JP5569722B2 - Motor control device - Google Patents

Description

  The present invention relates to a motor control device, and more particularly to a technique for estimating a mechanical constant such as a moment of inertia of a mechanical system including a motor and a load.

In order to make the motor speed control highly responsive, information on mechanical constants including the moment of inertia is required. For this reason, conventionally, various techniques for estimating the mechanical constant using an inverter that drives a motor have been provided.
For example, Patent Document 1 discloses a technique for estimating an inertia moment and a friction torque by a least square approximation method using a current command value for acceleration / deceleration, a speed and acceleration of a mechanical system.
Patent Document 2 discloses a technique for estimating a moment of inertia and a viscous friction coefficient by a sequential least square approximation method using a torque command differential value, an acceleration, and a jerk which is a differential value of acceleration.

Japanese Patent No. 3545487 (paragraphs [0008] to [0016], FIG. 1, etc.) JP 2006-217729 A (paragraphs [0036] to [0050], [0060] to [0064], FIG. 1 and the like)

  When the load of the motor is a fluid machine such as a fan or a pump, the load torque often has a square reduction load characteristic in which the load torque is proportional to the square of the speed. However, since the techniques disclosed in Patent Document 1 and Patent Document 2 do not consider a square reduction load as a mechanical system model, when these conventional techniques are applied to driving of a fluid machine, the machine constant is estimated. An error may occur.

On the other hand, the motor control apparatus may employ sensorless vector control in which the motor speed is estimated from the motor current and voltage, and the motor is operated using the estimated speed value. In this case, a delay occurs in speed detection due to the motor control principle, and an error caused by this delay may be included in the estimated machine constant value.
Furthermore, the sequential least square approximation method described in Patent Document 2 may overflow the variables in the control device depending on the operating conditions and the set values of the control constants.

  SUMMARY OF THE INVENTION Accordingly, the problem to be solved by the present invention is to provide a motor control device capable of estimating a machine constant with high accuracy regardless of the characteristics of the load, and further eliminating inconveniences in calculation such as overflow. .

In order to solve the above-mentioned problem, a motor control device according to claim 1 is a motor control device for driving a load whose load torque changes depending on speed by a motor.
A mechanical system model that simulates the mechanical system including the load is represented by the following mathematical formula 1.
Means for passing a torque equivalent value of the mechanical system through a low-pass filter to calculate a second torque equivalent value;
Moment of inertia J _m as a mechanical constant of the mechanical system model, speed coefficient D _mk (k = 1, 2,..., P, P is a positive integer), and load torque τ _L0 independent of speed Means for estimating the second torque equivalent value, the acceleration equivalent value and the speed equivalent value of the mechanical system by a sequential least square approximation method ,
The time constant of the low-pass filter is a value corresponding to the delay of the speed detection value as the speed equivalent value .

  A motor control device according to a second aspect is the motor control device according to the first aspect, wherein P = 2 in Formula 1.

The motor control device according to claim 3, Oite to the motor control device according to claim 1 or 2, the estimated gain matrix in the successive shape least square approximation method, once per cycle of the acceleration and deceleration operation of the motor , Means for initializing .

According to the first aspect of the present invention, it is possible to accurately estimate a mechanical constant such as a moment of inertia for a load whose load torque varies depending on speed, such as a square reduction load. Further , even when there is a large delay in speed detection as in sensorless vector control, the mechanical constant can be accurately estimated.
According to the invention of claim 2, the machine constant of the square reduction load can be estimated by a simpler calculation than that of claim 1.
According to the third aspect of the present invention, it is possible to prevent an overflow of a variable in the control device and to further speed up the convergence of the machine constant estimated value.

It is a block diagram of the control apparatus which concerns on the basic form of this invention. It is a block diagram of a control device concerning a 1st embodiment of the present invention. It is explanatory drawing of the timing which initializes an estimated gain matrix in each embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
First, FIG. 1 shows a block diagram of a control device according to the basic form of the present invention. In FIG. 1, the deviation between the speed command value ω _m ^* and the speed detection value ω _mdet of the mechanical system model 20 detected by the speed detection circuit 14 is calculated by the subtractor 11, and this deviation is amplified by the speed regulator 12. Calculate torque command value τ _m ^* . The mechanical system model 20 is a transfer function between a motor and a load mechanical system.
The torque control means 13 controls the motor torque τ _m so as to coincide with the torque command value τ _m ^* , and controls the motor current by an inverter (not shown). As a result, the output torque is set to a desired value. Functions to control.
Through the control as described above, the speed ω _m of the motor (mechanical system model 20) can be controlled to the speed command value ω _m ^* .

Further, the machine constant estimation unit 30 _obtains a machine constant estimated value vector Θ _est from the speed detection value ω _mdet and the torque command value τ _m ^* .
Hereinafter, Examples 1 and 2 of the machine constant estimation unit 30 will be described.

First, the calculation of the machine constant estimation unit 30 in the first embodiment will be described. The first embodiment corresponds to the invention according to claim 1.
When the magnitude of the load torque depends on the speed, the mechanical system model 20 can be expressed by a mathematical model such as the mathematical formula 1.

The machine constant estimation unit 30 uses the moment of inertia J _m , the speed coefficient D _mk , and the load torque τ _L0 that does not depend on the speed as the machine constant in Equation 1, the torque detection value τ _mdet , the acceleration detection value a _mdet, and the speed detection. The value is estimated from the value ω _mdet by the iterative least square approximation method.
First, the machine constant estimated value vector Θ _est , the signal vector z, and the output y are defined by Equation 2. In addition, T in the following numerical formula shows transposition.

The torque detection value τ _mdet in Equation 2 is substituted by the torque command value τ _m ^{* on} the premise that the torque control by the torque control means 13 of FIG. The torque detection value τ _mdet may be calculated from an output obtained by passing the torque command value τ _m ^* through a low-pass filter and simulating the control delay of the torque control means 13. The calculation may be performed using the detected current value of the motor and the electric constant of the motor.

The acceleration detection value a _mdet in Equation 2 is calculated by differentiating the speed detection value ω _mdet .
Next, the signal vector z and the output y in Expression 2 are normalized by Expression 3.

Further, the machine constant estimated value vector Θ _est in Equation 2 is obtained by using, for example, “adaptive control” (by Takashi Suzuki, Hyundai Control Series 7, p. 7) using the normalized signal vector z _N and the normalized output y _N. 87-p.90, August 2001, issued by Corona Inc.), based on arithmetic expressions of the recursive least square approximation method (e.g., equations (4.89)-(4.91) on p.89)) Thus, the calculation can be performed using Equation 4.

Alternatively, the machine constant estimated value vector Θ _est in Equation 2 is expressed as “Kalman filter and adaptive signal processing” (by Takashi Tanizaki, Digital Signal Processing Library 5, p. 81-p. 85, December 2005, issued by Corona). ) May be calculated by Formula 5 based on the calculation formula of the sequential least square approximation method described in () (Formulas (2.73) to (2.76b) on page 83).

  By the arithmetic processing described above, the machine constant can be accurately estimated for all loads including a square reduction load.

In the second embodiment, the mechanical constant of the square reduction load is estimated by a simpler calculation than in the first embodiment, and corresponds to the invention according to claim 2.
The mechanical system model 20 in the case of a square reduction load can be expressed by Expression 6 by setting P = 2 in Expression 1.

In this case, the machine constant estimated value vector Θ _est , the signal vector z, and the output y are defined as in Equation 7, and similarly to the first embodiment, using Equations 3 and 4 or Equations 3 and 5, Estimate vector Θ _est is estimated.

Next, FIG. 2 shows a block diagram of the control apparatus according to the first embodiment of the present invention.
This embodiment corresponds to the invention according to claim 1 and is configured so that the mechanical constant can be accurately estimated even when there is a large delay in speed detection as in sensorless vector control.

In FIG. 2, the speed detection delay compensator 31 is constituted by a low-pass filter, and the filter time constant is set to a value corresponding to the delay of the speed detection circuit 14. As a result, the delay time from the speed ω _m to the speed detection value ω _mdet can be made equal to the delay time from the torque τ _m to the torque detection value τ _mdet , and the machine constant can be estimated accurately.
In addition, since the calculation content by the machine constant estimation part 30 is the same as that of the basic form mentioned above , description is omitted here.

Next, the second embodiment of the present invention is for preventing the overflow of variables in the control device and speeding up the convergence of the machine constant estimated value in each of the above-described embodiments. The second embodiment corresponds to the invention according to claim 3 .

First, a case where the sequential least square approximation method is realized by the above-described Expression 4 will be described.
In Formula 4, forgetting coefficient σ is usually set to a value of zero or more (for example, 0 to 0.01).
When the forgetting factor σ is set to a positive value, as is clear from the equation of the estimated gain (adaptive gain) matrix Γ in Equation 4, when an element of the normalized signal vector z _N becomes zero, the estimated gain matrix Γ is Increase function. For this reason, when this calculation is realized by the control device, the estimated gain matrix Γ may overflow.

On the other hand, when the forgetting factor σ is set to zero, the estimated gain matrix Γ becomes a decreasing function, so there is no possibility that the estimated gain overflows. Therefore, in this embodiment, the forgetting factor σ is set to zero.
However, when the forgetting factor σ is set to zero, the estimated gain matrix Γ decreases with time, so that the convergence of the machine constant estimated value is delayed. Therefore, in the present embodiment, as will be described below, the estimated gain matrix Γ is initialized at regular intervals to prevent the estimated gain matrix Γ from becoming small, and the convergence of the machine constant estimated value is accelerated.

FIG. 3 shows the timing for initializing the estimated gain matrix Γ.
In each of the embodiments described above, the acceleration / deceleration operation of the motor is repeated in order to estimate the mechanical constant. In order to accurately estimate the mechanical constant, a period of one acceleration / deceleration operation is required. Therefore, as shown in FIG. 3, the initialization command is set to “1” for each cycle of the acceleration / deceleration operation, and the estimated gain matrix Γ is initialized.

As another application example of the second embodiment, a case in which the sequential least square approximation method is realized by the above-described Expression 5 will be described.
In Equation 5, the forgetting factor λ is normally set to a positive value of 1 or less (for example, 0.95 to 1.0).
If you set the forgetting factor λ less than 1, as is clear from equation 5, the estimated gain matrix Q certain elements of the normalized signal vector z _N is zero becomes an increasing function, there is a risk of overflow.
On the other hand, when the forgetting factor λ is set to 1, the estimated gain matrix Q becomes a decreasing function, so there is no fear of overflow. Therefore, in this embodiment, the forgetting factor λ is set to 1. Further, as in the case of the estimated gain matrix Γ, it is desirable to prevent the estimated gain matrix Q from becoming small by initializing the estimated gain matrix Q at every fixed period and to speed up the convergence of the machine constant estimated value.

  In the present invention, not only the detected values of torque, speed, and acceleration described in each embodiment, but also the calculated values (estimated values) and command values are used to calculate machine constants by a sequential least square approximation method. It may be estimated.

DESCRIPTION OF SYMBOLS 11 Subtractor 12 Speed regulator 13 Torque control means 14 Speed detection circuit 20 Mechanical system model 30 Machine constant estimation part 31 Speed detection delay compensator

Claims (3)

In a motor controller for driving a load whose load torque varies depending on the speed by a motor,
A mechanical system model that simulates the mechanical system including the load is represented by the following mathematical formula 1.
Means for passing a torque equivalent value of the mechanical system through a low-pass filter to calculate a second torque equivalent value;
The following J _m , D _mk and τ _L0 as machine constants of the mechanical system model are estimated by the successive least square approximation method using the second torque equivalent value, the acceleration equivalent value and the speed equivalent value of the mechanical system. Means to
Equipped with a,
A motor control device characterized in that a time constant of the low-pass filter is a value corresponding to a delay of a speed detection value as the speed equivalent value .

In the motor control device according to claim 1,
A motor control device characterized in that P = 2 in Formula 1 above. In the motor control device according to claim 1 or 2,
A motor control apparatus comprising: means for initializing an estimated gain matrix in the successive least squares approximation method once in one cycle of acceleration / deceleration operation of the motor.

Images