component, so that i ′ _c (n) and i ′ _c (m) components Can also be handled as a scalar quantity, containing only the X-axis component. Similarly, the voltage difference in the case of N-side shift at the two-phase modulation in Table 3 'order (-e ₁ includes only the X-axis _{component, i a V 0)' (} n) and i _a '(m) The component includes only the X-axis component and can be handled as a scalar quantity. Therefore, one of the axis components of the vector is zero, and the square and square root operations for obtaining the norm are not required. Thus, the calculation amount can be further reduced.

Summarizing the above, as | Δi | in the fourth embodiment, i ′ _nc and i ′ _mc that are scalar quantities are used when shifting to the P side, and i ′ _na and i ′ _ma are used when shifting to the N side. Assuming the norm as it is, only the remaining vector quantity i ′ _b (n) or i ′ _b (m) is vector-calculated and then the norm component is calculated.

After that, the same configuration as that of the second embodiment is adopted. In principle, the same function as in the second embodiment is to be realized, but the feature of the fourth embodiment is that the amount of calculation such as a current vector and norm can be reduced. Here, in equation (32) when calculating the norm component of the current ripple, k _x = 1 and k _y = 1 are still set. These coefficients are used in the fifth embodiment.

[Embodiment 5]
The fifth embodiment is an improvement over the third and fourth embodiments, and a new function can be generated by changing the current ripple evaluation method.

  In the third and fourth embodiments, the direction of the speed electromotive force vector is defined on the X axis, and the Y axis is defined on the orthogonal direction. Thereby, among the current ripple components, the X-axis component is a current corresponding to active power, and the Y-axis component is a current corresponding to reactive power. In the case of a rotating machine or the like, the pulsation of active power corresponds to torque ripple, and thus becomes a disturbance component to the mechanical system. Therefore, it is desired to reduce the average carrier frequency by controlling the current ripple to be constant as shown in the first to fourth embodiments, but it is also desired to suppress the current ripple for the effective amount at the same time. Therefore, instead of allowing the current ripple of the ineffective portion to be slightly increased from the limit setting value of the current ripple, it is considered that both suppression of torque ripple and suppression of increase in average switching frequency can be achieved. Embodiment 5 describes a method for realizing this.

The method to be applied is that when calculating the norm from the vector component of the current ripple according to the equation (32), if it is desired to suppress the current ripple corresponding to the active power, the X axis is expressed as k _x > 1 and k _y <1. This is a method of weighting the component and the Y-axis component.

  In this way, if the X-axis component of the current ripple is relatively larger than the Y-axis component, the maximum norm is overestimated by the weighting factor, so the carrier frequency is increased to increase the X-axis ripple component. Operates to suppress. On the contrary, when the Y-axis component is larger, the maximum norm is underestimated this time, so the carrier frequency is lowered and the Y-axis current ripple is increased. In this way, by setting different weighting factors for the X axis and Y axis, it is desirable to suppress current ripple when the effective axis component is the main component, so the carrier frequency is increased to suppress current ripple, and conversely the invalid axis When the component is the main component, an increase in current ripple is allowed and, instead, the switching frequency is reduced to make the average carrier frequency as low as possible. Thus, by making the variable control method of the carrier frequency different depending on the axis component of the current ripple, the current ripple corresponding to the active power is suppressed, and conversely, the reactive current ripple is increased. It is possible to realize control that can satisfy the conflicting requirement of suppressing the increase in average switching frequency as much as possible.

  In the fifth embodiment, the correction coefficient is set on the current ripple estimation side of the equation (32), but instead of applying a weight here, the allowable value of the current ripple is set in an elliptical shape and the ripples for the effective and ineffective portions are set. A method of individually controlling the components is also conceivable. Even if the command side is operated or the estimation side compared with the command is operated as in the fifth embodiment, a relatively substantially equivalent effect can be obtained. However, in order to correct the command side to be elliptical, there is a problem that the amount of calculation increases, such as requiring current ripple phase information, and therefore, the fifth embodiment employs a method of correcting the estimation side. ing.

  In order to show the effects of the first to fifth embodiments, an example in which a numerical simulation is performed will be described. Here, the following three types of numerical simulation results are shown as an explanation of the effect.

  Here, Embodiments 1 and 2 are methods for performing strict calculations in a fixed coordinate system, and Embodiments 3 and 4 further simplify the calculation by applying approximation. Therefore, Embodiment 1 and Embodiment 3 and Embodiment 2 and Embodiment 4 each show substantially the same function. Therefore, here, the effect is limited to the third and fourth embodiments. Further, although Embodiment 5 can be combined with both Embodiment 3 and Embodiment 4, here, as a representative example, an effect when Embodiment 4 and Embodiment 5 are combined is shown.

  Tables 13 and 14 list the data shown below. Table 13 is a list of comparison data for confirming the effects of the third embodiment in the case of three-phase modulation. Table 14 is a list of comparison data for confirming the effects of applying Embodiment 4 and the effects of combining Embodiment 4 and Embodiment 5 in the case of two-phase modulation.

  First, FIG. 35 and FIG. 43 show an operation example when the fifth embodiment is not applied for comparison. FIG. 35 shows a case where the frequency of the triangular carrier signal Cry is fixed at 3 kHz and three-phase modulation is applied as the zero-phase modulation method. FIG. 43 shows a case where the frequency of the triangular carrier signal Cry is fixed to 4 kHz and two-phase modulation is applied as the zero-phase modulation method. Since the average switching frequency is lower in the two-phase modulation, the carrier frequency is set to 4/3 times.

In this simulation, the following ideal conditions are set.
A) It is an ideal switch, and there is no dead time setting, no delay time, and no element voltage drop. B) The load condition (values of e ₀ , R, and L) is about 1.0 p. The vicinity of u and the power factor were set to be about 1.0. This is set because it is easier to understand when the voltage and current have the same phase, and there is no problem in operation even under other current conditions. C) The fundamental carrier frequency f ^* c = 3 kHz (three-phase modulation), 4 kHz (two-phase modulation), and the fundamental wave frequency = 50 Hz.

FIG. 35 and FIG. 43 show the following waveforms in order from the top.
(A) Three-phase voltage commands V _u0 , V _v0 , V _w0 and triangular wave carrier signal Cry
These are three-phase AC voltage commands V _u0 , V _v0 , V _w0 (after zero-phase modulation / sample hold) and a triangular wave carrier signal Cry that are input to the PWM comparator 4.

One cycle (20 ms) of voltage commands V _u0 , V _v0 , V _w0 (50 Hz), the triangular wave carrier signal Cry is 60 kHz (4 kHz) in a 20 ms period in the case of 3 kHz, and a 20 ms period in the case of 4 kHz. 80 periods (4 kHz) are drawn. The voltage component is shown by the unit method.

  Since zero-phase modulation is applied to three-phase modulation, the amplitude of the maximum value and the minimum value of the phase voltage are always equal. In the case of two-phase modulation, two types of zero-phase modulation modes of P side or N side are switched every 60 degrees in electrical angle.

(B) Three-phase PWM pattern PWM_u, PWM_v, PWM_w
The three-phase component (digital value) of the PWM control signal (H / L) that is the output of the PWM comparator 4, the upper level is “H (upper arm ON)”, and the lower level is “L (lower arm on)” ".

(C) Current waveform (rotating coordinate system, XY axis component) ix, iy
The current waveform of the three-phase AC in the lower stage (d) is shown in [5. The three-phase alternating current is converted into the orthogonal two-axis component in the rotation coordinate form represented by the XY coordinates of the voltage reference described in the case of treating the current ripple component as an active current and a reactive current]. The DC component is a fundamental wave component, and the triangular ripple is a current ripple component generated by PWM. It can be seen that the PWM ripple width pulsates at a current ripple width of 6 times the fundamental wave period. The vertical axis represents the unit method.

(D) Current waveform (three-phase AC waveform i _u , i _v / synchronous current sample waveform i _{u_smpl} , i _{v_smpl} , i _{w_smpl)}
It is an output current waveform of the inverter and shows a three-phase current component of a three-phase coordinate system. Two types of three-phase components are drawn, and the upper side shows a three-phase alternating current waveform, and the lower side shows a sampled and held current waveform at the top of the carrier as equivalent to a current detection signal. Here, since it is difficult to see if the waveforms overlap, the detection signal is -0.1 p. u. The offset is superimposed and shifted downward.

  Thus, since a sinusoidal current flows, a PWM modulation output equivalent to a sinusoidal voltage command is obtained, and a current detection method called “carrier synchronization current sampling” (sample timing control). ), It can be confirmed that the fundamental wave component from which the PWM current ripple is substantially removed can be detected. Again, the vertical axis is the unit method display.

(E) Integrated value of the number of occurrences of three-phase switching The change of each phase (H → L or L → H conversion) in the second stage (b) is 0.5 times each (L → H → L) As a single switching, the number of occurrences is counted for each phase. In three-phase modulation, since the carrier frequency fc = 3 kHz, switching occurs 60 times in 20 ms. In the two-phase modulation, the carrier frequency fc = 4 kHz, but the switching frequency is reduced to about 2/3 (about 2.7 kHz) since the period of about 1/3 is not switched, and about 53 for 20 ms. Times of switching occurs.

  FIGS. 37 and 40 show the locus of the current ripple component as orthogonal biaxial components of fixed coordinates (αβ coordinates). FIG. 37 shows the vicinity of the first electrical angle 60 deg of FIG. 35, and FIG. Only the vicinity of the first electrical angle of 43 deg. When one cycle is drawn, many trajectories are overlapped and cannot be identified. Therefore, the drawing is limited to about 1/6 cycle. This corresponds to one obtained by removing the fundamental wave component from the three-phase alternating current shown in FIG. 4D, or one obtained by removing the direct current component shown in FIG.

  From FIG. 37 and FIG. 40, it can be seen that the vector of the maximum norm point of the current ripple component by PWM is distributed elliptically. And in this invention, paying attention to distribution of this maximum norm point, these are controlled.

  The above is the operation example of the conventional example for comparison with the embodiment of the present invention.

(Embodiment 1 and Embodiment 3)
The first embodiment and the third embodiment are cases where three-phase modulation is applied, and the same effect can be obtained in principle although there is a difference in approximation method. Therefore, after the data of the third embodiment is shown as a representative. The effect will be described.

FIG. 36 is a time chart when the third embodiment is applied. Here, the reference carrier frequency is 3 kHz, but the carrier frequency of the output PWM changes according to the amplitude command of the current ripple. Therefore, the amplitude setting of the current ripple is set to 0.05 p. u. Set to. Further, since it is not used here, k _x = k _y = 1 is set.

FIG. 38 shows the current ripple components Δi _α and Δi _β in FIG. 36 as trajectories in XY coordinates.

  36 and 38, the effect of the third embodiment can be read.

(1) Change in current ripple width Focusing on the width of the ripple component of the XY-axis current in FIG. 36 (c), the ripple width of i _X pulsates in a period of 6 times in FIG. The ripple width of i _Y is substantially constant and the amplitude remains small. However, when the third embodiment is applied, the amplitude of the current ripple of i _X becomes small, and a 6-fold periodic component appears in the current ripple of i _Y. However, the current ripples of i _X and i _Y are not increased at the same time.

Also, looking at the waveform of the triangular wave carrier signal Cry in FIG. 36A, it can be seen that the period has changed in FIG. In particular, it can be seen that when the ripple component of i _Y is large, the cycle of the triangular wave carrier signal Cry becomes long, and the carrier frequency fc is reduced during this period. Conversely, when the ripple component of i _X is decreasing, the carrier frequency fc is slightly increased, but the current ripple component is suppressed. When this is compared as the XY axis component of the current ripple, the distribution of the maximum norm of the current ripple is elliptical in FIG. 37, but in FIG. 38, the major axis direction becomes smaller and the minor axis direction becomes larger and the circular shape becomes larger. The distribution has changed. That is, the maximum value of the maximum norm can be suppressed by suppressing the pulsation of the amplitude component of the maximum norm of the current ripple.

(2) Output voltage and current detection characteristics The change of the carrier frequency fc is updated on a carrier half cycle basis so that an accurate PWM voltage can be output. This confirmation is the same in the waveform of FIG. 36 (d) in which the current waveform has no distortion and is the same fundamental wave component, and the current detection value is sampled / held at the sample timing assuming current detection. A waveform equivalent to the actual current can be detected without distortion.

(3) Switching frequency reduction effect due to modulation factor Embodiment 3 has a large improvement effect when the voltage command is reduced, that is, the modulation factor is low. FIG. 39 shows the modulation rate on the horizontal axis and the average carrier frequency (switching frequency) when the current ripple is equal on the vertical axis. When the modulation rate is small, the current ripple due to PWM tends to be small, but when the third embodiment is applied, the carrier frequency can be reduced. Thereby, when the modulation rate is low, the effect of reducing the switching loss is high.

  In summary, the conventional method of changing the carrier frequency was realized by preset table information, switching operation, etc. In this embodiment, the carrier frequency is automatically set only by operating the current ripple setting value. Can follow and change. In addition, influences such as fluctuations in DC voltage can also be taken into consideration.

  In addition, by sequentially changing the carrier frequency according to the phase within the fundamental wave period, the maximum amplitude of the pulsating current ripple can be reduced and suppressed to a uniform current ripple, and the switching frequency can be lowered as much as possible. .

  As a result, it is possible to realize PWM modulation in consideration of both conflicting demands for reducing the switching loss of the main circuit element and reducing the copper loss and iron loss in the electrical equipment of the load due to current ripple.

  The above is the effect of the first and third embodiments.

(Embodiment 2 and Embodiment 4)
The difference between the first embodiment and the third embodiment, and the second embodiment and the fourth embodiment is that the modulation method is changed from three-phase modulation to two-phase modulation. Other than that, it has almost the same function.

  Similarly, this effect can be confirmed by comparing the XY-axis current components of FIGS. 43 and 44 and comparing the current ripple trajectories of FIGS. 40 and 41. When comparing FIG. 40 and FIG. 41, there is no change in the large current ripple component as in three-phase modulation, but the maximum norm component of the current ripple is smoothed (circular distribution) to a small extent. I can confirm that. This can also be confirmed from the fact that the current ripple locus in FIG. 41 is circular. Further, even in the case of two-phase modulation, it is an advantage of this method that the carrier frequency can be controlled in accordance with the modulation rate similar to that in FIG.

(Embodiment 5)
FIG. 45 shows a scheme in which the fourth embodiment is applied to the two-phase modulation and the weighting factor correction in the norm calculation of the fifth embodiment is further applied. The waveform corresponding to FIGS. 43 and 44 is FIG. 45, and the locus shape of the current ripple corresponding to FIGS. 40 and 41 is FIG. Here, k _x = 1.6 is set so as to suppress the active power component, and conversely, k _y = 1 / 1.6 is set so as to increase the reactive component. Then, the amplitude setting of the current ripple is adjusted so that the average switching frequency becomes substantially equal to that in FIG. u. It was.

In FIG. 42, the vector of the maximum point of the current ripple returns to an ellipse this time, and the difference between the major axis and the minor axis is larger than in FIG. The feature of FIG. 45 lies in the current ripple pulsation waveforms of i _X and i _Y , where 6 times the pulsation with respect to the fundamental wave is suppressed for i _X and the maximum ripple width is further suppressed. On the other hand, the amplitude and pulsation width of the current ripple of i _Y are increasing. In order to suppress the ripple component affecting the active power, that is, the torque, the carrier frequency is set high when the current ripple is large, and the average switching frequency is also increased in this state. The carrier frequency is reduced by allowing the ineffective portion to be expanded in a period where the frequency is small. In this way, PWM with a small torque ripple can be realized while maintaining the average switching frequency. This is the effect of this method.

  In addition, since the average switching frequency is almost the same as that in FIG. 44, it can be said that the switching loss is reduced as compared with the conventional method shown in FIG.

The features of the fourth and fifth embodiments are completely different, such as the difference in the shape of the current ripple waveform in FIG. 41 and FIG. These superiority and inferiority are evaluated differently depending on the function required by the system to which the advantage is applied. Embodiments 1 to 4 may be applied to a system that places importance on reducing the loss of the main circuit, and Embodiment 5 may be applied to the case where importance is placed on the torque ripple of the load. However, even when selectively using these methods, the amplitude setting .DELTA.i ^* the current ripple can be realized only by changing the two weighting factors k _x, k _y upon predictive calculation of the current ripple, the control program, etc. are the same as Can be used. Therefore, it is not necessary to manage the program for each product, and it can be said that one of the advantages is that the function can be changed only by setting in terms of software history management.

DESCRIPTION OF SYMBOLS 1 ... Biaxial voltage command calculating part 2 ... Carrier generating part 3 ... Phase voltage command generating part 4 ... PWM comparator 5 ... Gate drive part 6 ... Main circuit 10 ... Current ripple norm estimating part 11 ... Carrier correction coefficient calculating part

Claims (4)

An inverter control method for controlling an inverter that outputs a three-phase AC voltage by a PWM modulation method at a variable voltage variable frequency,
The voltage command is updated in synchronization with the upper and lower vertices of the triangular wave carrier signal, or a period that is an integral multiple of the voltage command is used as the update cycle.
Input the voltage command, correct the voltage command by applying zero-phase modulation by three-phase modulation method or two-phase modulation method based on the reference carrier frequency,
Using the corrected voltage command, the vector locus of the current ripple of the output current during the half cycle of the triangular wave carrier signal is estimated, the norm of the vector locus of the current ripple is calculated, and the maximum component that is the maximum component of the norm is calculated. Select the norm,
Calculate the carrier frequency correction factor so that the maximum norm value of the current ripple matches the allowable current ripple,
Set the carrier frequency command to a value obtained by multiplying the reference carrier frequency by the carrier frequency correction coefficient,
An inverter control method, comprising: outputting a PWM pattern for controlling on / off of a switching element using a corrected voltage command and a triangular wave carrier signal based on a carrier frequency command. When estimating the current ripple norm,
Taking the rotating coordinate system, the phase angle of the vector of velocity electromotive force due to the field magnetic flux is considered to be equal to the phase angle obtained by converting the voltage command into polar coordinates,
The value obtained by adding the reference phase used for rotation coordinate conversion and the phase angle obtained by converting the voltage command to polar coordinates is used as the reference axis of the rotation coordinate, and the corrected voltage command vector is selected as the reference axis of the rotation coordinate. Calculate the scalar quantity of the vector as this reference axis,
The current ripple trajectory is estimated from the voltage component of the difference between the scalar quantity of the output voltage vector of the reference axis and the scalar quantity of the speed electromotive force due to the field magnetic flux and the time interval at which the voltage vector is output. The inverter control method according to claim 1.   When calculating the norm of the vector locus of the current ripple, the current norm of the reference axis and its orthogonal axis are multiplied by different weighting factors, and the norm is calculated from each axis component after the weighting factor multiplication. 3. The method of controlling an inverter according to claim 2, wherein the maximum norm is selected from.   A voltage-type inverter that outputs a three-phase AC voltage using the inverter control method according to claim 1.

Images