JPH07337100A - Torque controller in vector control system for induction motor - Google Patents

Abstract

PURPOSE:To improve torque control performance when a low resolution speed detector is employed. CONSTITUTION:An induction motor 1 is subjected to PWM control through a PWM inverter 2 which is controlled through a PWM circuit 3. The PWM circuit 3 is controlled by means of the sum of outputs from a noninterference operating section 4 and ACR amplifiers 5, 6 for (d) and (q) axes. A slip frequency operating section 7 operates a slip frequency omegas by multiplying the quotient, obtained by dividing a torque command j1q* by an excitation command i1d*, by the quotient, obtained by dividing the secondary resistance R2 by the secondary inductance L2, The slip frequency omegas is added to omegar fed from a speed detection circuit 8 and the sum omega1 is delivered to the noninterference operating section 4 and the PWM section 3. A three-phase/two-phase coordinate converting section 9 is fed with iu, iw among outputs from the PWM inverter 2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a torque control device for a vector control system of an induction motor in which an extremely low speed speed estimation observer using a low resolution speed sensor is applied to a torque control system of an electric vehicle or the like.

[0002]

2. Description of the Related Art In a speed control system using a low resolution rotary encoder, an encoder pulse interval becomes longer than a speed control cycle in an extremely low speed range, and accurate speed information cannot be obtained during the speed control cycle. Therefore, in the extremely low speed range, the speed control system is likely to become unstable, and it becomes difficult to improve the response of the speed control. For these reasons, resolvers and encoders with high pulse outputs have been used as speed detectors (speed sensors) in applications requiring positioning accuracy such as servo motors and elevators. However, recently, several methods have been proposed in which the speed control in the extremely low speed range can be stably performed even by using the low resolution encoder by estimating the speed in the low speed range using an observer.

[0003]

Among these, the method of estimating the motor inertia that particularly affects the speed estimation accuracy by using the speed estimation method in the extremely low speed region using the minimum dimension load torque observer was examined. However, since the encoder pulse interval becomes long in the extremely low speed range, it is necessary to set the observer gain smaller than usual in consideration of the stability of torque estimation.

Therefore, when attempting to perform disturbance compensation using the load torque estimated value which can be estimated simultaneously by the speed estimation observer, there arises a problem that the disturbance suppressing effect in the high speed range is deteriorated.

Further, in recent years, in order to improve the global environment, research and practical application of electric vehicles have been actively promoted. Challenges to the practical application of electric vehicles include cost, efficiency, miniaturization, and reliability. Among them, studies are being made to increase the speed (10,000 rpm or more) in order to downsize the motor. However, at about 10,000 revolutions, the standard rotary encoder cannot be applied because of the mechanical strength of the speed sensor. Therefore, a non-contact gear pulse sensor that uses a magnetoresistive element and a magnetic gear has a pulse number of 6 per rotation due to the restriction of gear processing.
It becomes about 0 to 120 P / R. For this reason, accurate speed information cannot be obtained over a wide range of low speed regions, so that it is necessary to improve the characteristics using a speed observer or the like.

The present invention has been made in view of the above circumstances, and an object thereof is to provide a torque control device for a vector control system of an induction motor, which has improved torque control performance when a low-resolution speed detector is used. To do.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a PWM inverter,
An induction motor driven by this PWM inverter, a non-contact, low-resolution speed detector that sends the speed of this induction motor as a pulse output, a minimum-dimensional load torque estimated value observer, and a speed control cycle and a speed detection cycle. Converted to a discrete system model in, the speed estimation observer for estimating the speed at the pulse interval output from the speed detector, the torque command and the excitation command and the angular frequency output from the speed estimation observer is input, A vector control system for sending the output for driving the PWM inverter to the output is provided.

According to a second invention, the speed estimation observer is
First, the deviation between the torque command and the load torque estimated value is integrated by the observer model mechanical time constant to obtain the model output estimated value.
The calculation unit, the second calculation unit that obtains the average value in the pulse interval from the model output estimated value obtained by the first calculation unit, and the output of the second calculation unit and the pulse change output from the speed detector A first deviation unit that calculates a deviation from the average value speed, an observer gain unit that multiplies the deviation value obtained by the first deviation unit by an observer gain to obtain the load torque estimated value, and a model of the first calculation unit. Output estimation value and the first
It is characterized in that it comprises a second deviation unit for obtaining a deviation from the deviation value of the deviation unit, and outputs the deviation value obtained by the second deviation unit as an estimated speed value.

A third aspect of the invention is characterized in that the second arithmetic unit is constituted by means for correcting the timing deviation between the speed control cycle and the speed detection cycle.

A fourth aspect of the present invention is the load torque estimated value and the first aspect.
The difference between the speed deviations is calculated from the output of the deviation unit by the speed deviation difference calculation unit, and the calculated value is added as a correction value to the model output estimated value obtained by the first calculation unit. It is characterized in that it is supplied to the second arithmetic unit for obtaining the average value.

According to a fifth aspect of the invention, the speed deviation difference calculating section is
The observer gain reciprocal part that is switched at the next sampling time when the load torque estimated value is input and the output is integrated with the load torque estimated value, and the output of this reciprocal part and the first
The output of the reciprocal part is composed of a deviation part for obtaining a speed deviation from the deviation from the output of the deviation part, and an integrator for integrating the speed deviation difference obtained from the deviation part to obtain a correction value for the output. It is characterized in that the speed estimation value is supplied to the deviation portion and is output to the output thereof.

According to a sixth aspect of the present invention, a model output estimated value is added to an integrator which obtains a model output estimated value by integrating a velocity deviation difference obtained by a deviation unit with an observer model mechanical time constant only at the next sample. It is characterized in that

According to a seventh invention, the observer gain section is
It is characterized in that the speed is switched in two steps above and below a certain speed.

An eighth aspect of the invention is characterized in that the observer gain section is an integrated value of a fixed gain, a constant and a compensation gain.

[0015]

The angular frequency of the low-resolution speed detector is input to the vector control system together with the torque command and the excitation command through the speed estimation observer, and the torque control of the induction motor is performed by these. Therefore, good starting characteristics can be obtained even if a low-resolution speed detector is used.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. First, the speed estimation observer will be described.

(A) Principle of speed estimation The equation of the motor section when the motor is approximated to only the integral element is expressed by the equation (1). Further, the estimated torque value of the load torque observer of the minimum dimension derived from this is expressed by equations (2) and (3).

[0018]

[Equation 1]

FIG. 12 shows a block diagram of the motor section and the load torque observer of the minimum dimension from the above equations (1) to (3). In the minimum-dimensional load torque observer, the observer gain G ₁ is composed of only proportional elements. Therefore, when the load torque τ _L is applied, the motor model output estimated value ∧n _M '(the following estimated value is marked with ∧. ) And the actual motor speed n _M , there is a deviation. This deviation Δn can be expressed by the following expression (4) from the expression (2).

[0020]

[Equation 2]

Here, the physical meaning of the velocity deviation Δn will be examined. Using the expressions (1) to (3), Δn can be expressed by the following expression (5). However, it is assumed that the mechanical time constants of the motor and the model match (T _M = T _M *).

[0022]

[Equation 3]

Further, the load torque estimated value ∧τ _L can be expressed by the following equation (6) from the equations (4) and (5).

[0024]

[Equation 4]

From the above equations (5) and (6), Δn is obtained by integrating the deviation between the load torque and the load torque estimated value estimated by the primary delay with the time constant T _M *.

Here, in the method using the full-dimensional load torque observer for speed estimation, the observer gain G ₁
Is a PI element. Therefore, in the steady state, the motor model output estimated value ∧n _M '= n _M , but this does not hold in a transient state such as when the load torque changes suddenly. Also, the adjustment factor of the observer gain increases. Therefore, the speed estimation method using the load torque observer of the minimum dimension is described here. From the equation (4), the motor speed is expressed by the following equation.

[0027]

[Equation 5]

From the equation (7), the motor model output estimated value ∧
The velocity can be estimated by subtracting the velocity deviation Δn from n _M ′. This is the basic principle of speed estimation. This principle has no meaning in a continuous system, but in the extremely low speed range where no encoder pulse is obtained during the speed control cycle, (∧n
_M = _{∧n M '-Δn)} wherein the speed estimation from becomes possible. From the above equation (7), the principle diagram of the speed estimation circuit using the minimum-dimensional load observer is constructed as shown in FIG.

(B) Speed estimation observer When an encoder is used as the speed detector, the encoder pulse interval becomes longer than the speed control cycle in the extremely low speed range. The timing chart is shown in FIG. Figure 1
4, the A-phase and B-phase signals of the encoder are two-phase signals, and the combination of the rising and falling edge signals of each phase is the quadruple detection signal. Also, the A-phase rising edge signal is set as the 1f detection signal. (An edge signal other than the A-phase rising edge may be used) Since the two-phase signal of the encoder has an error of about 10% in the duty ratio of the on and off times, the speed detection signal is normally used in the high speed range. As a result, the influence of this error is reduced by using the 1f detection signal. Also, in order to detect speed information quickly in the low speed range,
A quadruple detection signal is used.

In the extremely low speed range as shown in FIG. 14, the latest speed information cannot be obtained at the speed control cycle between encoder pulses. The speed during this period is estimated using the speed estimation principle of FIG. When the principle diagram of the speed estimation observer in FIG. 13 is discretized with the speed control cycle T _S and the speed detection cycle T _P , the speed estimation observer shown in FIG. 15 is obtained. In FIG. 15, since the speed detection value [-n _M (j)] is the average speed between the encoder pulses (j-1) to (j), the motor model output average value [-∧n _M '(i)]. Also, the average value during this period is calculated and the motor model output average value [−
∧n _M '(i)] needs to be obtained. If the speed detection cycle is sufficiently longer than the speed control cycle, the motor model output average value can be expressed by the following equation (8).

[0031]

[Equation 6]

However, when n (j) is small, a more accurate averaging process needs to be examined.
Assuming an extremely low speed range in which (j) is sufficiently large, the motor model output average value is approximated by equation (8).

Next, switching of the observer gain will be described. In the minimum-dimensional load torque observer, the observer gain G ₁ is composed of proportional elements. Therefore, when the load torque τ _L is applied, the average value of the motor model output at the speed detection cycle j point [−∧n _M ' (J)] and the average value [-n _M (j)] of the actual motor speed have a speed deviation Δn (j). This speed deviation Δn (j) can be expressed by the following expression (9) from the expression (4).

[0034]

[Equation 7]

Here, it is assumed that the estimation of the load torque ∧τ _L (j) has already been completed and has converged to a certain value. (J
The average speed between j detected in time j -1), and Setsu換Ru unconditionally G ₁ (j) = G _1H the observer gain from _{G 1 (j-1) =} G 1L, Δn (j) Is constant, the load torque changes. Therefore, although there is no torque change, a transient phenomenon in which a torque change has occurred occurs, causing a disturbance to the speed control system. Therefore, it is necessary to switch the observer gain so that no transient phenomenon occurs.

Next, the switching method will be described. Now, when the observer gain is switched from G _1L to G _1H at time j, the estimation of the load torque has already been completed and a certain value ∧τ _L (j-1)
The load torque does not change during gain switching. Speed deviation Δn from (j-1) to j
It is assumed that (j) is obtained from the equation (9) at time j.

Here, the estimated load torque value ∧τ _L (j-
In order to keep 1) = ∧τ _L (j), gain switching cannot be executed at time j. Therefore, it is considered to delay the gain switching from G _1L to G _1H by one sample to the next speed detection cycle (j + 1). Therefore, at time j, G ₁
(J) = G ₁ (j−1) = G _1L . Now
At time (j + 1), G ₁ (j) = G _1L to G ₁ (j + 1)
Assume that the gain is switched to = G _1H .

Speed deviation Δn (j +
1) is expressed by the following equation (10).

[0039]

[Equation 8]

Estimated torque value ∧τ at time (j + 1)
_To keep _L (j + 1) = ∧τ _L (j), j and (j + 1)
Motor model output ∧n _M '(i)
It suffices to make a correction with the difference Δn _c (j) of the speed deviation shown in the equation 1).

[0041]

[Equation 9]

From the equation (11), Δn (j + 1) is given by the following (1
It is represented by the formula 2).

[0043]

[Equation 10]

Here, the speed deviation Δn (j) is the motor model output ∧n as the load torque estimation error shown in the equation (5).
Since it has already been added to _M ′ (i), the difference in speed deviation may be corrected to the motor model output. Further, since the motor model output is corrected by the difference between the speed deviations, Δn (j + of Expression (10) is used as the speed deviation for calculating the estimated speed value.
1) is used as a predicted value. FIG. 16 is a block diagram of the speed estimation observer in consideration of the gain switching described above.

In FIG. 16, the torque command τ _M * (i)
The load torque estimated value ∧τ _L (j) is supplied to the deviation device 111, and the deviation output thereof is input to the first calculation unit 112. The first calculation unit 112 includes a division unit 112a that divides the speed control cycle Ts by the model machine time constant T _M *, and an adder 112b that adds the output of the division unit 112a and the output of the integrator 112c. It consists of First computing unit 11
Motor model output estimated value calculated by 2 ∧
n _M ′ (i) is input to the second calculation unit 113 that obtains the average value in the pulse interval. The average value of the motor model output calculated by the second calculation unit 113 is supplied to the plus input end of the first deviation unit 114, and the average value of the speed detection outputs detected by the pulse encoder (not shown) is supplied to the minus input end thereof. Supplied.

The deviation output of the first deviation unit 114 is supplied to the observer gain unit 116, where it is multiplied by a predetermined value to obtain the load torque estimated value ∧τ _L (j) at the output. The deviation output of the first deviation unit 114 is supplied to the minus input terminal of the second deviation unit 117. The motor model output estimated value ∧n _M '(i) is supplied to the plus input end of the second deviation unit 117, and the speed estimated value ∧n _M (i) is obtained at its output.

Reference numeral 131 is a speed deviation difference calculation unit, and this speed deviation difference calculation unit 131 is a load torque estimated value ∧τ.
_{When L} (j) is input, the observer gain G ₁ (j
+1) Reciprocal part 132 and this observer gain reciprocal part 1
At 32, estimated load torque ∧τ _L (j) and 1 / G ₁ (j +
The output that integrates 1) and
Deviation unit 133 whose deviation output value is supplied to the minus end
And an adder 134 to which the deviation output value of the deviation unit 133 is supplied to the first input end, and an integrator 135 that integrates the output of the adder 134 and supplies the integrated output to the second input end. The integrated output of the observer gain reciprocal unit 132 is supplied to the negative end of the second deviation device 117, and the output Δn _c ′ (j) [addition of difference (integral element)] of the adder 134 is the same as the output of the adder 112 b. It is added at 136. In FIG. 10, since the load torque estimated value can also be detected by the speed estimation observer, disturbance compensation can be performed by adding this load torque estimated value to the torque command. As a result, the PI element appears equivalently. Therefore, the speed control amplifier has only the proportional element.

FIG. 17 is a block diagram used in the simulation. In FIG. 17, 21 is a motor unit and 2 is a block diagram.
Reference numeral 2 is an encoder model, and 23 is a speed detection unit. The output of the speed detection unit 23 is supplied to the speed estimation observer 24. The speed estimation value from the speed estimation observer 24 is deviated from the speed command value, and the deviation is given to the proportional portion 25. The output of the proportional portion 25 is added to the estimated load torque value from the speed estimation observer 24 to obtain the motor torque. The deviation between this torque and the load torque is supplied to the motor unit 21.

In the simulation shown in FIG. 17, the speed control calculation cycle Ts is set to 1 ms, and the speed detection calculation cycle Tp is executed only when an encoder pulse is input. In addition, the calculation of the motor section is a continuous system. The simulation conditions at this time are shown in Table 1 below.

[0050]

[Table 1]

Simulation results of the gain switching characteristic are shown in FIGS. This is a simulation result when the observer gain is switched after the load torque estimation is completed when the motor is started at an extremely low speed of 1/5000 of the rated speed while the load torque is applied at 1%. FIG. 18 shows the characteristics when the observer gain G ₁ is set to a large value of 100 and started, and the observer gain is not switched. As shown in FIG. 18, the stability of the load torque estimation deteriorates because the observer gain is large, and the speed fluctuation at the time of starting increases.

FIG. 19 shows the case where G ₁ = 50 is set at the time of starting and G1 = 100 is switched without gain switching compensation. In this case, the observer gain is set smaller than that in FIG. , The torque estimation at starting is stable. However, since there is no gain switching compensation, the estimated torque value changes during gain switching, resulting in speed fluctuation.

FIG. 20 shows the result when G ₁ = 50 is set at the time of starting and G ₁ = 100 is switched with the gain switching compensation. From this result, it was found that the stability of torque estimation at the time of starting and gain switching was secured, and the effectiveness of the observer switching method could be confirmed. The speed detection cycle at this time is 25 ms at a speed of 1/5000.

21 and 22, the motor is started at an extremely low speed of 1/2000 of the rated speed while the load torque is applied at 1%, and after the load torque estimation is completed, the load torque is reduced to 3%. in the simulation result when changing, 21 and at time G ₁ = 50 constant, FIG. 22 shows the results when switching to G ₁ = 100 in the state of there gain switching compensation. It was found that the gain switching method improves the disturbance suppression effect.

Next, an embodiment of the present invention in which the above-described speed estimation observer is incorporated into the vector control system of the induction motor will be described. FIG. 1 is an embodiment showing a configuration diagram of the vector control system, and the simulation is executed by replacing the motor portion of FIG. 17 with the vector control system of FIG. Figure 1
In the above, 1 is an induction motor IM, and 2 is a PWM inverter for performing PWM control of IM1. A PWM circuit 3 controls the PWM inverter 2. The PWM circuit 3 is controlled by adding the output from the non-interference calculation unit 4 and the outputs of the d and q axis ACR amplifiers 5 and 6. _{Reference numeral} 7 denotes a slip frequency calculation unit. The slip frequency calculation unit 7 divides the torque command i _1q * by the excitation command i _1d * and the secondary resistance R ₂ by the secondary inductance L ₂ . The multiplication is performed to obtain the slip frequency ωs. Slip frequency ωs
Is added to ωr obtained from the speed detection circuit 8, and the added value ω ₁ is given to the non-interference calculation unit 4 and the PWM circuit 3. 9 is a 3φ → 2φ coordinate conversion unit, and this conversion unit 9 has P
Of the outputs of the WM inverter 2, i _u and i _w are input.

The output of the conversion unit 9 is supplied to the minus ends of the deviations 10 and 11 provided at the inputs of the d and q-axis ACR amplifiers 5 and 6, and the deviations 10 and 11 are provided with i at the plus ends.
_1q * and _i1d * are supplied. The non-interference operation unit 4 corrects the primary resistance R ₁ with a value obtained by multiplying the leakage inductance L _s and the primary inductance L ₁ by ω _1, and outputs the d and q-axis ACR amplifiers 5 and 6 and the adders 12 and 13. It is added and supplied to the PWM circuit 3.

In the embodiment of the vector control system configured as shown in FIG. 1, a simulation was executed by providing a speed estimation observer instead of the speed detection circuit 8. FIG. 2 is a simulation result of a starting characteristic when the speed control system is configured, and this characteristic is a result when the motor is started at a low speed of 1/200 while a load torque of 1% is applied. From Figure 2, speed detection cycle is speed 1
It becomes about 44 ms at / 200. The simulation conditions at this time are shown in Table 2 below.

[0058]

[Table 2]

Good starting characteristics can be obtained by assuming a motor having a maximum speed of about 10,000 rpm and setting the number of pulses per rotation of the speed detection signal to 60 P / R. FIGS. 3 to 11 show simulation results of starting characteristics when the torque control system is used. The simulation conditions at this time are shown in Table 3 below.

[0060]

[Table 3]

FIGS. 3 to 8 show the results when the motor is started with the torque command set to 25 to 35% with the load torque of 20% applied. When there is no speed estimation observer, the actual speed starts increasing by the time the encoder pulse is input, so the slip of the motor becomes smaller and the torque decreases. When the encoder pulse is input and the speed can be detected, the torque recovers to the set value, but the acceleration time becomes very long. When the torque command is increased, the acceleration characteristic is improved, but since the torque does not occur according to the command value, the acceleration time becomes long. On the contrary,
When the speed estimation observer is used, the load torque cannot be estimated until the encoder pulse is detected, so the estimated speed value is estimated to be larger than the actual speed. This causes the output frequency to rise above the required output frequency, resulting in greater motor slippage. As a result, acceleration torque is generated and the vehicle accelerates smoothly. When the encoder pulse is input and the speed can be detected, the load torque estimation is started and approaches the generated torque command value. Further, the acceleration characteristic is good even when the torque command value is small.

9 to 11 are motor model machine time constants T.
Fix _M * = 2s and double the motor mechanical time constant T _M , 1 /
This is the starting characteristic when the value is changed to 2 times or 1/4 times. With the load torque of 20% applied, set the torque command value to 25
Set to% and started the motor. It can be seen that when T _M = 2T _M *, the torque estimation overshoot is large because the speed estimation error until the encoder pulse is input is large, but stable startup is possible. Also, T
_{When M} = 1/2 · T _M * and T _M = 1/4 · T _M *, the inertia error cancels the load torque estimation error until the encoder pulse is input, so the speed estimation error is small. And good starting characteristics are obtained.

From the above, it can be seen that the starting characteristics of the torque control system in the load torque applied state are unlikely to be affected by inertia error. However, in order to obtain even better starting characteristics, it is necessary to establish a load torque compensation method at the time of starting.

[0064]

As described above, according to the present invention,
Since the speed estimation observer is applied to the torque control of the induction motor vector control system, good starting characteristics can be obtained even if a low resolution speed detector is used.

[Brief description of drawings]

FIG. 1 is a structural explanatory view showing an embodiment of the present invention.

FIG. 2 is a starting characteristic diagram of a speed control system.

[Fig. 3] Torque command value 25% without speed estimation observer
Starting characteristic diagram at the time of.

FIG. 4 Torque command value 30% without speed estimation observer
Starting characteristic diagram at the time of.

FIG. 5: Torque command value 35% without speed estimation observer
Starting characteristic diagram at the time of.

Fig. 6 Torque command value 25% with speed estimation observer
Starting characteristic diagram at the time of.

FIG. 7: Torque command value 30% with speed estimation observer
Starting characteristic diagram at the time of.

FIG. 8: Torque command value 35% with speed estimation observer
Starting characteristic diagram at the time of.

FIG. 9 is a starting characteristic diagram when there is an inertia error T _M = 2T _M *.

FIG. 10 is a starting characteristic diagram when there is an inertia error T _M = 1 / 2T _M *.

FIG. 11 is a starting characteristic diagram when there is an inertia error T _M = 1 / 4T _M *.

FIG. 12 is a configuration diagram of a load torque observer.

FIG. 13 is a principle diagram of a speed estimation observer.

FIG. 14 is a timing chart of speed control.

FIG. 15 is a block diagram of a discretized velocity estimation observer.

FIG. 16 is a block configuration diagram of a speed estimation observer in consideration of gain switching.

FIG. 17 is a block diagram of simulation.

FIG. 18 is a characteristic diagram of a simulation result without switching observer gain.

FIG. 19 is a characteristic diagram of simulation results without observer gain switching compensation.

FIG. 20 is a characteristic diagram of simulation results with observer gain switching compensation.

FIG. 21 is a characteristic diagram of a simulation result of a disturbance suppression effect without observer gain switching.

FIG. 22 is a characteristic diagram of a simulation result of a disturbance suppression effect with observer gain switching compensation.

[Explanation of reference numerals] 1 ... Induction motor 2 ... PWM inverter 3 ... PWM circuit 4 ... Non-interference calculation unit 5 ... d-axis ACR amplifier 6 ... q-axis ACR amplifier 7 ... Slip frequency calculation unit

Claims (8)

[Claims] 1. A PWM inverter, an induction motor driven by this PWM inverter, a non-contact low-resolution speed detector that sends the speed of this induction motor as a pulse output, and a minimum-dimensional load torque estimated value observer. Is converted to a discrete system model in the speed control cycle and the speed detection cycle, and a speed estimation observer that estimates the speed at the pulse interval output from the speed detector, and a torque command, an excitation command, and the speed estimation observer are output. And a vector control system that outputs the PWM inverter driving output to the output, and a torque control device for a vector control system of an induction motor. 2. The speed estimation observer includes a first arithmetic unit that obtains a model output estimated value by integrating a deviation between a torque command and a load torque estimated value with an observer model mechanical time constant.
A second arithmetic unit that obtains an average value in the pulse interval from the model output estimated value obtained by the first arithmetic unit; and an average value speed obtained when the pulse output from the second arithmetic unit and the speed detector changes. A first deviation part for calculating the deviation of the first deviation part, an observer gain part for multiplying the deviation value obtained in the first deviation part by the observer gain to obtain the load torque estimated value, and a model output estimated value of the first calculation part. 2. The induction motor according to claim 1, wherein a second deviation unit for obtaining a deviation from the deviation value of the first deviation unit and a deviation value obtained by the second deviation unit are output as a speed estimation value. Vector control system torque controller. 3. The torque control device for a vector control system of an induction motor according to claim 2, wherein the second arithmetic unit comprises means for correcting a timing deviation between a speed control cycle and a speed detection cycle. . 4. A speed deviation difference calculation unit calculates a difference in speed deviation from the load torque estimated value and the output of the first deviation unit,
The calculated value is added as a correction value to the model output estimated value obtained by the first calculation unit, and the added value is supplied to a second calculation unit that obtains an average value in the pulse interval. A torque control device for a vector control system of an induction motor according to the description. 5. The observer gain reciprocal unit, which is switched at the next sample time when the estimated load torque value is input and the integrated value with the estimated load torque value is obtained at the output, and the reciprocal unit thereof. Of the output of the first deviation unit and a deviation unit that obtains a speed deviation from the output of the first deviation unit, and an integrator that obtains a correction value at the output by integrating the speed deviation difference obtained from this deviation unit. 5. The torque control device for a vector control system of an induction motor according to claim 4, wherein the output of the above is supplied to the second deviation portion, and the estimated speed value is sent to the output. 6. The model output estimated value is corrected by adding it only to the next sample to an integrator that obtains a model output estimated value by integrating the velocity deviation difference obtained by the deviation unit with an observer model mechanical time constant. The torque control device for a vector control system of an induction motor according to claim 4. 7. The torque control device for a vector control system of an induction motor according to claim 4, 5 or 6, wherein the observer gain section is switched between two stages above and below a certain speed. . 8. The torque control device for a vector control system of an induction motor according to claim 7, wherein the observer gain section is an integrated value of a fixed gain, a constant and a compensation gain.