JP3336791B2 - Motor load characteristic identification device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motor control device for driving a mechanical load, and more particularly to a motor load characteristic identifying device suitable for identifying a parameter representing a vibration characteristic of a mechanical load.

[0002]

2. Description of the Related Art Conventionally, in the control of a mechanical load drive motor having vibration characteristics, as a method of identifying a parameter representing the vibration characteristic, the vibration characteristic is modeled by a two-inertia system, and the parameter is defined by the motor generated torque and the motor speed. A method of identifying from a detected value is known. The Euro in 1993
"Self-tuning State Control for a Poo" announced at the pean Power Electronics Conference (EPE'93)
rly Damped MechanicalSystem ”(author: N.KlaaBen, pp.
According to 394-398), a parameter of a two-inertia system in which a motor and a mechanical load are connected by axial torsional rigidity is calculated by identifying a coefficient of a transfer function of a motor speed with respect to a motor generated torque. The transfer function is described as a relationship between a data train at each sampling point for detecting the motor generated torque and the motor speed. That is, a data sequence τ _m (z) of the motor generated torque and a data sequence ω _m (z) of the motor speed
Is represented by the following equation.

[0003]

(Equation 1)

At this time, a transfer function G (z) is defined by the following equation.

[0005]

(Equation 2)

By calculating the coefficients a _n and b _n , it is possible to identify the two inertial system parameters of the motor and the mechanical load (the moment of inertia of the motor and the mechanical load, the torsional rigidity of the shaft, etc.).

Here, the coefficients a _n and b _n of the transfer function are calculated by using the a _n and b _n calculated from the detected N data strings.
This is performed using a systematic method so that the error between the actual value and the actual value is minimized. This is well known as a method called the least squares method, and is described by Ichinose in "Error Theory" (First Edition, Sho 28
Year, Baifukan) pp. 38 to 54 describe the calculation method.

[0008]

In a mechanical load drive system using a motor, the speed of the motor varies not only with the torque generated by the motor but also with the load torque acting on the mechanical load. Therefore, in order to calculate the transfer function of the motor speed with respect to the motor generated torque, it is necessary to correctly model the characteristics of the load torque acting at this time.

However, in this known example, the effect of the load torque on the motor speed detection value is modeled by including the measurement error in the motor speed detection. Assuming that noise having no correlation in time series is superimposed on the detected motor speed value due to this measurement error, the coefficient of the transfer function of the motor speed with respect to the motor generated torque is calculated by the least square method. For this reason, as seen in a mechanical load drive system using an actual motor, in a system in which disturbance due to load torque acts due to gear play or non-linear friction of a mechanical part during parameter identification, the vibration characteristics of the mechanical system cannot be correctly identified. .

An object of the present invention is to correctly identify parameters representing the vibration of a mechanical system even when a load torque acts on the mechanical system.

Further, according to this known example, a transfer function of a motor speed with respect to a motor generated torque is calculated as a relational expression of a data sequence at each sampling time. Therefore, the identified coefficient is calculated as a value depending on the sampling period. In order to identify a physical quantity representing the vibration characteristic of the mechanical system, it is necessary to convert the calculated coefficient in consideration of the sampling time, and if the sampling time is not set appropriately, the identification accuracy of the vibration characteristic deteriorates.

An object of the present invention is to accurately identify vibration characteristics by directly calculating a transfer function from a motor generated torque to a motor speed without depending on a sampling time.

[0013]

In order to achieve the above object,
The effect of the load torque acting on the mechanical system on the motor speed detection value is represented by a time function, and taking this into account, a transfer function of the motor speed with respect to the motor generated torque is identified.

Further, the time integral value of the motor generated torque and the motor speed is used as a new variable, and the coefficient of the transfer function is identified by the relational expression.

[0015]

By modeling the effect of the load torque on the detected motor speed as a time function, the vibration characteristics of the mechanical system can be identified separately from the measurement error included in the detected speed. As a result, even when the backlash of the gear or the non-linear friction acts, the vibration characteristics can be correctly identified.

Further, by using a value obtained by integrating the value instead of an instantaneous data sequence for each sampling as a new variable, the coefficient of the transfer function can be calculated without being affected by the sampling time. Physical parameters (moment of inertia, torsional rigidity of the shaft, etc.) representing the vibration characteristics can be identified well. Further, by using the integral value, parameter identification that is less susceptible to measurement errors can be achieved as compared with the case where calculation is performed from a data string of instantaneous values.

[0017]

An embodiment of the present invention will be described with reference to FIG. A mechanical load drive system using a motor includes a motor 1, a mechanical load 2, and a connecting shaft 3 for both, and the motor 1 and the mechanical load 2 vibrate due to the torsion of the connecting shaft. The physical parameters identified as load characteristics are the motor 1 and the mechanical load 2
, The torsional rigidity of the connecting shaft 3 and the viscosity coefficient.

The motor 1 is supplied with electric power by the power converter 4 to generate a driving torque. Here, the power converter 4
Torque current i _t flowing through the motor 1 by the current detector 5
It is detected by, the power converter 4 in accordance with the desired torque current command value i _r to the motor 1, and the current control system is provided to flow the torque current i _t. Further, a speed detector 6 is attached to the motor 1, and the motor speed ω
_{Find m} . Motor speed command generator 7 generates a velocity command omega _r for the motor drive system. Motor speed controller 8, the speed command value omega _r and the motor speed detected value omega _m, with a motor torque current detection value i _t, so that the motor speed omega _m is equal to the command value omega _r, power converter calculating a torque current command value i _r for 4.

Now, for such a motor drive system,
A motor load characteristic identification device 9 is provided. The identification device 9
Using the detection value of the motor torque current i _t and the motor speed omega _m, to identify the parameter value when modeling the vibration characteristics of the motor drive system with two-inertia system. Identification device 9, the coefficient multiplier 901 to the motor torque current detection value i _t 3
One of the integrators 902, 903, 904, and two integrators for the motor speed detected value omega _m 905 and 906, in addition, simulating the effect of the load torque tau _d acting on the mechanical load 2 has on the motor speed detected value A load torque simulating means 907 is provided.
X ₁ , x ₂ , x ₃ , y ₁ , y ₂ which are the outputs of the integrators and δ which is the output from the load torque simulating means 907
Using (t), an operation for identifying the parameter value is executed.

Next, a description will be given of the vibration characteristics of the motor drive system.
FIG. 2 shows a block diagram of the inertial system model. Where J
₁ is a motor inertia values, the moment of inertia value J ₂ mechanical load, K _s is the axial torsional rigidity, C _s represents a viscosity coefficient, S is meant Laplace operator. Further, τ _m represents the motor generated torque, ω _m represents the motor speed, and ω _l represents the speed on the mechanical load side. In the identification device 9, the motor torque current i
motor generated torque τ _m proportional to _t and motor speed ω _m
, J ₁ , J ₂ , K _s , and C _s are identified.

[0021] At this time, transfer coefficient of the motor speed omega _m motor torque tau _m, the load torque tau _d is given by 2 inertia model in FIG. 2 as follows.

[0022]

[Equation 3]

Here, ω _n represents the resonance angular frequency, ω _z represents the anti-resonance angular frequency, and ζ _n and ζ _z correspond to the damping factors at the resonance and anti-resonance frequencies, respectively. These have the following equation with the two inertial system parameters J ₁ , J ₂ , K _s and C _s .

[0024]

(Equation 4)

In the present invention, the motor generated torque τ _m
Identifying parameter as a transfer function of the motor speed omega _m for. From equation (3), the transfer function F (s) is as follows.

[0026]

(Equation 5)

Here, each coefficient q ₁ , q ₂ , p ₀ , p ₁ ,
The following relationship exists between p ₂ and the physical parameters.

[0028]

(Equation 6)

Therefore, by identifying the coefficients of the transfer function of the equation (5), the two inertial system parameter values J ₁ , J ₂ , K
_s, C _s can calculated from equation (6).

Now, according to the present invention, the equation (5) is transformed into
This is expressed as the following equation.

[0031]

(Equation 7)

Here, 1 / s represents integration, and 1 / s ² ,
1 / s ³ corresponds to double and triple integration, respectively. Using equation (7), equation (3) is expressed by the following equation.

[0033]

(Equation 8)

When this relationship is expanded, the following relationship is obtained.

[0035]

(Equation 9)

Here, δ (t) represents the effect of the load torque τ _d on the motor speed ω _m . The integrated values of τ _m and ω _m are expressed as x ₁ , x ₂ , x ₃ , y ₁ , y
_{Assume 2} .

[0037]

(Equation 10)

From this, the relational expression to be identified is obtained as follows.

[0039]

Y (t) =-q1y2 (t) -q2y1 (t) + p0x3 (t) + p1x2 (t) + p2x1 (t) + δ (t) + e (t) (11) where e (t) Represents an error associated with the identification, and identifies each coefficient so as to minimize the error. That is, the coefficients are identified by the identification calculator 908 in FIG. 1 based on this relational expression.

Next, the effect δ (t) of the load torque τ _d on the motor speed is modeled as the following time function.

[0041]

(Equation 12)

Here, c = (c ₁ c ₂ ... C _ν ) is a coefficient of a ν-order time function. By newly introducing this load torque characteristic model, the mechanical system in which load torque disturbance acts , The parameters of the vibration model,
J ₁ , J ₂ , K _s , and C _s are identified.

The relational expression (11) to be identified by modeling δ (t) as shown in expression (12) can be described by the following expression.

[0044]

(Equation 13)

Based on this relational expression, the coefficient of the expression (13) can be identified according to the well-known least squares method. FIG. 3 shows a block diagram based on the least squares method.
Thus, the coefficients q ₁ , q ₂ , p ₀ , p ₁ , p ₂ , and c ₁ , c are set such that the square error e ² shown in the following equation is minimized.
₂ , ..., _cv are identified.

[0046]

[Equation 14]

As described above, according to this embodiment,
The influence of load torque disturbance tau _d, which has been neglected in the prior art, taking into account as ν next time function, since the coefficients of the transfer function from the motor torque to the motor speed identified by the least squares method, accurately The two inertial vibration parameters can be identified. Further, by modifying the relational expression of the transfer function of the motor speed with respect to the motor generated torque, the motor generated torque,
Since the coefficient is identified as a relational expression of each integral value of the motor speed, there is an advantage that identification can be performed without being affected by measurement errors and noise as compared with the conventional identification method based on a data string of instantaneous values at each sampling time. .

As a second embodiment of the present invention, (1
A method of identifying the relational expression of the expression 3) by repeated calculation will be described below. The block diagram is shown in FIG. First, the equation (14), (13) a solved by the least square method type, the results of Motoma' and _{N i-1, D i-} 1. The variables corresponding to this are x ＾ and y ＾. Next, at the i-th point, x ＾, y
N _i and D _i are obtained from ＾. By repeating these as i = 1, 2,..., N and D can be calculated more accurately. The relational expression can be expressed by the following expression.

[0049]

(Equation 15)

Where i represents the number of repetitions,
Identification accuracy can be improved by increasing i.

As described above, according to this embodiment, since the least square method is repeated to improve the identification accuracy, there is an advantage that the two inertial system parameters can be identified with higher accuracy.

[0052]

As described above in detail, according to the present invention, the effect of a load torque component acting as a disturbance torque on a mechanical system driven by a motor is modeled by a time function. The coefficient of the transfer function to the motor speed with respect to the generated torque can be identified. Therefore, even when a disturbance load torque acts during parameter identification by giving a torque current command value to the motor, the coefficient of the transfer function can be identified separately from the influence of this. This coefficient is a physical parameter value in the two inertial vibration model, J ₁ , J ₂ ,
Since they correspond to K _s and C _s , there is an effect that parameters can be accurately identified from these coefficient identification values.

Further, according to the present invention, since the coefficient of the transfer function is identified by converting it into the relation between the integral values of the motor generated torque and the motor speed, the coefficient identification is executed from the relational expression of the instantaneous value at each sampling time. Compared with the method of the example, parameter identification in which the influence of measurement noise included in the detected value of the motor speed or the motor generated torque is eliminated can be achieved. As a result, identification of vibration characteristics can be performed with higher accuracy than in the conventional method.

Using these parameter identification values,
By controlling a motor mechanical system having vibration characteristics, vibration can be suppressed and motor control with good response characteristics can be achieved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a motor load characteristic identification device according to an embodiment of the present invention.

FIG. 2 is a two-inertia model of a motor load characteristic which is an example of a control target of the present invention.

FIG. 3 is a block diagram of coefficient identification by the least squares method of the present invention.

FIG. 4 is a block diagram of a least-squares method according to another embodiment of the present invention;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Motor, 2 ... Mechanical load, 4 ... Power converter, 5 ... Current detector, 6 ... Speed detector, 7 ... Motor speed command generator, 8 ... Motor speed control device, 9 ... Motor Load characteristic identification device, 901: coefficient multiplier, 902 to 906, integrator, 907: load torque simulation means, 908: identification arithmetic unit.

Continuation of the front page (56) References JP-A-5-176580 (JP, A) JP-A-6-78579 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01R 31 / 34 H02P 5/00

Claims (2)

(57) [Claims] 1. A based on the torque current i _t and the motor speed omega _m
Controlled motor and mechanical load connected to the motor
In motor load characteristic identification apparatus for driving the device, the motor load characteristic identification unit receives a means for detecting the torque current i _t of the motor, the detection signal of the means for detecting the motor speed omega _m,
Means for calculating a motor torque tau _m from the motor torque current i _t, the calculation value of the motor torque tau _m, to calculating an integral value
A first integral calculating means, and a second integral calculating means for calculating a double integral value .
And integral calculation unit, a third integral calculation for calculating the triple integral value
Means and, of the motor speed detection value omega _m, fourth for calculating an integral value
A fifth calculating means for calculating a double integral value;
Comprising of an integral calculating means, and a load torque simulating means is the integral value of the motor torque, the first integral calculation
An integral value of the means, an integral value of the second integral operation means,
An integral value of the third integral operation means, an integral value of the motor speed, an integral value of the third integral operation means,
And the load torque simulating means.
Of load torque, which is the output of motor, on motor speed δ
(T) and enter the, by the least squares method, to identify the coefficients of the transfer function of the motor speed to the motor torque, the motor inertia value J ₁ is a 2-inertia system parameters of the motor mechanical system on the basis of the identification results , the identifying the machine load inertia moment value J _2, a shaft torsional stiffness K _s
A motor load characteristic identification device, comprising: a constant operation unit . 2. A control constant of a motor speed control device according to claim 1, wherein the identified two inertial system parameter values are used.
A motor load characteristic identification device characterized by adjusting the following .