JP2003219690A - Detection method for output current of pwm inverter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce switching noise and PWM ripples in a voltage PWM inverter which has a current control system and subjects an induction motor to variable-speed control. <P>SOLUTION: A moving average computing portion 8 computes the moving average of the multi-point sampled data of the output currents from an A/D converter 8, and one cycle interval of PWM carriers is taken as the moving average window. In addition, a method is included wherein a PWM ripple model is created based on the induced voltage and applied voltage of a motor and leakage inductance, and PWM ripple portions are reduced by reducing PWM ripple portions from the sampled data of detection current. Further, a method is included wherein an instantaneous current detection value lightly affected by PWM ripples in the previous PWM carrier period is evaluated by current detection based on a PWM ripple model, and the leakage inductance is automatically corrected in such a direction that PWM ripples are reduced. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a PWM inverter having a current control system for variable speed control of an electric motor, and more particularly to an output current detection method with reduced PWM switching noise and PWM ripple.

[0002]

2. Description of the Related Art In a voltage type PWM inverter, a PWM waveform is generated by comparing a carrier signal (triangular wave) and a frequency command (sine wave), and a switching element of an inverter main circuit is turned on / off by using this as a gate signal. By doing so, a sine wave output is obtained.

On the other hand, a PWM inverter controller is provided with a current control system in a minor loop of a speed control system or a torque control system, and is further digitized by using a high-speed arithmetic digital signal processor (DSP) to improve control performance. I am trying.

In such a PWM inverter, switching noise due to a carrier signal and generation of noise at a current inflection point due to a Hall CT which is a current detecting element poses a problem. Further, there is a problem that the digital current control is easily affected by the PWM ripple.

Therefore, as shown in FIG. 9 showing the timing of the conventional output current detection method of the PWM inverter, there is a synchronous sampling method for sampling in synchronization with the apex of the PWM carrier signal. This synchronous sampling method uses PW
It is characterized in that sampling is performed at a zero voltage point so as not to be affected by the PWM ripple generated in synchronization with one cycle of the M carrier signal.

As another method, as shown in FIG.
There is a multi-point sampling method in which the WM carrier signal is sampled multiple times within a half cycle and averaged. This multi-point sampling method is characterized in that sampling is performed at many points in order to reduce spike noise generated when the power element is switched, and the influence of noise is reduced.

[0007]

Although the conventional synchronous sampling method and the multipoint sampling method have the above-mentioned characteristics, they also have the following disadvantages.

(1) The synchronous sampling method cannot be used in applications requiring high-speed detection such as servo applications because the detection cycle depends on the carrier cycle because the sampling can be performed only at the apex of the carrier and the detection is delayed. Further, when the connecting cable between the inverter and the motor is long, noise at the time of switching increases due to the inductance component of the cable, and current detection may become unstable particularly when a high voltage is output. In this case, expensive filters have to be used.

(2) In the multi-point sampling method, when sampling is simply performed at multi-points, switching noise can be reduced by a large number of samples, but a large sample window (number of samples) increases detection delay. Also,
Due to the relationship between carrier period and sampling period, P
The effect of the WM riffle becomes large.

For the above reasons, the conventional current detection method is not suitable for high-distance sampling such as servo applications.

An object of the present invention is to provide an output current detection method for a voltage type PWM inverter which is capable of reducing switching noise, being unaffected by PWM ripple, and capable of high-speed current sampling.

[0012]

In order to solve the above-mentioned problems, the present invention uses a method in which a moving average window at the time of multi-point sampling is set as one period section of a carrier, an induced voltage and an applied voltage of a motor, and a leakage inductance. The PWM ripple component is obtained, and the PWM ripple component is subtracted from the detected current sample to obtain the instantaneous current detection value with less influence of the PWM ripple, or the current detection by the PWM ripple model shows the influence of the PWM ripple component in the previous PWM carrier cycle. Evaluate less instantaneous current detection value,
This is a method of automatically correcting the leakage inductance in the direction of less WM ripple, and is characterized by the following configuration.

(1) A current detection system of a voltage type PWM inverter having a current control system for variable speed control of an induction motor, wherein current detection system detects detected current by multipoint current sampling. Means and a moving average calculating means for determining a moving average value of a multipoint current sample detected by the current sampling means as a detected current, the moving average window being one period section of the PWM carrier. .

(2) A current detection system of a voltage type PWM inverter having a current control system for variable speed control of an induction motor, wherein current detection system detects detected current by multipoint current sampling. Means, a PWM ripple model that creates a PWM ripple model from the induced voltage of the induction motor, the applied voltage, and the leakage inductance, and finds the instantaneous current component of the PWM ripple; and subtracting the instantaneous current component of the PWM ripple from the detected current sample value. And a subtractor for obtaining a detection current with a reduced PWM ripple amount.

(3) The PWM ripple simulator is provided with means for evaluating the instantaneous current detection value, which is less affected by the PWM ripple in the previous PWM carrier cycle, and automatically correcting the leakage inductance in the direction in which the PWM ripple is smaller. It is characterized by

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment 1) FIG. 1 is a circuit diagram of a main part of a voltage type PWM inverter showing a first embodiment of the present invention.

The current control unit 1 obtains a current control signal by comparing the current command value and the detected current signal, generates a sine wave having a frequency that matches the frequency command value with this current control signal as an amplitude, and further generates a carrier signal. A gate signal having a PWM waveform is obtained by comparison with.

The inverter body 2 constitutes a main circuit by bridge-connecting semiconductor elements such as power transistors and IGBTs, and each semiconductor element is on / off controlled according to a gate signal from the current control section 1 to generate an alternating current of PWM waveform. The output is obtained, the AC output current of a sine wave is detected, and the induction motor 3 is driven.

In the current detection, the output current of the inverter main body 2 is detected by the Hall CT5, this is sampled by the A / D converter 6 and converted into a digital value, which is taken out as a current sample. The sampling pulse given to the A / D converter 6 uses a signal obtained by multiplying the carrier signal by the digital PLL 4.

In such carrier synchronized multipoint current sampling method, the PWM ripple appears in synchronization with one cycle of the carrier. Therefore, in this embodiment, a moving average is used in which the average window at the time of multipoint sampling is set to one period section of the carrier, and a sampling method that is not affected by the PWM ripple even at the time of multipoint sampling is adopted.

Therefore, in this embodiment, the moving average of the detected current is calculated by passing the current sample from the A / D converter 6 through the moving average calculator 8. Moving average calculator 8
Is composed of, for example, a FIFO function that sequentially takes in current samples at a sampling pulse cycle and an adder function that adds each current sample, and is synchronized with the sampling pulse from the digital PLL 4 to perform a 1-cycle section moving average calculation of the carrier. The calculation result is output as a detection current.

FIG. 2 shows the phase relationship of the carrier synchronization multipoint sampling system according to this embodiment. In the figure, the case where the sampling timing of 8 pulses is set in one cycle period of the carrier is shown, and the moving average calculator 8 obtains the average value of the immediately preceding 8 samples as the current detection value at each sampling timing to thereby obtain the PWM ripple. And reduce switching noise.

(Embodiment 2) FIG. 3 is a circuit diagram of a main part of a voltage type PWM inverter showing Embodiment 2 of the present invention.
The same parts as those in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted.

In this embodiment, the PWM ripple simulator 9 obtains a PWM ripple current component by performing a PWM ripple model calculation from the induced voltage of the induction motor 3, the applied voltage and the leakage inductance. Subtractor 10
Is a current detection value obtained by reducing the PWM ripple by subtracting the PWM ripple current component from the PWM ripple simulator 9 from the current detection output of the A / D converter 6.

The calculation of the PWM ripple current component by the PWM ripple simulator 9 will be described below.

First, the instantaneous output voltage of the three-phase voltage type inverter has eight kinds of voltage vectors including two kinds of zero vectors. As shown in FIG. 4, the PWM pulse pattern has adjacent voltage vectors V with a phase difference of 60 °.
The three units of λ, Vμ and the zero vector V ₀ are used as the minimum units, and the output time of each voltage component is determined so that the combined value of the PWM voltage becomes equal to the voltage command value V ₁ . Here, FIG.
The PWM cycle T _C is divided into a period in which the three-phase voltage is only turned on and a period in which the three-phase voltage is turned off, and a PWM half period (T _C / 2)
Is the minimum unit of the PWM pulse pattern. This PWM
In this method, PWM calculation and PWM signal output processing are
Execute every WM half cycle. The zero vector period is divided into two equal parts and arranged on both sides of Vλ and Vμ.

Next, PWM in the case of an induction motor (IM)
The principle of ripple current detection will be described (the same applies to PM). In the rotating coordinate system that rotates at the power supply angular frequency ω, the voltage equation of the induction motor using the primary magnetic flux can be expressed by the following equation (1).

[0028]

[Equation 1]

Where V₁: Primary voltage vector, i₁:one
Next current vector, λ₂: Secondary magnetic flux vector, ω_s: Sliding angle
Frequency vector, R₁: Primary resistance, R₂: Secondary resistance,
L₁, L ₂, M: primary, secondary, exciting inductance, P =
∝d / d_t: Differential operator, x: vector cross product.

Assuming that the induction motor is subjected to slip frequency type vector control, the secondary magnetic flux vector of the above equation (1) can be controlled to be λ ₂ / M = i ₀ = constant. The primary voltage vector V _{1 at} this time is given by the following expression (2).

[0031]

[Equation 2]

From the equation (2), if the differential term is shifted to the left side, the following equation (3) is obtained.

[0033]

[Equation 3]

However, E ₁ is set as follows.

[0035]

[Equation 4]

Here, the equation (3) is differentially approximated by a minute time ΔT that reaches a PWM half cycle, and the current i ₁ is an initial value component I ₁
The following expression (5) is obtained by substituting the change component Δi ₁ with the change component Δi ₁ .

[0037]

[Equation 5]

Therefore, during ΔT, the voltage on the rotating coordinates is
Flow vector movement Δi₁Is (V₁-E ₁) In the direction of the current component
Occurs.

Here, compared with the radius of the current vector, Δi
Assuming that the movement amount of ₁ is small, i ₁ ≈I ₁ and approximate paper, M ′
E ₁ during the ΔT period is approximated to be constant by considering that the fluctuation of the i ₀ current flowing in is constant.

[0040]

[Equation 6]

Then, E ₁ is the sum of the primary impedance voltage drop due to the current I ₁ and the induced electromotive force of the secondary circuit due to i ₀ . This is a current component that keeps the primary current i ₁ vector constant at the current position on the rotating coordinates, and can be said to be a component that makes the magnitude of the current vector and the angular frequency ω constant when viewed in a fixed coordinate system.

Next, the locus of the current vector during one PWM cycle as shown in FIG. 5 will be examined. Here, Vλ and Vμ are voltage vectors fixed on the stator coordinates, but the following approximation is performed in order to handle them on the rotating coordinates.

A transistor is used as the main circuit element, and the carrier frequency is about 2 kHz. Then, if ΔT is ½ of the carrier period of the inverter, ΔT = 250 μs. This is V, which is sufficiently shorter than the fundamental wave period and moves on the rotational coordinate at a speed of −ω.
Since the moving angles of λ and Vμ are also small, Vλ and Vμ are approximated to be stationary.

V in FIG. 4 equivalent to V ₁ during the PWM half cycle
Since three types of PWM voltage vectors of λ, Vμ, and V ₀ are output, the amount of current change Δi ₁ can be expressed by the sum of the three components of the following formula (7) from the above formula (5).

[0045]

[Equation 7]

Here, ΔT = T _C / 2 = Tλ + Tμ + T _{0 From the} above equation (7), three types of voltage vectors Vλ, V
When μ, V ₀ are output, each vector output time Tλ, T
Between μ and T ₀ , the current vector is (Vλ−E ₁ ) / Lσ,
It moves at a speed of (Vμ-E ₁ ) / Lσ and (V ₀ -E ₁ ) / Lσ. Lσ is the leakage inductance. FIG. 6 shows the current change component of the equation (7) drawn on the rotational coordinates in the output order of FIG.
It is FIG.

FIG. 6 shows a steady state in which the current is direct current (ω = 0).
Then, the current vector i ₁ passes on the same locus every PWM cycle and draws a parallelogram like the current vectors a to h. Then, at the intermediate time of the T ₀ period (corresponding to the circle in FIG. 6), the current vector passes through the center of this parallelogram.

Further, when a constant frequency (ω ≠ 0) is output, the angle between the E ₁ vector and Vλ, Vμ changes, so that the parallelogram remains flat at the same center position or becomes square. As it becomes, the deformation and rotational vibration are repeated. However, since the center of the parallelogram does not steadily change, it can be seen that the average of the current vector during one PWM cycle can be detected by sampling the current at the boundary time of the PWM half cycle.

Changes a to h of these current vectors Δi ₁
Corresponds to the PWM ripple, and this Δi ₁ can be strictly simulated from the PWM voltage vectors Vλ, Vμ, V ₀ , the frequency ω, the current I _{1, and the} like.

FIG. 7 shows a pattern of the current vector Δi ₁ at the time of transition such as sudden change in torque. During the transition, the PWM current command V ₁ changes, so that Vλ, Vμ,
The output time ratio of the voltage pattern of V ₀ is different. The current vector locus at this time is as shown in FIG. 7, and the starting point of a and the arrival point of d do not match. This change is (5)
Due to the equivalence of the equation and the equation (7), it becomes equal to the movement amount when the V ₁ vector voltage command is output at T _C / 2 during the PWM half cycle.

That is, the PWM which is the midpoint of the zero vector
If the current is sampled at the boundary time between the half cycles, the current vector is sampled at the intersection of the current locus when the command voltage V ₁ is output and the current locus when the PWM voltages Vλ, Vμ, and V ₀ are output in time division. You are doing it.

Also in these cases, the changes a to h of the current vector Δi ₁ correspond to the PWM ripple, and this Δi ₁ is P
WM voltage vector Vλ, Vμ, V ₀ , frequency ω, current I ₁
It is possible to perform a strict simulation from

From the above, when the induced voltage of the PWM ripple feedforward compensation motor is known, the PWM ripple model can be created from the applied voltage and the leakage inductance.

Therefore, in the present embodiment, the PWM ripple simulator 9 calculates the PWM ripple current component to obtain the PWM ripple current component, and subtracts it from the current detection output of the A / D converter 6 to obtain P
A current detection value with reduced WM ripple is obtained.

By using the multi-point samples having a short average window together, it is possible to obtain the current detection value in which the switching noise and the PWM ripple are reduced and the delay due to the filter is small.

Further, since the current sample does not interfere with the carrier period, it is not necessary to synchronize the current control calculation with the carrier signal. Therefore, the current response can be speeded up.

(Third Embodiment) FIG. 8 is a circuit diagram of a main part of a PWM ripple simulator 9A used in a voltage-type PWM inverter showing a third embodiment of the present invention. A portion different from the simulator 9 in FIG. The point is that an automatic inductance adjustment function was added.

In FIG. 8, the induced voltage model 11 uses the frequency command value and the applied voltage command value to determine the PW of FIGS. 4 and 5.
The patterns of the periods Tλ, Tμ, T ₀ and the voltage vectors Vλ, Vμ, V ₀ are calculated from the M pattern, and the primary voltage vector V ₁ is obtained from the detected current value i ₁ and the motor constant.

The subtractor 12 subtracts the applied voltage command value E ₁ from each voltage vector. The multiplier 13 has a leakage inductance Lσ set by the leakage inductance setting unit 14.
Multiply the coefficient of. The PWM ripple model calculation unit 15
The PWM ripple current Δi ₁ is obtained by performing the calculation of the equation (7) using the periods Tλ, Tμ, T _{0 and the} like.

In this embodiment, the ripple detecting section 16 automatically adjusts the leak inductance of the leak inductance setting section 14. The ripple detection unit 16 evaluates the current detection value with a small PWM ripple detected in the previous PWM cycle from the instantaneous current detection value, and further corrects the leakage inductance value in the direction in which the PWM ripple is smaller, thereby reducing the leakage inductance. It is not necessary to set it, and the leakage inductance can be automatically measured.

[0061]

As described above, according to the present invention, the PWM ripple component is obtained from the method in which the moving average window at the time of multipoint sampling is set to one period section of the carrier, the induced voltage of the electric motor, the applied voltage and the leakage inductance, By subtracting the PWM ripple from the detected current sample value, PW
A method for obtaining an instantaneous current detection value that is less affected by M ripple,
Or current detection by PWM ripple model
In the WM carrier cycle, the instantaneous current detection value that is less affected by the PWM ripple is evaluated, and the leakage inductance is automatically corrected in the direction in which the PWM ripple is smaller. Therefore, the switching noise and the PWM ripple can be reduced.

Application of Lead Compensation Filter In the conventional detection method, lead compensation (differential compensation) cannot be performed because the current detection is affected by switching noise and PWM ripple. However, according to the present invention, since the switching noise is reduced and the influence of the PWM ripple is not exerted, advance compensation can be performed. This allows it to be used in high speed sample applications.

[Brief description of drawings]

FIG. 1 is a main part circuit diagram showing a first embodiment of the present invention.

FIG. 2 is a phase relationship diagram of the carrier synchronization multipoint sampling method according to the first embodiment.

FIG. 3 is a circuit diagram of a main part showing a second embodiment of the present invention.

FIG. 4 is an explanatory diagram of PWM voltage output time.

FIG. 5 is a PWM pattern and current sampling timing.

FIG. 6 is a steady-state current vector locus of the PWM inverter.

FIG. 7 is a current vector locus in a transient state of the PWM inverter.

FIG. 8 is a main part circuit diagram showing a third embodiment of the present invention.

FIG. 9 is an explanatory diagram of a conventional synchronous sampling method.

FIG. 10 is an explanatory diagram of a conventional multipoint sampling system.

[Explanation of symbols]

1 ... Current control unit 2 ... Inverter body 3 ... Induction motor 4 ... Digital PLL 5 ... Hall CT 6 ... A / D converter 7 ... Filter 8 ... Moving average calculator 9 ... PWM ripple simulator 10 ... Subtractor 11 ... Induced voltage model 12 ... Subtractor 13 ... Multiplier 14 ... Leakage inductance setting device 15 ... PWM ripple model calculator 16 ... Ripple detector

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 5H007 AA12 BB06 CA01 CB02 CB05                       CC23 DA03 DA06 DB02 DB12                       DC02 EA02                 5H576 BB06 BB09 DD02 DD04 EE01                       EE11 GG04 HA02 HB02 JJ03                       JJ08 JJ16 LL22

Claims (3)

[Claims] 1. A current detection system of a voltage-type PWM inverter having a current control system for variable speed control of an induction motor, wherein current sampling means detects a detection current applied to the current control system by multipoint current sampling. And a moving average calculating means for calculating a moving average value of a multipoint current sample detected by the current sampling means as a detected current, with the moving average window as one period section of the PWM carrier. Inverter current detection method. 2. A current detection system of a voltage-type PWM inverter having a current control system for variable speed control of an induction motor, wherein current sampling means detects a detection current applied to the current control system by multipoint current sampling. By creating a PWM ripple model from the induced voltage of the induction motor, the applied voltage, and the leakage inductance and obtaining the instantaneous current component of the PWM ripple, and by subtracting the instantaneous current component of the PWM ripple from the detected current sample value, And a subtracter for obtaining a detection current with a reduced PWM ripple component.
M inverter current detection method. 3. The PWM ripple simulator comprises means for evaluating an instantaneous current detection value that is less affected by the PWM ripple in the previous PWM carrier cycle, and automatically correcting the leakage inductance in the direction in which the PWM ripple is smaller. The PWM according to claim 2, wherein
Inverter current detection method.