JP2000148210A - Gain calculating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To actualize technology which can calculate a gain giving excellent response and stability to the motion of an operation part when, for example, an industrial robot has an easy-to-deform member (low-rigidity member) interposed between an actuator and the operation part. SOLUTION: A mechanical-electric system comprises a mechanical system 10 having an arm 14 connected to a motor 11 through a speed reducer 13, an electronic controller 20 which performs feedback control over the mechanical system 10, a gain calculating device 30 which calculates the gain used for the electric controller 20, etc. The gain calculating device 30 has an electric system mathematical expression model 31a mathematically equivalent to the electric controller 20 and a mechanical system mathematical expression model 31b mathematically equivalent to the mechanical system 10, calculates an evaluated value 32a by a calculating circuit 32 according to an operation part motion value 31d obtained from them, and evaluate the evaluated value 32a by a search circuit 33 to calculate a proper gain value. This proper value is transferred to the electronic system controller 20 to control the mechanical system 10.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for calculating an appropriate gain value used in a control device for controlling the movement of an electromechanical system such as an industrial robot. The electromechanical system referred to in this specification refers to a system in which a mechanical system and an electric system are combined. For example, a robot body (mechanical system) that moves an operation unit such as a hand by an actuator and an electric system that controls the operation of the actuator. It refers to a system that is combined with a control device (electrical system) to form one system.

[0002]

2. Description of the Related Art A method of calculating an appropriate value of a gain used in an electric control device of the above-mentioned electromechanical system is disclosed in Japanese Patent Application Laid-Open No. Hei 8-1620.
No. 5 publication. This publication discloses an electric control device of an electromechanical system that controls the motion of a mechanical system that transmits the motion of an actuator such as a motor to an operation unit such as a printer head or a robot hand using an electric actuator control device. Describes a technique for finding an appropriate value of the gain used in. According to the technology described in this publication, an electrical mathematical model that provides an equivalent relationship to the relationship between the input value and the output value of the electrical control device, and a relationship equivalent to the relationship between the power applied to the actuator and the motion generated on the output shaft of the actuator are described. Using the resulting mechanical mathematical model, simulation is performed while changing the gain in the electrical mathematical model, and a value of the gain that provides a suitable response and stability to the output shaft of the actuator is searched for.

[0003]

The above-described gain calculation technique is extremely effective in calculating an appropriate value of a gain that provides a suitable response and stability to an output shaft of an actuator. However, if a member having low rigidity is interposed between the actuator and the operation unit, the response and stability required in the operation unit cannot be obtained even if the optimum gain is obtained by the above technique. This is because the deformation of the low rigidity member causes the movement of the output shaft of the actuator and the movement of the operation unit not to match, and even if the output shaft of the actuator has good responsiveness and stability, This is because another movement occurs in the operation unit. For example, in the case of an industrial robot, usually, a robot arm (hereinafter, referred to as an arm) is attached to an output shaft of an actuator (a motor or the like) via a power transmission member (a gear train or a belt). In this case, the power transmission member is easily deformed, and a deviation occurs between the rotation angle of the output shaft of the actuator and the rotation angle of the arm by the amount of the deformation, realizing good responsiveness and stability of the rotation angle of the actuator. However, the movement of the operation unit (hand) at the tip of the arm does not correspond to the movement of the actuator, and the response and stability generated in the operation unit may be insufficient. Such a phenomenon is a very serious problem in controlling a robot or the like. The problem to be solved by the present invention is the above-mentioned problem. When an easily deformable member (a low-rigidity member) is interposed between the actuator and the operation unit, good responsiveness to the movement of the operation unit is obtained. It is to realize a technique capable of calculating a gain that brings about stability.

[0004] Two methods can be considered to control the operation section of the mechanical system in which the low rigidity member is interposed. First
Is a method of performing feedback control by detecting the current value such as the position or speed of the operation unit instead of detecting the current value such as the rotation angle or rotation speed of the actuator and performing feedback control (this is This is a method of detecting the object to be controlled, in this case, the current value of the operation unit, and performing feedback control, which can be referred to as closed-loop feedback control). If the technique described in JP-A-16205 is used, a gain value that provides the required responsiveness and stability to the operation unit will be calculated. However, for this purpose, it is necessary to set sensors for detecting the position and speed on the operation unit. Normal actuators often include a sensor such as an encoder, and the current value of the actuator is easy to detect, but when trying to detect the current value of the operation unit,
There is a problem that new sensors need to be set. Further, the sensor for detecting the current value of the actuator is originally wired, but if the current value sensor is attached to the operation unit, it is necessary to newly provide a wiring for the current value sensor, which is often troublesome.

The second method is a method of detecting the current value of the actuator and performing feedback control regardless of the deformation of the low-rigidity member (this method detects the amount of deformation of the low-rigidity member and reflects this in the feedback control). Therefore, the feedback control loop is not closed, but the amount of deformation of the low-rigidity member is small, and the current value of the actuator and the current value of the operation unit do not differ so much. In other words, it is a technology that secures the responsiveness and stability of the operation unit while maintaining the semi-closed loop feedback control. If this can be realized, it is not necessary to attach a sensor to the operation unit, and the wiring will be troublesome. I do not need it.

An object of the present invention is to develop a feasible technique capable of accurately controlling the movement of an operation unit while maintaining a semi-closed loop feedback control system.

[0007]

According to a first aspect of the present invention, there is provided a mechanical system in which a low-rigidity member is interposed between an actuator and an operation unit, and a deviation between a current value and a target value of the actuator. The present invention relates to a device for calculating an appropriate value of the gain in an electromechanical system including an electric actuator control device for controlling a current value of the actuator to be the target value based on a value obtained by multiplying the multiplied value by the multiplied value. This gain calculation device is generated in a mechanical mathematical model that realizes a mathematically equivalent relationship to the relationship established between the power applied to the actuator and the motion generated in the operating unit, and in the operating unit obtained from the mechanical mathematical model. A circuit for calculating an evaluation value for evaluating the exercise; and a circuit for searching for an appropriate value of the gain from a change in the evaluation value calculated by the calculation circuit. By using the gain calculating device according to the first aspect, it is possible to compensate for the deformation generated in the low-rigidity member and to highly evaluate the motion generated in the operation unit as a result, while performing the feedback control of the semi-closed loop type. it can. That is, an appropriate gain is calculated so that the motion generated in the operation unit satisfies the desired responsiveness and stability. When the actuator is feedback-controlled using the gain calculated in this way, the semi-closed loop feedback control that does not detect the amount of deformation occurring in the low-rigidity member, but the low-rigidity member is used in calculating the target value given to the actuator. It is not necessary to consider the amount of deformation that occurs, and only by using the target value calculated on the assumption that there is no deformation, as a result, it is possible to control the motion of the operation unit affected by the deformation to a value that can be highly evaluated.

The value corresponding to the power to be applied to the actuator input to the mechanical mathematical model can be obtained by calculation, or can be actually obtained without calculation.
The invention according to claim 2 adopts the former method, wherein a relationship established between a current value and a target value (the two of which are at least input values) and an output value of the actuator input to the electric control device. Further comprising an electrical mathematical formula model that realizes a mathematically equivalent relationship, inputting the output value of the electrical mathematical formula model to the mechanical mathematical formula model, and outputting the output value of the mechanical mathematical formula model to the calculation circuit. input. According to this method, the gain is adjusted to an appropriate value before the electric actuator control device is actually manufactured.
It is possible to verify whether or not the motion generated in the operation unit can be controlled so as to obtain a high evaluation, and if it turns out that it is rejected, it can be known in advance that a satisfactory product cannot be obtained even if it is manufactured as it is I can do it. In this case, for example, measures such as increasing rigidity by replacing parts of low-rigidity members or changing control processing contents scheduled by the control device can be performed at an early stage.

Further, the invention according to claim 3 corresponds to a method of actually obtaining a value corresponding to the power to be applied to the actuator, which is input to the mechanical mathematical model, without using a calculation. The output value of the actuator control device is input to the mechanical mathematical model, and the output value of the mechanical mathematical model is input to the calculation circuit. According to this method, it is possible to easily calculate an appropriate value of the gain even when the relationship between the input value and the output value of the electric control device is very complicated and a mathematical model cannot be easily obtained. Will be possible.

[0010] The important factors of the movement characteristics of the operation unit depend on the use environment. According to a fourth aspect of the present invention, an evaluation can be correctly performed in accordance with a use environment, and a calculation circuit calculates an evaluation value by adding a value obtained by multiplying a plurality of evaluation element values by respective weights. , Each weight can be increased or decreased. According to the present invention, for example, when responsiveness is emphasized, a large weight is given to the index indicating the responsiveness, and when positioning accuracy is emphasized, evaluation with a large weight given to the index indicating the positioning accuracy can be performed. Become.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a schematic conceptual diagram of this embodiment. FIGS. 2 and 3 show partial conceptual diagrams of the conceptual diagram shown in FIG. This embodiment allows movement of the actuator 122 to be controlled by the low rigidity member 124.
The mechanical system 120 transmitted to the operation unit 126 through the control unit satisfactorily is controlled. The low rigidity member 124 refers to a member that is easily deformed, such as a gear train, a pulley, and a belt.
24, the actuator 122 and the operation unit 1
The movement of 26 is shifted. This actuator 122
An electric control device 116 is provided for controlling the actuator 12.
2 and the target value 114 (the current rotation angle and rotation speed if the actuator 122 is of a rotary type, and the advance / retreat amount and speed if the actuator 122 is linear).
Actuator 1 based on the value obtained by multiplying the deviation of
The power 118 applied to the motor 22 is increased / decreased and adjusted so that the current value 128 of the actuator 122 matches the target value 114. As viewed from the actuator 122, the electric control device 116 constitutes a closed loop feedback control device. However, since the operation unit 126 moves relatively differently from the actuator 122 due to the deformation of the low-rigidity member 124, it can be said that a substantially closed feedback control system is configured when viewed from the operation unit 126. This is called semi-closed loop feedback control. As is well known, the feedback-type electric controller 116 adjusts power such as electric power to be applied to the actuator 122 based on a value obtained by multiplying a gain by a difference between the current value 128 and the target value 114. At this time, it is necessary to determine an appropriate gain value. If an improper gain is used, the response of the mechanical system 120 becomes extremely slow or the operation becomes unstable.

In order to calculate an appropriate gain value, this technique uses a mechanical mathematical model 106. What is characteristic here is that the mechanical rigidity model 1
24 includes an element that is deformed, and the relationship established between the power 118 applied to the actuator 122 and the resulting motion value (for example, a value of a position or a speed) of the operation unit 126 and the mechanical system. Formula model 106
Equivalent value 108 input to the machine and the mechanical system mathematical model 10
The relation established with the operation unit motion value 110 calculated in 6 is mathematically equal. Double arrows 112 in FIG. 1 indicate that they are mathematically equivalent. Mechanical formula model 10
The operation unit movement value 110 calculated in step 6 is calculated by the calculation circuit 104.
Is converted to a value to be evaluated. Calculation circuit 10
The evaluation value calculated in 4 is sent to the search circuit 102, and the search circuit 102 searches for a gain value that provides a good evaluation value. The gain calculated in this manner is used by the electric control device 116. When the actuator 122 is feedback-controlled using this gain, the operation unit 1
26 will be well controlled.

As shown in FIG. 2, a mechanical mathematical model 10
The power equivalent value 108 to be added to the electric control device
Alternatively, a value obtained by an electric mathematical expression model 132 equivalent to the above may be input. The double arrow 134 in FIG. 2 indicates that they are mathematically equivalent. In this case, the electric control device 116
The relationship established between the input value (current value 128 and target value 114) input to and the output value (power equivalent value 108) and the relationship between the input value and the output value in the electrical mathematical model 132 are mathematically described. Has been equal to. Alternatively, the electric control device 1
The 16 output values can be directly added to the mechanical mathematical model 106.

As shown in FIG. 3, the calculation circuit 104 includes an operation unit motion value 110 calculated by a mechanical mathematical model 106.
, 146, 148, 1
50 and weight 140, 142, 1 for each element.
The result is multiplied by 44 and the obtained result is added to obtain one evaluation value, which is sent to the search circuit 102. The search circuit 102
For example, a suitable value search technique such as a downhill simplex method is used to search for a gain that provides a good evaluation value.

Next, a first embodiment of the present invention will be described with reference to FIG.
A description will be given with reference to FIG. FIG. 4 is a diagram showing a configuration example of the first embodiment, and FIG. 5 is a diagram showing a configuration of a mechanical system in the embodiment shown in FIG. As shown in FIG. 4, the electromechanical system according to the first embodiment calculates a mechanical system (robot) 10 that rotates one axis, an electric control device 20 that controls the mechanical system 10, and a gain used in the electric control device 20. And the like.

The structure of the mechanical system 10 will be described with reference to FIG. As shown in FIG. 5, a mechanical system 10 controlled in this embodiment is a single-axis turning robot, and includes a motor (rotary actuator) 11, a rotary encoder 12, a speed reducer 13 having a gear train, and an arm 14. ing.
The motor 11 is connected to the arm 14 via the speed reducer 13, and a motor that can obtain a torque proportional to the applied current value is used. Rotary encoder 12
Is connected to the rotating shaft of the motor 11,
Generates a pulse signal in accordance with the rotation of. The pulse signal is supplied to an electric control device 20 described later. The speed reducer 13 is a low-rigidity member having a low rigidity that is slightly deformed when transmitting a force. To show this low rigidity, a spring 13a is shown in the speed reducer 13, but actually the spring 13a
a is not necessarily interposed. The distal end of the arm 14 is an operation unit 14a of the mechanical system, and the mechanical system 10 performs necessary work by the operation unit 14a. The position of the operation unit 14a can be described in a rectangular coordinate system [x, y] having a coordinate origin at the turning center 14b of the arm 14. For example, the coordinate position [Hand
_x, Hand_y]. Operation unit 1
The speed of 4a can be described as a function of the rotational angular speed of the arm 14. Further, the position of the operation unit 14a can be described as a function of the rotation angle of the arm 14.

FIG. 4 shows the configuration of the electric control device 20.
This will be described below. As shown in FIG. 4, the mechanical system 10 is feedback-controlled by the electric control device 20.
Figure 4 shows a signal processing method of the electric control device 20 when the feedback control of the signal 20b indicating the rotation angle theta _M of the motor 11 as the current value, electric control means 2
0 is a target value θ _ref of the rotation angle of the motor 11 as an input value.
20a indicating the rotation angle of the motor 11 and the current value θ of the rotation angle of the motor 11
_M is input, and a current I _M to be applied to the motor 11 is output from these input values to control the motor 11. As shown in FIG. 4, the electric control device 20 is a digital control device in which feedforward control is added to well-known feedback control, and has a position proportional gain Kp,
The control is performed using five kinds of gains: a speed proportional gain Kvp, a speed integration time constant Ti, and feedforward gains F ₁ and F ₂ . The reciprocal of the speed integration time constant Ti can be considered as a gain. Then, three types of gains (position proportional gain K
p, speed proportional gain Kvp, reciprocal of speed integration time constant Ti)
Is multiplied by a value relating to the deviation between the target value θ _ref of the rotation angle of the motor 11 and the current value θ _M to obtain the control amount of the current I _M. Gains F ₁ and F _{2 for} feedforward control
Is multiplied by a value related to the target value θ _ref of the rotation angle of the motor 11 and an element of feedforward control is added.
The electric control device 20 can be mathematically described by the following equation (1).

[0018]

(Equation 1)

Here, in the equation (1), z ^-1 is a delay operator, k is the number of samples, st is a sample time, θ
_ref (k) is the target value of the rotation angle of the motor 11 at the time of k samples, θ _M (k) is the rotation angle of the motor 11 at the time of k samples,
I _M (k) indicates a current value applied to the motor 11 at the time of k samples. Accordingly, an electric mathematical expression model 31a mathematically equivalent to the electric control device 20 can be realized by the following expression (2).

[0020]

(Equation 2)

Here, the signs “′” of θ _M (k) and I _M (k)
Indicates the rotation angle of the motor 11 at the time of k samples and the current value of the motor 11 at the time of k samples obtained by the mechanical mathematical model. As can be seen from Expressions (1) and (2), the electric control device 20 and the electric mathematical expression model 31a constituted by Expression (2) are mathematically equivalent.

On the other hand, the equations of motion of the robot that rotates one axis, which is the mechanical system 10, are expressed by the following equations (3), (4), and (5).
Is described. In the mechanical mathematical model 31b that is mathematically equivalent to the mechanical system 10, when the position of the tip of the arm 14 other than the angle of the motor 11, that is, the coordinate position [Hand_x ', Hand_y'] of the operation unit 14a is output. Is also used in equation (5).

[0023]

(Equation 3)

In the equations (3) to (5), ω _M ′ represents the angular velocity of the motor 11, θ _A ′ represents the angle of the arm 14, and ω _A ′ represents the angular velocity of the arm 14. K _T is the torque constant of the motor 11, J _M is the rotor inertia of the motor 11, D _M is the viscosity coefficient of the motor 11, K _S is the spring constant of the speed reducer 13, J _A is the inertia of the arm 14, and D _A is Arm 1
4, L is the length of the arm 14, τ _M is the motor 1
It represents 1 Coulomb friction. As can be seen from equation (3), the rigidity of the low-rigidity reducer 13 is equal to the spring constant K _S
Equation (3) describes the equation of motion of a mechanical system including a low-rigidity member. Equation (3)
The various coefficients therein can be measured by measuring the behavior of the mechanical system 10. The pulse transfer function (shown in equation (6)) is obtained by discretizing equations (3) and (4) and converting them into transfer function expressions.
Is required. On the other hand, a pulse transfer function (shown in equation (7)) of the same order (fourth order) as equation (6) is obtained by system identification of the mechanical system 10. Therefore, the coefficients ([ni0, ni1, ni2, ni3, ni3,
ni4, di1, di2, di3, di4]), the coefficients ([nm0, nm1, nm2, nm3, nm4, dm1, dm2, dm3, dm) in equation (6)
4]), and further, each coefficient in equation (3) is determined. The input of the pulse transfer functions in Equations (6) and (7) is the current value given to the motor 11, and the output is the rotation angle of the motor 11. Therefore, the mechanical mathematical model 31b can be realized by the above equations (3) to (5).

Here, the angle θ of the motor 11_M, Motor 1
1 angular velocity ω_M, The angle θ of the arm 14 _A, Arm 14 corner
Speed θ_AIs used in the mechanical mathematical model 31b.
Indicates that the value is acceptable.

FIG. 4 shows the configuration of the gain calculating device 30.
This will be described below. As shown in FIG.
0 is a response estimator 31, a calculation circuit 32, and a search circuit 3
3 is provided. The response estimator 31 is an electric control device 2
This is a form in which an electrical mathematical model 31a mathematically equivalent to 0 and a mechanical mathematical model 31b are combined. further,
Since the electric control device 20 has a time delay of output, it is necessary to consider this time delay. Here, z ^-1 is added as a delay operator 31c between the electrical mathematical model 31a and the mechanical mathematical model 31b in order to model the output time delay. The operation unit motion value 31d of the operation unit 14a calculated by the mechanical mathematical model 31b is sent to the calculation circuit 32.

The calculation circuit 32 calculates the operation section motion value 31d.
The evaluation value 32a can be calculated based on
The evaluation value 32a can be described by the following equation (8).

[0028]

(Equation 4)

As shown in the equation (8), the evaluation value 32a
(J) is a value obtained by multiplying the four types of evaluation element values obtained from the operation section motion value 31d by weights α1, α2, α3, and α4, respectively, and adding the obtained four terms. In the equation (8), the first evaluation element value 32b is the overshoot amount of the operation unit 14a, and the second evaluation element value 32c is the operation unit 14a.
At the settling time, the third evaluation element value 32d is added to the squared error area of the operation unit 14a during the transition, and the fourth evaluation element value
2e corresponds to an error square area of the operation unit 14a in a steady state. Operation unit 14a for elapsed time t
Relationship of the target position 15 (y _{re f)} and the response position 16 (Hand_y) can be shown as in FIG. Operation unit 14a
Starts to move with respect to the target position (curve 15),
It converges while oscillating with respect to the target position that has reached the steady state. In FIG. 6, the time from when the target position 15 reaches the steady state until the operation unit 14a converges to the predetermined position is referred to as a settling time 32c. The overshoot amount of the response position after the target position 15 has reached the steady state is 32b. The error area during transition refers to the area of the region indicated by 32f, and the error area during steady state refers to the area of the region of 32g. The evaluation element values 32d, 32e are 32f, 3
This is a value obtained by squaring 2 g. Weights α1 to α1 in equation (11)
By variously changing α4, the weighting of each of the four terms can be changed. For example, the operating unit 1 in a steady state
When the purpose is to suppress the displacement of 4a, α
It is assumed that 1 = α2 = α3 = 0 and α4 = 1. In this case, the evaluation value 32a in which the fourth evaluation element value 32e is emphasized is calculated. When the purpose is to suppress the overshoot amount and settling time of the operation unit 14a, α1 = α2 =
1, α3 = α4 = 0.

The search circuit 33 searches for an appropriate gain based on the evaluation value 32a calculated by the calculation circuit 32. That is, five types of gains K that minimize the evaluation value 32a
Search for p, Kvp, Ti, F ₁ , and F ₂ using an appropriate value search technique such as the downhill simplex method. The gain searched in this manner is set as a gain candidate. When the gain candidate satisfies the convergence condition, the adjustment of the gain is completed, and the gain at that time is transferred to the electric control device 20 to control the mechanical system 10. After the adjusted gain is transferred to the electric control device 20, the switch 41 shown in FIG. 4 is switched from the A contact to the B contact. When the gain candidate does not satisfy the target convergence condition, the gain value of the response estimator 31 is changed to a new gain candidate,
The evaluation value 32a is calculated and evaluated again by the calculation circuit 32 and the search circuit 33. If the convergence condition is not satisfied, this operation is repeated.

The actuator of the present invention is constituted by the motor 11 and the like. Further, the speed reducer 13 corresponds to the low rigidity member of the present invention.

Next, a method of calculating a gain according to the first embodiment of the present invention will be described with reference to FIG. FIG. 7 is a diagram showing an operation flowchart of the first embodiment.
As shown in FIG. 4, first, the switch 41 is connected to the A side,
Control gains _{Kp, Kvp, Ti, F 1} , and inputs to the estimator 31 the initial value of F ₂ (step S1-1).

Next, the target angle θ _ref of the motor 11 is inputted, and the coordinate position of the operating portion 14a of the arm 14 at this time is calculated by using equations (2), (3), (4) and (5). Hand_
x ′, Hand_y ′] is obtained (step S1-2).

Next, an evaluation value 32a is obtained by the calculation circuit 32 from the evaluation function represented by the equation (8) (step S8).
1-3).

As the convergence condition, the convergence determination is performed by assuming that the gain candidate that minimizes the value of the evaluation value 32a has the same value, for example, ten times in succession (step S1-4).

When the above convergence condition is not satisfied, the search circuit 33 obtains a candidate value of the gain for making the value of the evaluation value 32a smaller, for example, by using a downhill simplex method which is a kind of optimization method. (Step S1-
5). Thereafter, the value of the gain of the response estimator 31 is changed to a new candidate value, the process returns to step S1-2, the response of the position of the operation unit 14a of the arm 14 is obtained, and the value of the evaluation value 32a is obtained by the calculation circuit 32. It is determined whether or not convergence has occurred.

When the convergence condition is satisfied, the adjustment of the gain is terminated, and the value of the gain at which the value of the evaluation value 32a becomes the minimum value is transferred to the electric control device 20 as an appropriate value of the gain. Switch from A side to B side,
Put the mechanical system 10 in an operable state (step S1-
6).

The weight in the evaluation function of equation (8) is α1 =
α2 = α3 = 0, α4 = 1, and the operation unit 14a in a normal state
FIG. 10 shows an example of the case where the adjustment is performed for the purpose of suppressing the positional deviation. FIG. 10 shows changes in the target value θ _ref and the current value θ _M of the motor 11 with respect to the elapsed time in the upper part,
Target value y _{r ef} of the operation unit 14a with respect to the elapsed time in the lower part
And the change of the current value Hand_y. Both the upper and lower rows are the results when the real target is operated with the gain adjusted by the gain calculator 30. At this time, a gain is obtained that minimizes the square value of the area of the area indicated by the hatching in FIG. 10, and according to the gain, the operating unit 14a in a steady state (when the target value is steady) is obtained. Will be suppressed. FIG. 11 shows an example in which weights in the evaluation function of Expression (11) are set to α1 = α2 = 1 and α3 = α4 = 0, and adjustment is performed with emphasis on suppressing the overshoot amount and settling time of the operation unit 14a. Shown in In FIG. 11, similarly to FIG. 10, a change in the target value θ _ref and the current value θ _M of the motor 11 with respect to the elapsed time is shown in the upper part, and a target value y _ref and the current value Han of the operation unit 14 a with respect to the elapsed time are shown in the lower part.
This shows a change in d_y. The evaluation value 32a at this time is based on the overshoot amount 32b of the operation unit 14a and the operation unit 14a.
Of the settling time 32c. By using the gain thus obtained, the overshoot amount and the settling time of the operation unit 14a are suppressed. 10 and 11
In the settling time 32c, the position of the operation unit 14a with respect to the target position is ±
It can be the time required to converge to 0.1 mm.

As described above, according to the gain calculation method of the first embodiment of the present invention, it is possible to calculate an appropriate gain so that the motion generated in the operation section 14a of the arm 14 satisfies a desired response. it can. In addition, prior to actually manufacturing the electric control device 20, by adjusting the gain to an appropriate value, it is possible to verify whether or not the motion generated in the operation unit 14a can be controlled to one that can obtain a high evaluation. it can. Further, the evaluation value 32a can be calculated in consideration of the response to the control targeted by the operation unit 14a, the positioning accuracy, and the like.

In the first embodiment of the present invention, the electric control device 20 is a control device in which feedforward is added to feedback control, but may be a control device having only feedback control. Further, the mechanical system 10 is a single-axis turning robot, but the configuration of the mechanical system can be variously changed. Of course, the present invention can also be applied to a robot that rotates multiple axes. Although the power of the motor 11 is electric power, the power of the motor 11 can be variously changed. For example, high-pressure air can be used as the power. The above-described system is also effective for obtaining an appropriate value of a gain that is strong against disturbance. In this case, as shown in FIG. 4, a disturbance is input to a mechanical mathematical model, and an operating unit generated at that time is input. A control system can be constructed after evaluating the motion and obtaining a gain that is hardly affected by disturbance. Alternatively, the magnitude of the load applied to the operation unit is the arm inertia J _A of the mechanical system mathematical model of Expression (3).
By calculating the behavior when this value is changed in order to influence the magnitude of the gain, it is possible to select an appropriate value of the gain that can satisfactorily cope with a change in the load applied to the operation unit. Further, in the case of a multi-axis mechanical system, since the reaction force of the actuator on the tip side is applied to the actuator on the root side, the mathematical model can be modified while taking this element into consideration. An appropriate gain value for each axis of the system can be calculated.

Next, a second embodiment of the present invention will be described with reference to FIG. FIG. 8 is a diagram illustrating a configuration example of the second embodiment. As shown in FIG. 8, the second embodiment is an application of the first embodiment, and has substantially the same configuration as the first embodiment. That is, the electromechanical system according to the second embodiment includes the mechanical system 10 for one-axis turning as shown in FIG.
Is a response estimator 71, a calculation circuit 72, and a search circuit 73
It has. The response estimator 71 combines a current I _M output from an electric control device 60 similar to the electric control device 20 of the first embodiment with a mechanical system mathematical model 71b that is mathematically equivalent to the mechanical system 10. It is a form. As shown in FIG. 4, the electric control device 60 is a digital control device in which feedforward is added to well-known feedback control, and includes a position proportional gain Kp, a speed proportional gain Kvp, and a speed integral. Control is performed using five types of gains, a time constant Ti and feedforward gains F ₁ and F ₂ . The mechanical mathematical model 71b is the same as the mechanical mathematical model 31b of the first embodiment,
It can be described by (5). The calculation circuit 72 and the search circuit 73 have the same configuration as the calculation circuit 32 and the search circuit 33 of the first embodiment. Therefore, the evaluation value 72a of the calculation circuit 72 can be described by Expression (8), and the four types of evaluation element values 72 obtained from the operation section motion value 71d are obtained.
b, 72c, 72d, and 72e respectively have weights α1 and α
It is indicated by a value obtained by multiplying 2, α3, and α4 and adding the obtained four terms. The search circuit 73 calculates five types of gains K when the evaluation value 72a calculated by the calculation circuit 72 is minimized.
For example, p, Kvp, Ti, F ₁ , and F ₂ are searched using a suitable value search technique such as a downhill simplex method, and the searched gain is set as a gain candidate. If the gain candidate satisfies the target convergence condition, the adjustment of the gain ends, and the mechanical system 10 is controlled by the gain at that time.
When the control of the mechanical system 10 is actually performed, the switches 42 and 43 shown in FIG. 8 may be switched. If the gain candidate does not satisfy the target convergence condition, the gain value of the response estimator 71 is changed to a new gain candidate, and the evaluation value 72 is again calculated by the calculation circuit 72 and the search circuit 73.
a is calculated and evaluated. If the convergence condition is not satisfied, this operation is repeated.

As described above, the second embodiment differs from the first embodiment in that the second embodiment differs from the response estimator 71 in the first embodiment.
In estimating the response of the operation unit 14a, the output of the electric control device 60 itself is input to the mechanical formula model 71b and used instead of the electric mathematical model equivalent to the electric control device. Therefore, the second embodiment is effective when an electric mathematical model equivalent to the electric control device 60 cannot be easily derived.

Next, a method of calculating a gain according to the second embodiment of the present invention will be described with reference to FIG. FIG. 9 is a diagram showing an operation flowchart of the second embodiment. FIG.
As shown in (1), first, the switches 42 and 43 are connected to the A side, and the initial values of the gains Kp, Kvp, Ti, F ₁ and F ₂ are input to the electric control device 60 of the response estimator 71 (step S).
2-1).

Next, the target angle θ _ref of the motor 11 is input, the current I _M output from the electric control device 60 is detected, and the detected value is used as the equation (1), (3), (4). , (5)
, The coordinate position of the operation unit 14 a of the arm 14 [Ha
nd_x ′, Hand_y ′] (step S2-2).
The current I _M output from the electric control device 60 is
Let it correspond to I _M ′ in equation (3).

Next, the value of the evaluation value 72a is obtained by the calculation circuit 72 from the evaluation function represented by the equation (8) (step S2-3). As in the first embodiment, by appropriately changing the weights α1 to α4 of the evaluation function 72a for control purposes, as shown in FIG. 10 and FIG. It is possible to adjust the gain such that

The convergence condition is such that the candidate for the gain with the minimum value of the evaluation value 72a has the same value, for example, ten times in succession, and the convergence is determined (step S2-4).

When the above convergence condition is not satisfied, the search circuit 73 obtains a candidate value of the gain which makes the value of the evaluation value 72a smaller, for example, by using a downhill simplex method which is a kind of optimization method. (Step S2-
5). Then, the value of the gain of the response estimator 71 is changed to a new candidate value, the process returns to step S2-2, the response of the position of the operation unit 14a of the arm 14 is obtained, and the value of the evaluation value 72a is obtained by the calculation circuit 72. Repeat until convergence.

When the convergence condition is satisfied, the adjustment of the gain is terminated, and the value of the gain at which the value of the evaluation value 72a becomes the minimum value is used for controlling the mechanical system 10. Switches 42, 43
Are switched from the A side to the B side, and the mechanical system 10 is made operable (step S2-6).

As described above, according to the gain calculating method of the second embodiment of the present invention, the motion generated in the operation section 14a of the arm 14 is set to an appropriate gain so as to satisfy the desired response or stability. Can be calculated. Further, even when the relationship between the input value and the output value of the electric control device 60 is extremely complicated and an electric mathematical model cannot be easily obtained, an appropriate value of the gain can be easily calculated. . Further, the evaluation value 72a can be calculated in consideration of responsiveness to the control targeted by the operation unit 14a, positioning accuracy, and the like.

In the second embodiment of the present invention, the electric control device 60 is a control device in which feedforward is added to feedback control, but may be a control device having only feedback control. Further, the mechanical system 10 is a single-axis turning robot, but the configuration of the mechanical system can be variously changed. Of course, the present invention can also be applied to a multi-axis turning robot. Although the power of the motor 11 is electric power, the power of the motor 11 can be variously changed. For example, high-pressure air can be used as the power. The above-described system is also effective for obtaining an appropriate value of a gain that is strong against disturbance. In this case, as shown in FIG. 4, a disturbance is input to a mechanical mathematical model, and an operating unit generated at that time is input. A control system can be constructed after evaluating the motion and obtaining a gain that is hardly affected by disturbance. Alternatively, the magnitude of the load applied to the operation unit is the arm inertia J _A of the mechanical system mathematical model of Expression (3).
By calculating the behavior when this value is changed in order to influence the magnitude of the gain, it is possible to select an appropriate value of the gain that can satisfactorily cope with a change in the load applied to the operation unit. Further, in the case of a multi-axis mechanical system, since the reaction force of the actuator on the tip side is applied to the actuator on the root side, the mathematical model can be modified while taking this element into consideration. An appropriate gain value for each axis of the system can be calculated.

[0051]

As described above, by using the gain calculating device according to the first aspect, the deformation that occurs in the low-rigidity member is compensated for while the semi-closed loop feedback control is performed, and as a result, the operation unit is provided. It is possible to calculate an appropriate value of the gain that satisfies the desired response or stability of the generated motion. In addition, by using the gain calculating device according to claim 2, by adjusting the gain to an appropriate value, it is possible to verify in advance whether or not the motion generated in the operation unit can be controlled to one that can obtain a high evaluation. it can. Claim 3
In the case of using the gain calculation device described in (1), even when the relationship between the input value and the output value of the electric control device is very complicated and a mathematical model cannot be easily obtained, it is possible to easily set an appropriate gain value. Can be calculated. In addition, the use of the gain calculating device according to claim 4 enables tuning according to the necessity of the response and positioning accuracy of the operation unit.

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic concept of an embodiment of the present invention.

FIG. 2 is a diagram showing a schematic concept of an embodiment of the present invention.

FIG. 3 is a diagram showing a schematic concept of an embodiment of the present invention.

FIG. 4 is a diagram illustrating a configuration example of a first embodiment of the present invention.

FIG. 5 is a diagram illustrating a configuration example of a mechanical system.

FIG. 6 is a diagram illustrating an example of a time response of an arm.

FIG. 7 is an operation flowchart of the first embodiment of the present invention.

FIG. 8 is a diagram illustrating a configuration example of a second embodiment of the present invention.

FIG. 9 is an operation flowchart of the second embodiment of the present invention.

FIG. 10 is a diagram illustrating an example of a time response of a mechanical system.

FIG. 11 is a diagram showing an example of a time response of a mechanical system.

[Explanation of symbols]

 10, 120 ... mechanical system 14a, 126 ... operating unit 20, 60, 116 ... electric control device 30, 70 ... gain calculation device 31, 71 ... response estimator 31a, 132 ... electric system mathematical model 31b, 71b, 106 ... Mechanical system mathematical model 32, 72, 104 ... calculation circuit 32a, 72a ... evaluation value 33, 73, 102 ... search circuit

 ──────────────────────────────────────────────────続 き Continued on the front page F-term (reference)

Claims (4)

[Claims] 1. A mechanical system in which a low-rigidity member is interposed between an actuator and an operation unit, and a current value of the actuator is set based on a value obtained by multiplying a deviation between a current value and a target value of the actuator by a gain. A device for calculating an appropriate value of the gain in an electromechanical system including an electric actuator control device for controlling to be the target value, wherein between a power applied to the actuator and a motion generated in the operation unit. A mechanical mathematical model that realizes a mathematically equivalent relationship to the relationship that holds, a calculating circuit that calculates an evaluation value that evaluates a motion generated in the operation unit obtained from the mechanical mathematical model, and a calculation circuit that calculates the evaluation value. A search circuit for searching for an appropriate value of the gain from a change in the evaluation value to be obtained. 2. The gain calculation device according to claim 1, wherein the relationship between a current value input to the electric actuator control device and an output value with respect to a target value is mathematically equivalent. A gain calculating apparatus further comprising a system formula model, wherein an output value of the electrical system formula model is input to the mechanical system model, and an output value of the mechanical system model is input to the calculation circuit. 3. The gain calculating device according to claim 1, wherein an output value of the electric actuator control device is input to the mechanical mathematical model, and an output value of the mechanical mathematical model is calculated by the calculating circuit. A gain calculating device for inputting the data to a device. 4. The gain calculation device according to claim 1, wherein the calculation circuit calculates a plurality of evaluation element values based on output values of the mechanical mathematical model, and further calculates the plurality of evaluation element values. A gain calculating apparatus for calculating an evaluation value by adding a value obtained by multiplying an evaluation element value by each weight, and capable of increasing or decreasing each weight.