JPH03215182A - Current controlling system of variable speed driving gear - Google Patents

Abstract

PURPOSE:To perform exact current control for enabling responding to be performed at a high speed by detecting current data synchronized with sample hold signal, and by forecasting current after delay by a control time from the current data, to perform currant control operation. CONSTITUTION:To a voltage type PWM inverter for performing vector control, current feedback is applied, and in other words, in a PWM inverter device provided with a vector controlling section 1, a current controlling section 2, and a PWM pattern generator 3, the generator 3 is provided with a timer for enabling the output of a sample hold-timing to be generated at the intermediate point of a O vector voltage component period, and current data at its time synchronized with this sample hold signal are detected, and by the current controlling section 2, current at a following sampling time is forecast through the current data of a detecting circuit, and current control operation is performed.

Description

[Detailed Description of the Invention] A. INDUSTRIAL APPLICATION FIELD OF THE INVENTION The present invention relates to a current control method for a variable speed drive device that operates as a current control inverter by adding current feedback to a voltage type PWM inverter that performs vector control.

B. Summary of the Invention The present invention provides a PWM inverter device including a vector control section, a current control section, and a PWM pattern generator.
The current data at that time is detected at the midpoint of the 0 vector voltage component period of the M pattern, and the current control unit uses this current data to predict the current at the next sampling time and performs current control calculations, thereby achieving smooth control with little error. This allows current control to be achieved.

C. Current control (ACR) in conventional technology inverters is a PI system that adjusts the output voltage according to the current error, as shown in Figure 12.
Control method is common. However, in this case, the PI control gain is determined by adjusting it on the actual machine, and adjustment is difficult.

This occurs because the relationship between the output voltage and the increase/decrease in current is not constant, and if this is formulated, no adjustment will be necessary. In other words, as shown in Figure 13, ACRP
The output of the I amplifier may be regarded as a command to change the current, and a voltage calculation formula may be used to change the current according to this command.

In Figure 13, the feedforward term outputs a voltage component necessary to maintain the command current, and the proportional term outputs a component that changes the current so that the command value and detected value match,
The integral term compensates for deviations due to errors in constants, etc.

Conventionally, ACR control has been performed using an analog system as shown in Figure 14, but since coordinate conversion can be easily performed in a digital system, the coordinates are converted to rotational coordinates as shown in Figure 15.
ACR control is now available. Furthermore, if digital control is performed as shown in FIG. 14, wasted time such as sampling and CPU calculation time will occur, resulting in a phase delay even in a steady state. Even if you try to increase the loop gain, a limit cycle will occur due to the effect of dead time, making it easy to become unstable.

Therefore, currently, a method is used in which coordinates are transformed onto rotational coordinates and ACR is applied by a DC amount, as shown in FIG. As a result, errors such as phase delay were improved.

Next, this coordinate transformation will be explained.

As shown in FIG. 16, the two orthogonal axes fixed on the motor stator winding are α and β axes, and the power supply angular velocity ω is on the α and β axes. Let the coordinate system rotated by the d-q axes.

The basic voltage-current equation in this rotating coordinate d-q axis is expressed by equation (1).

Margin below v,. ,V,, 111S+IIll λ, 6, λ,, R, R, M L.　　： L： Primary voltage Primary current secondary magnetic flux primary resistance secondary resistance Excitation inductance primary inductance secondary inductance ωr ω.

P Two-rotor angular velocity (electrical angle) : Power supply angular velocity (electrical angle) : Differential operator (5) and The underlined part of formula (5') is common, summarize and (6) Equation (7) is obtained.

Next, equation (7) for vector control is determined.

Vector control is λta/M=CONSt=I+a ・−・−18) λ
tq/M-0 ・・・・・・・・・(9)
In order to satisfy this condition, the slip frequency is controlled using equation (10).

ω5: Slip angular velocity (8), (9), (10) in the second and third terms on the right side of equation (7)
By substituting and transforming the equation, equation (1 1) is obtained.

In equation (11), current is expressed by a command value and an error term as shown in equation (12).

1=I" +ΔI ・・・・・・・・・ (12
) I": Command value, Δ1: Error or differential. As shown in equation (5), this error is approximated by the difference in a minute time ΔT. Also, if it is assumed that the change in the command value during a minute time is small, , PI=P (I"+ΔI) (12) Equation. Substituting equation (13) into equation (1l), (l
4) Equation is obtained.

Therefore, the first term in equation (14) is a term that is proportional to the current error if L 71 1 ΔT is a constant. The third term is the motor constant and output frequency ω. and the detected value of the current command are unnecessary feed-forward terms. Therefore, although these two components can be easily realized, the second term is a nonlinear equation because it includes the term ω0. Here, since this generally holds true when the ACR cycle: ΔT is made as high as the PWM cycle, the second term can be regarded as smaller than the first term and can be omitted. In addition to this, an integral term is added to compensate for steady-state deviations due to errors in constants, etc., as shown in FIG.

D. Problems to be Solved by the Invention The first problem to be solved is that the actual current cannot be detected accurately unless the PWM ripple current waveform and sampling timing are good, for example as shown in Figure 17(a). .

In order to avoid this, it is necessary to increase the sampling timing as shown in the diagram (b), or to perform integral calculation using an analog system. If you increase the sample timing, A/
The speed of the D converter and the like must also be increased, making the circuit complex and increasing the cost. Additionally, adding an analog circuit increases the number of adjustment steps and makes it susceptible to temperature fluctuations, among other problems.

The second problem is that in Fig. 13, the current command value and the error in the detected value are simply PI-controlled and a feed-forward term is added, but the voltage output is delayed after one sample period. Therefore, there is a problem in that the ACR gain cannot be increased due to wasted time from current detection to compensation voltage output.

The present invention has been made in view of the above-mentioned problems of the conventional technology, and its purpose is to detect current without detection delay without including PWM current or pull current from the current vector locus. enables faster response,
An object of the present invention is to provide a current control circuit for a variable speed drive device that is capable of more accurate current control.

E. Means for Solving the Problems In order to achieve the above object, the current control circuit of the variable speed drive device of the present invention includes a vector control section using data such as a speed or torque command and speed detection, and this vector control section. A current control unit that performs current control calculations based on current commands from the current controller, and a PWM pattern generator that calculates a three-phase PWM pattern based on the voltage amplitude and phase commands of the voltage space vector from this current control unit, and issues PWM commands based on the PWM pattern. In the PWM inverter device equipped with the PWM
The pattern generator is provided with a timer capable of outputting a sample and hold timing at an intermediate time of the 0 vector voltage component period, and the current data at that time is detected in synchronization with this sample and hold signal, and the current control section controls the current of the detection circuit. The current control calculation is performed by predicting the current after the control time delay based on the data.

F. Since the operational sample and hold circuit samples the output current at the midpoint of the current O vector period according to the timing from the PWM pattern generator, it is possible to detect a current that approximates the average current of the IPWM period, which is less affected by the PWM non-noble current.

The proportional term of the current control section performs current prediction calculation to predict the current value at the next sampling time based on the current detection value and current change rate command. The current change rate required to follow the current command is calculated from the difference between the current detection value, the current preliminary correction amount, and the current command, and the voltage required to achieve this current change rate is calculated. calculate. The combined value of the voltage due to this proportional term, the feedforward voltage component, and the integral voltage component for steady-state error suppression is output. By performing current prediction here, wasted time required for control calculations, etc. from current detection to voltage output is reduced. Current control can be performed with less influence and less error.

G. Example <principle> First, the principle of the present invention will be explained.

(1) PWM output voltage pattern creation There are various methods for creating PWM patterns, but in this example, as shown in Figure 3, vectors within one pWM period are divided into 60 degrees including the 0 vector (v0). Three voltage vectors V, Vp, V. A circular locus method is used in which these periods are limited to and variable.

If the carrier frequency of PWM is fe[Hzl, then ? Ru. If we replace the voltage averagely equivalent to the voltage command vector V1 of this half-wave period Tel with two voltage vectors V, ■, and O vector V0 with a 60° difference, v■, V,
,V. The time taken by the vector is t0=TCt tA
t#. ．． ．． ．． ．． ．． ．． ．． ．． ．． ．． ．． (17)
v6c is the DC power supply voltage Vde= I VA l =
l v, Tct is the half period of PWM θ9 is V. An angle from to ■ is sufficient. By equations (15) to (l7), va is equivalent to V,
,v,,v. Find the period.

Next, the order of V, V, and vector is changed every half wave. Further, the O vector is divided into two and distributed on both sides of the λ and μ vectors as shown in FIG.

In this way, the voltage is replaced with a PWM pattern.

(2) Locus of current vector when outputting the above voltage Consider equation (11).

Shift the differential term to the left side of equation (11) The current at the start time of the PWM half cycle is (I..) (1.q). . shall be.

(11') Expressed by difference approximation in TCt time, here here, T. ，! The voltage taken during the time is ■yo T yo time ■， T2 hours V. -T. T e2 (T, +T,) time Therefore, each three voltage vectors Expressed by the sum of the periods of tors. ST (l8) The formula becomes

Substituting the voltage vector and time during the period of equations (15) to (17) into equation (14'), we get Δlld+Δi,9: Amount of current change between Tct Here, expressed as a vector for simplification, L= =d−i+a+
From 'll・IIQ (d.'It is a unit space vector), Δj・=−L
{(V・1El)t・+(V,−El) t, 10((
D-El) t 2]...(19) Shiiichi'' [(Vl-tA+V,-t,+C)-L)-E,(
t. +t, +t, )]L. 1 ΔA=-C (V,-E.) Tc2] -...
...(20)L. Therefore, after a PWM half cycle Tct, the current vector becomes [V. -E. l] in the direction of T e
, /L, will move for the time multiplied by .

The right side of equation (19) is: the first term is the distance traveled during the period ■ in Figure 4, the second term is the distance traveled during the period ■, and the third term is the distance traveled during the period (■ + ■). It shows the distance traveled.

When this is expressed on a vector diagram, the current movement directions in each period are (V.-E.), (V, -E,), (Cl
-E,), which is parallel to each vector direction seen from E1. Then, the composite value of ■ to ■ moves by a distance expressed by equation (20).

In the half-wave section ■→■ where the order of λ and μ in FIG. 4 is reversed, the result is as shown in FIG. 6. That is, although the order of the λ period and the μ period are reversed and the vector locus shapes are different, the composite movement vectors Δα from ■ to ■ are the same.

(3) Current sampling Here, in steady state, the VI vector and E1 vector in Fig. 5 match, and the trajectory of the current vector is
Since the ending times of
As shown in the figure, it becomes a parallelogram (t+jll{O has exactly such a proportion). Here, O consisting of ■ and ■
The current vector at the middle time of the vector period (the time at the border between ■ and ■) indicates the center of the parallelogram, and the PWM
Since it takes a figure-8 triangular locus in the I period, it passes through the center twice.

From this, if to is applied to both ends of the λ vector and μ vector, and the current is sampled at the point of mark 9 in Figure 5, the current value of one cycle of the PWM current Linople can be calculated by sampling at one point. It can be seen that data equivalent to the average value can be obtained.

From the above, if the current is sampled at the midpoint between the PWM output using the circular locus method and the 0 vector period, the average value of the PWMI cycle can be measured with only one sampling.

Furthermore, when the current vector is moving as shown in FIG. 6, the amount of movement of the center of this parallelogram can be determined.

In addition, the trajectory of the current linopule moves on a quadrilateral, but since the center is captured, the average value, which does not include current ripple, can be measured with a small number of sampling times (about twice the PWM cycle) with a time delay. Become.

Also, if we look at equation (20) from the perspective of controlling the current, we can see that when the current vector is moved by 80 after T ct time from the current vector, V of equation (21), which is a modification of equation (20), is All you have to do is output .

L.　　　.

V, one (E, + one Δ[1) − (2 1) (21)
If VI of the equation is output for half a cycle, the current vector moves to the position of (fi+Δfi).

Moreover, when maintaining the current without moving, it is sufficient to set Δfi=c) in equation (21), and V, -E. Just keep outputting.

From the above, even if there is PWM current ripple, PWM
If you focus on points every M half cycles, you can easily move in any direction in half cycles.

(4) Current control method When PWM voltage output is performed using the circular locus method using the CPU,
ACR calculation time is required after current sampling, and P
Only a voltage delayed by a half-wave period of WM can be output. The time t from when the voltage is output until the current reaches the target value takes another half PWM cycle as shown in FIG.

In other words, an IPWM cycle is required from current detection to correction response.

(2) Set the detected value of 1o to input-ACR.

■Set the ACR calculation output voltage to the PWM timer.

(2) Output voltage during the period Ly1*1→tn42. This changes the current.

(When the current is H*2, the current follows the command value.

Two samplings are required from detection to response in 1 hour (
PWM1 period).

Here, the ACR operation performed by ■ is in→I n+1 of 2
It does not calculate the control amount between samplings.

In other words, 1. ．． →The change between 1 and 1 was output to the PWM timer by the previous ACR calculation [based on 1 n-1]. →I. It is determined by the voltage during the period , ,
This cannot be changed. As an ACR calculation,
After predicting the amount of change in the period 1n→L ne1, the remaining t 1n*+)→tln. It is necessary to perform calculations to reach the target value in the interval , . This will be discussed in detail.

As a premise, (a) At time t when the vector control condition is satisfied.

The current value at is expressed by coordinates where the secondary magnetic flux at time tn and the d-axis coincide. This is expressed as a space vector called L.

Hereafter, vectors are based on the dq coordinates at time 1n.

(b) The ACR output voltage after L detection is limited by the PWM output method, so it can only output the value in the interval tn*l→tn''2 after one sample.The interval (tn→1-+) is due to calculation delay. Therefore, it cannot be controlled.

(c) To make the explanation easier to understand, it is assumed that the rotation speed is constant and the magnetic flux is also constant.

(d) In the above state, a command of a composite vector a1 of excitation current and torque current commands is input from the ASR control circuit, and the ACR performs follow-up control by regarding this value as a command value. (Since the magnetic flux is constant, Ild is constant and I IQ"
However, to simplify the notation, it will be expressed as a vector dI. ) Current: Vector definition [Current time t,...] dn: Time t. Current detection value j, . 1": Current command value A at time t (n*l1)
n. t': Time t,. ．． The target value 1/\, ! set by ACR so that the current vector exists at , , ! an*+: t7 using the detected value of the current in at time tn
．． Current prediction when the current changes at time , Here, the predicted value etc. is calculated from other sampling times tx at current time tn.
It also handles the current of , and is more accurately expressed as follows.

Clx] tn X is the predicted time, and 1n is the time at which the prediction is made. However, if Dtn is omitted for the sake of simplicity, all data is defined to be processed at the time tn. Please note that time has two meanings: the time when the predictive calculation is performed and the predicted time.

Based on the above assumption, FIG. 8 will be explained.

The result of ACR calculation using the detected current value d at time n is the voltage command υ for the period jy1*1→t y1*2.
．． ．． ,,→0. , , but what can be changed by this voltage is the amount of change in the current during the period (n+1)→(n+2).

The problem here is how to change the amount of change in current. When using a discrete value system, the current command and current prediction must be defined. As long as the amount of change in current is known, the voltage output etc. required to realize it can be calculated using equation (20).

For example, if the output voltage for the period n→(n+l) can be output immediately after the current is detected at time 1n (In), there is no time delay component and there is no need for current prediction, etc., so [ΔI1n-4-1nll ” ] t-= [(1”
E t- (L) t- """ As shown in (22), the difference between the current current command and current detection at 1n is expressed in the interval tn→t0.. between the current and the next sample. The ACR may control the voltage so as to change the voltage.

However, in reality, there is a calculation delay, so t. ．． ,,→t
．． Control must be performed using n*t+. Therefore, the current 11. , it is necessary to predict.

[About current prediction]

After detecting the current l11, ACR calculation is performed and the voltage is output. ．． ．． Until then, a dead time of one sampling period occurs. If the period after the next sample time (fl...)
．． n. ．． There is no problem if it can be measured, but the future value cannot be determined, so the amount necessary for control is determined by t. , the predicted value is used instead of the detected current value.

, /\, Prediction of command value [11. ．． ．． ”) tn... At time tn, the predicted current command at t(A1111) and the predicted current detected value at t(n*++ are given the symbol of the predicted value, and 8 is the command value. .

Actually, since the voltage is calculated based on the magnetic flux phase, it is necessary to predict the magnetic flux, but since vector control is established and the rotational speed is constant, this does not appear on the rotation chamber elevation.

Regarding prediction of two command values] Current command and child measurement are difficult (just as ASR cannot predict unless it can output both current and current differential command),
-nZ, it is assumed that the command value continues for the next period jn++.

Regarding the second detection (direct,) prediction, rule] The current detection value can be predicted using the following 90,000 method.

to:] n'l:ln-:1 n)+. One [Δyu. =+n
+Il]In-1'...(24) Current value: Y. : , 0 and previous (tn 1) current change command (according to ACR output) [Δb - Is -
Let the composite value with 'l be the predicted value.

In this way, t. ．． The current command and current detection (actual value prediction) at time ll are predicted. In this way, when the predetermined value of al at the time of voltage output is determined regarding the current command and the detected value, the ACR operation is performed. ．． ,》→t. (25) in the period of 2 (if the current is varied, j
When fn''l1, the command value and the detected value match. In other words, the tracking is completed after 2 samplings after detection.The error with the command value including the current prediction is expressed by equation (26).

Convex〒 [ko, n8 two, n {[yu. ], n-I-CΔL-fnl+”]In-.
1 }・・・・・・・・・(25) Here, when comparing the left side of the equation (next follow-up command value) and the third term on the right-hand side (previous follow-up command value), it is replaced with n-1 (d) n. It turns out that it is 12 otsu. In other words, the times (1
It can be seen that it is sufficient to use the value of Δd' obtained in (o−+) as the predicted correction amount for the next current detection.

It can be said that by repeating this, the current prediction equation (24) can be obtained using the left side of equation (25) for the next step.

CACR proportional gain] In equation (25), although the current is delayed by two sample times,
After that, he was considered a follower. In other words, the tracking amount was regarded as the error between the predicted command and the predicted detected value.

In reality, accurate tracking is not possible due to output voltage accuracy and constant setting errors, so a relaxation gain of k (0<k<1.0) is added to formula (25) to prevent overcompensation and instability. Multiply.

This k is AC This is R proportional gain.

[ΔA fnll-in+21'E k・[[:L”]..

{[Yu. ],. One [Δd. 1+ll”E+.-1}]
(26) Therefore, after 2 samplings, the command value C fi n”)t
n was not reached, and ACR set a new target value for mitigation [do. 2],, (lt, +k[CL″],.

(1: l nJ jn+ EΔL−1−,,, .
I″〕、.

}” (27) will begin to follow.

If the current change control accurately follows the target value signal indicated by the circle, the relaxation gain k will be 1.
If it is smaller, the current detection value starts tracking with a delay of one sample, and then follows with a first-order delay, as shown in FIG.

[Voltage calculation]

As for the voltage component, r. Therefore, by substituting into equation (14'), t fn"ll-"p
(Calculate the voltage in the section H + 1 2 1. In addition to these, the foot-feed term in equation (l・1): used when calculating the first term.) Steady-state deviation caused by an error in the motor constant. etc. These are uncertain, and if the response is made too fast, it will become unstable, so it will be compensated for by adding the integral term footbath.

The physical meaning of formula (14') will be explained.

Yu. ．． 1 to 1, . , when looking at the fixed coordinates (α-β axes), as shown in FIG. 10, the phase on the dq axes Lfn * +→? . ．． , and the phase changes by the sum of the rotation angle Δθ - ω1·Δt of the d-q axes themselves.

(Δt: 1 sample time, Δθ: advancing phase between 1 sample on the d-axis.) A nll → as seen on the α-β axis (fixed coordinates) in Figure 10
It can be said that the change in fl not is divided into the following two components, each of which is handled by the term in equation (14').

(a) 5 on the d-q axis. ．． Component that rotates by 80 degrees on the fixed coordinates while maintaining [2nd term of equation (14')] (b) Component that moves to (Ii.,, → (n6!l) on the d-q axis [ (14') 1st term] If we consider it in more detail, the above (b) component (14') 1st term is the current command value (
(or target value) and the detected value is divided by one sample time and regarded as the command value for the current rate of change. here,
This current error includes a component that occurs due to a change in the command value and a component that occurs due to a change in the detected value.

The former is a term that is predictable and allows feedforward control to be performed using the differential amount of the command, but if
Even without this feedforward voltage, a current error of this amount occurs as the latter error, and then current control (A
CR) adds that amount to the current follow-up command and corrects it, so current follow-up is possible. However, compared to the case where feed-forward is performed, once an error occurs, it is dealt with using the ACR time constant, resulting in a larger current response delay.

(Here, the feedforward term is the derivative of the current,
Since it is susceptible to noise and has stability problems, the example shown in Fig. 1,
It is omitted in 2. ) To summarize the above, υ. . Ill→(tno21 can be said to have the following characteristics.

■The current error caused by disturbance etc. is calculated by t at time tn.
Command value at time of n&+ [: fi n. ,") Predicted value of tn, detected value ui" at the time of t,.,] Obtained from the predicted value of In and multiplied by the relaxation coefficient k as the target value, the voltage Δυtoad back component that controls the current is calculated. include.

(2) The predicted value of the command value of jy+*1 [(In. 11) is used to feed forward the voltage component maintained by this.

? It is also possible to calculate feedforward voltage components from differential components such as torque commands and excitation, but this time we omitted them to prevent instability due to noise and other causes. .

Then, the voltage υ,'' including ■ to ■ is calculated using the detection data of d at time t0, and tLn.. → t.
It is output as a voltage during the period (2).

C Convert the voltage command on the d-q axis to the α-β axis (U, V,
When considering the compensation voltage calculated at time t7 when C(lJtn was detected) from the fixed coordinates of the α and β axes, there is a time delay of one sample before the output of the compensation voltage starts. Therefore, at time t0., the d-q axis itself has moved by Δθ0 on the α-β axis.Therefore, although it was unnecessary on the d-q axis, the α-β axis In the above, this phase change must also be taken into account.Furthermore, the period t during which the compensation voltage is output.
-+tn. , the voltage vector is also rotating on the α-β coordinates.

In other words, the average value of the phase change between ΔT and Δθ is further considered.

The voltage equation (14') in the previous section is a value expressed on the dq axis, and indicates the average value of the PWM period. However, this average value is υ when viewed from the fixed coordinates of the α-β axis that actually outputs the PWM voltage. a+) →. ,. , , are also rotating at the speed of ω. Therefore, to create a PWM pattern, a time average value is used as an approximate value of this rotating vector voltage. , V,, calculate the output time of V0.

In this way, on the fixed coordinates, the PWM pattern must be approximated by taking into account this rotating component, so as shown in equation (36), the voltage output phase at time 1n is 20. =(φJ (n”ll (..2.) phase).

<Example> An embodiment of the present invention is shown in FIG. This will be explained with reference to FIG.

In FIG. 1, l is a vector calculation unit, 2 is a current control unit, 3 is a PWM pattern generator, 4 is a dead compensation circuit, 5 is a gate signal circuit, 6 is a drive circuit, 7 is an inverter main circuit, and 8 is an induction motor, 9 is a voltage detector, 10 is a current detector, 1l is a speed detector, and 12 is an A/D converter of the output current at that time in synchronization with the 0 vector intermediate sample and hold signal from the PWM pattern generator. 13 is an A/D converter that converts the analog current signal from the sample and hold circuit into a digital quantity so that the CPU can handle it; 22 is the rotor angular velocity ω7 from the speed detector 11; Power source angular velocity ω from angular velocity ω5. This is an adder that obtains . Incidentally, the angular velocity ω8 is determined by the calculation circuit 21 of equation (10) using Ii6el as shown in FIG.

The vector control calculation unit 1 calculates the primary current command α1 and the slip angular velocity ωS using data such as a speed command or torque command and speed detection.

As shown in FIG. 2, the current control unit 2 inputs the current command 1 and the current data from the A/D converter 13 into the current command. Current data 1 whose coordinates have been converted by the FIJ two-phase conversion rotational coordinate calculation unit l4. Using the equation (25), the circuit l5 predicts the current In'l. 16.17.18 is the feedforward term according to equation (l4). Proportional and integral terms! There are 8. From the feedforward term l6. ) In the voltage component for maintaining the combined current, the voltage change amount ΔV P (nol-
n+21 on the integral term l8;] voltage change Δ■. . +
l = n + 2 Add 1? '! r voltage change Δ■an-■. , , is added and f two voltage components V In
Obtain *I=n*2, transform this voltage component into polar coordinates l9
The voltage amplitude on the d-Q coordinate is E=V*
+ 1 n * 2 1・and phase φV + n + +
one. ．． , , is obtained. Using the power supply angular velocity ω0 from the adder 22 for this phase, Δθ is calculated, and the integrator 20 calculates Δθ.
Phase angle θ [ n * I = n + 2 1 and the advance of the voltage vector? Adding the compensation amount Δθ, the phase (θV
). ■》→《. ．． Get 2. That is, since there is a PWM delay. one. ．． , calculate the voltage E in equation (28).

The PWM pattern generator 3 generates a voltage amplitude E and a phase (θ, .) of the voltage space vector from the current controller 2. 1. 10
．． 2+, a PWM calculation control section that calculates an equivalent PWM pattern using the circular orbit method, and a
The time of the M pattern is compared with the reference time obtained from the basic clock and counter to generate the command value of the 3-phase PWM signal, and the midpoint of the 0 vector voltage component period (
The sample-and-hold timing is output to the sample-and-hold circuit 12 during the O vector period of the voltage that compensated for the dead time delay.

The dead time compensation circuit 4 measures the delay time between the PWM switching command output from the PWM pattern generator 3 and the switching time of the actual voltage output for each switching, and always calculates this delay time or the set delay amount. The switching time is corrected so that it remains constant. Also,
The PWM generator 3 also outputs sample and hold timing to the sample and hold circuit 12.

Then, the gate signal circuit 5 obtains a gate signal based on the switching command corrected by the dead time correction circuit 4, drives the inverter main circuit 7 through the drive circuit 6, and rotates the induction motor 8.

}■． Effects of the Invention Since the present invention is configured as described above, it produces the following effects.

Current control is performed by predicting the current at the next sampling time, so there is no PWM delay, and detection equivalent to the average of one sample that does not include PWM resolvable current can be obtained, making it possible to remove limp. Since a low-pass filter is not required and smooth current detection is possible, it has high-speed current control characteristics with little time delay, and the voltage generator can also use a timer etc. to short pulse width times. Since it is possible to obtain an output voltage with high resolution, it has current control characteristics with a small amount of current overshoot.

[Brief explanation of drawings]

FIG. 1 is a block circuit diagram of a variable speed drive device showing an embodiment of the present invention, and FIG. 2 is a circuit diagram showing a current control section in the same circuit in functional blocks. Figures 3 to 11 relate to the explanation of the principle of the present invention, and Figures 3, 5 to 7
Figures 10 and 11 are vector diagrams, Figure 4 is a diagram explaining the order of vectors, Figure 8 is a diagram explaining the relationship between current command and voltage output, and Figure 9 is a diagram explaining the response characteristics of current command. be. 12 to 17 show the principle of conventional current control, and FIGS. 12 and 13 are block diagrams of the current control system. 1... Hector control section, 2... Current control section, 3...
PWM pattern generator, 4... Denode time compensation circuit, 5... Gate signal circuit, 6... Drive circuit, 7
...Inverter main circuit, 8...One-induction motor, 9...
- Voltage detector, 10... Current detector, 11... Speed detector, 12... Sample hold circuit. Entering μ μ−λ Figure 6 Figure 7

Claims (1)

[Claims] (1) A vector control unit that uses data such as speed or torque commands and speed detection, a current control unit that performs current control calculations based on current commands from this vector control unit, and a current control unit that uses voltage space vectors from this current control unit. In a PWM inverter device equipped with a PWM pattern generator that calculates a three-phase PWM pattern using voltage amplitude and phase commands and issues a PWM command based on the PWM pattern, the PWM pattern generator is provided with a sample at the midpoint of a zero vector voltage component period. A timer capable of outputting hold timing is provided, and the current data at that time is detected in synchronization with this sample and hold signal, and the current control section predicts the current at the next sampling time from the current data of the detection circuit and performs current control calculation. A current control method for a variable speed drive device characterized by performing the following.