阅读说明：本技术 振动抑制装置、振动抑制方法以及程序 (Vibration suppression device, vibration suppression method, and program ) 是由 西田吉晴 四方田真美 木田直希 于 2019-08-19 设计创作，主要内容包括：一种振动抑制装置,其对具有固有振动模式的机械系统中的动作部的振动进行抑制,该机械系统包括动作部、使动作部工作的致动部、以及将动作部和致动部连结的弹性体,其中,振动抑制装置的特征在于,具备：生成机构,其生成驱动致动部的驱动信号；推定机构,其对与机械系统相关的计测量进行推定；补正机构,其基于由推定机构推定出的计测量,对由生成机构生成的驱动信号进行补正；以及变更机构,其在机械系统的模型化误差增大的期间对在推定机构中使用的增益进行变更,以使得模型化误差的增大的影响变小。(A vibration suppression device that suppresses vibration of an operating unit in a mechanical system having a natural vibration mode, the mechanical system including the operating unit, an actuating unit that actuates the operating unit, and an elastic body that connects the operating unit and the actuating unit, the vibration suppression device being characterized by comprising: a generation mechanism that generates a drive signal that drives the actuator; an estimation means for estimating a measurement amount related to the mechanical system; a correction means for correcting the drive signal generated by the generation means on the basis of the measurement amount estimated by the estimation means; and a changing unit that changes the gain used in the estimating unit so that the influence of the increase in the modeling error is reduced while the modeling error of the mechanical system increases.)

1. A vibration suppressing device for suppressing vibration of an operating portion in a mechanical system having a natural vibration mode, the mechanical system including the operating portion, an actuating portion for actuating the operating portion, and an elastic body for connecting the operating portion and the actuating portion, wherein the vibration suppressing device is characterized in that,

the vibration suppression device is provided with:

a generation mechanism that generates a drive signal that drives the actuator;

an estimation means for estimating a measurement amount related to the mechanical system;

a correction means for correcting the drive signal generated by the generation means based on the measurement value estimated by the estimation means; and

and a changing unit that changes the gain used in the estimating unit so that an influence of an increase in the modeling error of the mechanical system is reduced while the modeling error increases.

2. The vibration suppressing device according to claim 1,

the period in which the modeling error increases is a period before and after the direction of the angular velocity of the actuator is reversed.

3. The vibration suppressing device according to claim 1,

the estimation mechanism estimates the measurement value using a differential value of a state quantity in a state equation of the mechanical system.

4. The vibration suppressing device according to claim 1,

the correction means corrects the drive signal by performing positive feedback based on the measurement of the meter on the drive signal.

5. The vibration suppressing device according to claim 1,

the meter measurement is an estimated value of the vibration torque of the actuator or an estimated value of a differential value of the vibration torque of the actuator.

6. The vibration suppressing device according to claim 5,

the estimation means calculates an estimated value of the vibration torque or an estimated value of a differential value of the vibration torque from the disturbance by filtering the estimated disturbance.

7. The vibration suppressing device according to claim 6,

the changing unit uses a convergence gain in the filtering process as the gain.

8. The vibration suppressing device according to claim 1,

the meter measurement is an estimated value of the vibration torque of the actuator or a differential value of the estimated value of the vibration torque of the actuator.

9. The vibration suppressing device according to claim 8,

the estimating means performs weighted least square estimation of a forgetting factor on the estimated disturbance, thereby calculating an estimated value of the vibration torque from the disturbance.

10. The vibration suppressing device as recited in claim 9,

the changing unit uses the weight in the weighted least square estimation and the forgetting factor as the gain.

11. The vibration suppressing device according to claim 1,

the measurement is an estimation of the angular velocity of the operating portion.

12. The vibration suppressing device as recited in claim 11,

the changing means uses a convergence gain used when calculating the estimated value of the angular velocity as the gain.

13. The vibration suppressing device according to claim 1,

the gauge measurement is an estimate of the angular velocity of elastic deformation of the elastomer.

14. A vibration suppressing method for suppressing vibration of an operating portion in a mechanical system having a natural vibration mode, the mechanical system including the operating portion, an actuating portion for actuating the operating portion, and an elastic body for connecting the operating portion and the actuating portion, wherein the vibration suppressing method is characterized in that,

the method comprises the following steps:

a step of estimating a measurement amount related to the mechanical system by changing a gain so that an influence of an increase in a modeling error of the mechanical system is reduced while the modeling error increases, and by not changing the gain except for the period; and

and correcting a drive signal for driving the actuator based on the estimated measurement.

15. A program for causing a computer to function as a vibration suppression device for suppressing vibration of an operating portion in a mechanical system having a natural vibration mode, the mechanical system including the operating portion, an actuating portion for actuating the operating portion, and an elastic body for connecting the operating portion and the actuating portion,

the program is for causing the computer to function as a generating means, an estimating means, a correcting means, and a changing means,

the generation mechanism generates a drive signal that drives the actuator,

the estimation means estimates a measurement amount related to the mechanical system,

the correction means corrects the drive signal generated by the generation means based on the measurement value estimated by the estimation means,

the changing means changes the gain used in the estimating means so that the influence of the increase in the modeling error of the mechanical system is reduced while the modeling error increases.

Technical Field

The present invention relates to a vibration suppression device, a vibration suppression method, and a program for suppressing vibration of an operating unit in a mechanical system having a natural vibration mode, the mechanical system including the operating unit, an actuating unit for actuating the operating unit, and an elastic body for connecting the operating unit and the actuating unit.

Background

Known methods for suppressing ringing vibration are as follows: the vibration torque (current value) generated by the elastic body is obtained by passing a current corresponding to a disturbance torque obtained from a disturbance observer of a motor mounted in a drive system including the elastic body through a band-pass filter, and the current value is set to have a gain K_CThe torque is amplified and subtracted from a current command (torque command) of the motor (see, for example, patent document 1).

Documents of the prior art

Patent document

Patent document 1: japanese laid-open patent publication No. 7-337058

Disclosure of Invention

Problems to be solved by the invention

Here, in the case of a configuration in which the disturbance torque obtained by the disturbance observer is subjected to filter processing to extract the vibration torque and the torque command is corrected by open-loop control, if the influence of the modeling error of the mechanical system is large, there is a possibility that an erroneous vibration torque component is extracted by the filter processing and erroneous vibration suppression is performed.

The purpose of the present invention is to reduce the possibility of erroneous vibration suppression due to modeling errors of a mechanical system.

Means for solving the problems

In view of the above object, the present invention provides a vibration suppressing device for suppressing vibration of an operating unit in a mechanical system having a natural vibration mode, the mechanical system including the operating unit, an actuating unit for actuating the operating unit, and an elastic body for connecting the operating unit and the actuating unit, the vibration suppressing device comprising: a generation mechanism that generates a drive signal that drives the actuator; an estimation means for estimating a measurement amount related to the mechanical system; a correction means for correcting the drive signal generated by the generation means on the basis of the measurement amount estimated by the estimation means; and a changing unit that changes the gain used in the estimating unit so that the influence of the increase in the modeling error is reduced while the modeling error of the mechanical system increases.

Here, the period during which the modeling error increases may be a period before and after the direction of the angular velocity of the actuator is reversed.

In addition, the estimation means may estimate the measurement using a differential value of the state quantity in the state equation of the mechanical system.

The correction means may correct the drive signal by performing positive feedback based on the measurement of the drive signal.

The measured value may be an estimated value of the vibration torque of the actuator or an estimated value of a differential value of the vibration torque of the actuator. In this case, the estimation means may calculate the estimated value of the vibration torque or the estimated value of the differential value of the vibration torque from the disturbance by filtering the estimated disturbance. The changing means may use the convergence gain in the filtering process as the gain.

The measured value may be an estimated value of the vibration torque of the actuator or a differential value of the estimated value of the vibration torque of the actuator. In this case, the estimation means may calculate the estimated value of the vibration torque from the disturbance by performing least square estimation of the estimated disturbance by a forgetting factor and weighting. The changing means may use, as the gain, a weight in the weighted least square estimation and the forgetting factor.

The measured value may be an estimated value of the angular velocity of the operating portion. In this case, the changing means may use a convergence gain used when calculating the estimated value of the angular velocity as the gain.

Alternatively, the measured value may be an estimated value of the elastic deformation angular velocity of the elastic body.

Further, the present invention provides a vibration suppressing method for suppressing vibration of an operating portion in a mechanical system having a natural vibration mode, the mechanical system including the operating portion, an actuating portion for actuating the operating portion, and an elastic body for connecting the operating portion and the actuating portion, the vibration suppressing method including the steps of: estimating a measurement amount related to the mechanical system by changing the gain so that an influence of an increase in the modeling error is small while the modeling error of the mechanical system increases, and by not changing the gain except for the period; and correcting a drive signal for driving the actuator based on the estimated measurement.

The present invention also provides a program for causing a computer to function as a vibration suppression device, the vibration suppressing device suppresses vibration of an operating portion in a mechanical system having a natural vibration mode, the mechanical system includes an operating part, an actuating part for operating the operating part, and an elastic body for connecting the operating part and the actuating part, wherein the program causes the computer to function as a generating means for generating a drive signal for driving the actuator, an estimating means for estimating a measurement amount related to the mechanical system, a correcting means for correcting the measurement amount based on the measurement amount estimated by the estimating means, the drive signal generated by the generation means is corrected, and the change means changes the gain used in the estimation means so that the influence of the increase in the modeling error is reduced while the modeling error of the mechanical system is increasing.

Effects of the invention

According to the present invention, it is possible to reduce the possibility of erroneous vibration suppression due to modeling errors of a mechanical system.

Drawings

Fig. 1 is a diagram showing a structure of a machine system 10 to which the present embodiment is applied.

Fig. 2 is a graph showing a simulation result of a vibration behavior in a case where vibration suppression is not performed, among vibration behaviors at the time of a minute motion.

Fig. 3 is a graph showing a simulation result of vibration behavior in the case where vibration suppression is performed by the related art among vibration behaviors at the time of a minute motion.

Fig. 4 is a block diagram showing a configuration example of the machine control system of the first embodiment.

Fig. 5 (a) is a graph showing a result of interference estimation in a periodic interference observer of the related art, and fig. 5 (b) is a graph showing a result of interference estimation in the periodic interference observer of the first embodiment.

FIG. 6 is a graph showing the feedback gain G_dSet to 1 and estimate the vibration torque d_ωGraph of results obtained with positive feedback.

FIG. 7 is a graph showing the feedback gain G_dSet to 3 and estimate the vibration torque d_ωGraph of results obtained with positive feedback.

FIG. 8 is a graph showing the feedback gain G_dSet the value to 10 and estimate the vibration torque d_ωGraph of results obtained with positive feedback.

Fig. 9 is a flowchart showing an operation example of the convergence gain changing unit, the periodic disturbance observer, and the like of the controller according to the first embodiment.

Fig. 10 is a block diagram showing a configuration example of a machine control system according to a second embodiment.

Fig. 11 is a flowchart showing an example of operations of the weight changing unit, the least square estimating unit, and the like of the controller according to the second embodiment.

Fig. 12 is a block diagram showing a configuration example of a case where a periodic disturbance speed observer is provided in the machine control system according to the third embodiment.

Fig. 13 is a flowchart showing an operation example of a convergence gain changing unit, a periodic disturbance speed observer, and the like of the controller according to the third embodiment.

Fig. 14 is a block diagram showing a configuration example of a machine control system according to the fourth embodiment.

Fig. 15 is a flowchart showing an example of operations of a weight changing unit, a least square estimating unit, and the like of the controller according to the fourth embodiment.

Fig. 16 is a block diagram showing a configuration example of a machine control system according to a fifth embodiment.

Fig. 17 is a block diagram showing a flow for calculating an arm angular velocity estimation value from the motor angular velocity and the torque.

Fig. 18 is a graph showing the difference between the actual value of the arm angular velocity and the estimated value of the arm angular velocity when the convergence gain is not changed in accordance with the target angular velocity.

Fig. 19 is a graph showing the result of using the arm angular velocity estimation value in the case where the convergence gain is not changed in accordance with the target angular velocity and performing feedback.

Fig. 20 is a graph showing the difference between the actual value of the arm angular velocity and the estimated value of the arm angular velocity when the convergence gain is changed in accordance with the target angular velocity.

Fig. 21 is a graph showing the result of using the arm angular velocity estimation value in the case where the convergence gain is changed in accordance with the target angular velocity and performing feedback.

Fig. 22 is a flowchart showing an operation example of the convergence gain changing unit, the arm angular velocity estimation observer, and the like of the controller according to the fifth embodiment.

Detailed Description

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[ background of the present embodiment ]

Fig. 1 is a diagram showing a structure of a machine system 10 to which the present embodiment is applied. As shown in the drawing, the mechanical system 10 is a mechanical system having a natural vibration mode and configured by coupling the arm 1 and the elastic body 3 such as a reduction gear for the motor 2. The rotation angle of the arm 1 is represented by θ_ALet the rotation angle of the motor 2 be θ_MLet τ be a command torque for the motor 2. Returned by motor 2 to rotation angle theta_M. Here, the arm 1 is an example of an operation portion, the motor 2 is an example of an actuator portion, and the command torque τ to the motor 2 is an example of a drive signal for driving the actuator portion. In addition, the command torque τ and the rotation angle θ are described below_MReferred to as actuation information. At this time, the motion equation of the mechanical system 1 is given by the following equation.

[ number 1]

In this equation, the natural frequency of the arm 1 is √ (K/J)_A). In addition, J_AIs the inertia of the arm 1 (hereinafter referred to as "arm inertia"), J_MIs the inertia of the motor 2 (hereinafter referred to as "motor inertia"), K is the rigidity of the elastic body 3, and f is_AIs the frictional force of the arm 1 (hereinafter referred to as "arm frictional force"), f_MIs a frictional force of the motor 2 (hereinafter referred to as "motor frictional force"), f is an excitation force acting on the arm 1, and e is an elastic deformation (═ θ ∈)_A-θ_M). In the mathematical expressions and drawings in the specification, "·" shown directly above the text represents a first order differential of time, and "·" shown directly above the text represents a second order differential of time. On the other hand, in the text of the specification, the first derivative of time is represented by "d/dt", and "d" is used²/dt²"represents the second order differential of time.

Here, as a conventional technique, a technique may be considered in which a disturbance observer is configured, an output of the disturbance observer is filtered to extract an oscillation torque component, and a drive signal is corrected by open-loop control.

In a well-controlled robot or the like, usually, a vibration torque component in a drive signal is minute and is easily affected by a modeling error. In particular, when a robot or the like performs a small motion, the influence of the frictional force becomes a dominant factor, and the frictional force is abruptly changed at the time of the folding motion. In this case, the generated modeling error is increased sharply, and the included frequency components also include complicated high-frequency components.

When the conventional technique is applied under such a situation, the output of the disturbance observer is greatly affected by a modeling error due to a frictional force at the time of folding, and even if a filtering process is performed to extract the vibration torque component, an erroneous component is extracted. In particular, when the phase deviation is 90 ° or more, the vibration suppression effect is lost, and the vibration may be excited. Due to the influence of the modeling error, even the vibration torque component estimated by the conventional technique is subjected to a phase shift of 90 ° or more from the true value, and as a result, the excitation is performed. That is, during the folding back, the modeling error component is much larger than the vibration torque component, and the control is performed based on the erroneous estimation result in the conventional technique.

Fig. 2 and 3 are graphs showing the results of the simulation of the vibration behavior in the case of a minute motion. Fig. 2 shows the vibration behavior in the case where vibration suppression is not performed, and fig. 3 shows the vibration behavior in the case where vibration suppression is performed by the related art. As is clear from these figures, the conventional technique does not suppress vibration at all.

Fig. 3 shows the simulation result when the excitation force f does not act, but when the excitation force f acts, the phase shift is caused by the influence of the modeling error, and the conventional technique may instead make the vibration redundant.

Therefore, in the present embodiment, at a location where the modeling error (particularly, including a high-frequency component) is large, the influence of the modeling error is reduced by reducing the gain of the estimation unit. Specifically, the influence of the modeling error is reduced by reducing the convergence gain of the observer or by reducing the weight in the least-squares estimation before and after the direction reversal of the motor angular velocity in which the change in the modeling error of the frictional force is severe.

[ first embodiment ]

Fig. 4 is a block diagram showing a configuration example of the machine control system 100 according to the first embodiment. As shown, machine control system 100 includes a machine system 10 and a controller 20 that controls machine system 10. The controller 20 realizes each function by reading a program from a storage unit (not shown) such as a ROM and executing the program, for example, by a CPU (not shown). In the present embodiment, the controller 20 is provided as an example of the vibration suppression device.

First, a functional configuration of the machine system 10 will be described.

As shown, the machine system 10 includes operators 11 to 15.

The arithmetic unit 11 obtains the arm friction force f obtained by subtracting the arm friction force f output from the arithmetic unit 12 from the excitation force f_A(dθ_ADt) and the result obtained by subtracting K epsilon outputted from the arithmetic unit 15. Then, the result is divided by the arm inertia J_AAnd the obtained result is integrated to calculate the arm angular velocity d theta_ADt and outputs the calculated arm angular velocity d theta_A/dt。

The arithmetic unit 12 obtains the arm angular velocity d θ output from the arithmetic unit 11_AAnd/dt. Then, based on the arm angular velocity d θ_ACalculating arm friction force f from/dt_A(dθ_ADt) and outputs the calculated arm friction force f_A(dθ_A/dt)。

The arithmetic unit 13 obtains a motor friction force f obtained by subtracting the motor friction force f output from the arithmetic unit 14 from the command torque τ output from the controller 20_M(dθ_MDt) and added with K epsilon outputted from the arithmetic unit 15. Then, the result is divided by the motor inertia J_MAnd the obtained result is integrated to calculate the motor angular velocity d theta_M/dt, and outputs the calculated motor angular velocity d theta_M/dt。

The arithmetic unit 14 obtains the motor angular velocity d θ output from the arithmetic unit 13_MAnd/dt. Then, based on the motor angular velocity d θ_MCalculating motor friction force f by using/dt_M(dθ_MDt) and outputs the calculated motor friction force f_M(dθ_M/dt)。

The arithmetic unit 15 obtains an elastic deformation speed d ε/dt according to an arm angular speed d θ output from the arithmetic unit 11_AThe angular speed d theta of the motor output from the arithmetic unit 13 is subtracted from/dt_MAnd/dt. Then, K ∈ is calculated by integrating the result obtained by multiplying the elastic deformation speed d ∈/dt by the rigidity K, and the calculated K ∈ is output.

Next, a functional configuration of the controller 20 will be explained.

As shown in the figure, the controller 20 includes a PID control portion 21, an inertia compensation portion 22, and a friction compensation portion 23.

The PID control unit 21 obtains a target angular velocity d θ of the motor 2 (hereinafter referred to as "target angular velocity") instructed to the controller 20_DThe angular speed d θ of the motor fed back by the mechanical system 10 is subtracted from/dt_MThe result obtained by/dt. Then, this result is PID-controlled to output a feedback torque.

The inertia compensation unit 22 obtains a target angular velocity d θ instructed to the controller 20_DAnd/dt. Then, by adjusting the target angular velocity d θ_D(J) calculating inertial compensation using arm and motor inertial models_A^+J_M^)d²θ_D/dt²", thereby outputting a feedforward torque based on the inertial model.

The friction compensation unit 23 obtains a target angular velocity d θ instructed to the controller 20_DAnd/dt. Then, by adjusting the target angular velocity d θ_DThe frictional compensation f is calculated by applying an arm friction model and a motor friction model_A^(dθ_D/dt)+f_M^(dθ_DDt) ", thereby outputting a feedforward torque based on the friction model.

Note that, although the cap symbol is shown directly above the character in the numerical expression and the drawings of the specification, the cap symbol is shown behind the character in the text of the specification. In the present embodiment, a PID control unit 21, an inertia compensation unit 22, and a friction compensation unit 23 are provided as an example of a generation means for generating a drive signal.

Further, the controller 20 includes a vibration torque estimation unit 30. The vibration torque estimation unit 30 obtains the motor angular velocity d θ output from the machine system 10_MDt and an estimated value d of vibration torque generated by the elastic body 3 and acting on the motor 2_ωAnd a. Specifically, the vibration torque estimation unit 30 includes an arithmetic unit 31 and a periodic disturbance observer 32.

The computing unit 31 obtains the motor angular velocity d θ output from the machine system 10_MAnd/dt. Then, the motor angular velocity d theta is adjusted_MDt times arm inertia J_AAnd motor inertia J_MThe result of the sum of the two is differentiated to output an inertial force.

The periodic disturbance observer 32 obtains a disturbance torque d obtained by subtracting the command torque τ to the motor 2 from the sum of the inertia force output from the arithmetic unit 31 and the torque based on the friction model output from the friction compensation unit 23. Then, the disturbance torque d is subjected to a filtering process for extracting only the vibration component of the natural frequency ω represented by the following equation, thereby calculating the vibration torque estimated value d_ωAnd outputs the estimated value d of the vibration torque obtained by the calculation_ω^。

[ number 2]

In the present embodiment, the estimated value d of the vibration torque is used as an example of the measurement related to the mechanical system_ωAn oscillation torque estimating unit 30 is provided as an example of an estimating means for estimating the measurement amount.

The controller 20 includes an arithmetic unit 33. The arithmetic unit 33 obtains the estimated value d of the vibration torque output from the vibration torque estimating unit 30_ωAnd a. Then, the value d is estimated by estimating the vibration torque_ωMultiplying feedback gain G_dThe vibration suppression torque is calculated, and the calculated vibration suppression torque is output.

Thus, the feedforward torque based on the inertia model output by the inertia compensation unit 22 is added to the feedback torque output by the PID control unit 21, and the vibration suppression torque output by the calculation unit 33 is added to the result. In this sense, the arithmetic unit 33 is an example of a correction mechanism that corrects the drive signal based on the measurement. Then, the feedforward torque based on the friction model output by the friction compensation unit 23 is added to the result, and the result becomes the command torque τ for the motor 2.

However, in the first embodiment, the controller 20 includes the convergence gain changer 34 in addition to these configurations. The convergence gain changing unit 34 compares the target angular velocity d θ indicated to the controller 20 with the convergence gain change value_DThe value of ζ corresponding to the convergence gain of the filter used in the periodic disturbance observer 32 is corrected accordingly, thereby being free from the influence of the turning-back time when the modeling error including the high-frequency component increases. Specifically, when the target angular velocity d θ_DAnd when the/dt is small, the convergence gain zeta is reduced. At the target angular velocity d θ_DWhen the absolute value of/dt is 0.01rad/s or less, the convergence gain ζ is set to 0, and the influence of the modeling error is eliminated. In the present embodiment, the target angular velocity d θ is used as an example of a period during which the modeling error of the mechanical system increases_DWhile dt is decreasing, a convergent gain ζ is used as an example of a gain used in the estimating means, and a convergent gain changing unit 34 is provided as an example of a changing means that changes a gain so that an influence of an increase in a modeling error of the mechanical system is decreased while the modeling error is increasing.

Fig. 5 (a) is a graph showing a result of interference estimation in the periodic interference observer 32 of the related art, and fig. 5 (b) is a graph showing a result of interference estimation in the periodic interference observer 32 of the first embodiment. As is clear from fig. 5 (a) and 5 (b), the excitation force cannot be estimated at all due to the influence of the modeling error in the conventional technique, but the excitation force can be estimated accurately in the first embodiment.

In addition, FIG. 6 shows a feedback gain G_dSet to 1 and estimate the vibration torque d_ωPositive feedback, and resonance ratio control in a manner weaker than interferenceGraph of the results obtained. As is apparent from fig. 6, the vibration is suppressed to 1/2 compared to the case where the vibration suppression shown in fig. 2 is not provided.

Fig. 7 and 8 show that the vibration torque estimated value d is obtained by increasing the feedback gain_ωAnd (b) performing positive feedback, and performing resonance ratio control to obtain a result chart. As can be seen from FIG. 7, the feedback gain G is set_dWhen the value is 3, the vibration can be reduced to about 1/4, and as is apparent from fig. 8, the feedback gain G is set to_dWhen 10 is set, the vibration can be reduced to about 1/11.

Fig. 9 is a flowchart showing an operation example of the convergence gain changing unit 34, the periodic disturbance observer 32, the arithmetic unit 33, and the like of the controller 20 according to the first embodiment.

As shown in the drawing, in the controller 20, the convergence gain change unit 34 determines the target angular velocity d θ instructed to the controller 20_DWhether the absolute value of/dt is below a threshold (e.g., 0.01rad/s) or not (step 101). If it is determined that the target angular velocity d θ is reached_DWhen the absolute value of/dt is equal to or less than the threshold, the convergence gain changer 34 changes the convergence gain ζ in the periodic disturbance observer 32 (step 102).

Next, the periodic disturbance observer 32 calculates the estimated value d of the vibration torque by performing a filtering process for extracting only the vibration component of the natural frequency ω from the disturbance torque d_ωAnd (step 103). That is, when the convergence gain ζ is changed in step 102, the vibration torque estimated value d is calculated using the changed convergence gain ζ_ωAnd a. On the other hand, if the convergence gain ζ is not changed, the estimated vibration torque value d is calculated using the convergence gain ζ of a predetermined value_ω^。

Next, the computing unit 33 estimates the value d based on the vibration torque calculated in step 103_ωAnd (E) calculating vibration suppression torque (step 104).

As a result, the controller 20 outputs the command torque τ corrected based on the vibration suppression torque calculated in step 104 to the motor 2 (step 105).

[ second embodiment ]

In the first embodiment, the value d is estimated in the vibration torque_ωThe periodic disturbance observer 32 is used for estimation, but the same result is obtained by using an estimation unit (hereinafter referred to as "least square estimation unit") that performs least square estimation of a forgetting factor and weighting in the second embodiment.

Fig. 10 is a block diagram showing a configuration example of a machine control system 200 according to a second embodiment. As shown, machine control system 200 includes a machine system 10 and a controller 20 that controls machine system 10. The controller 20 realizes each function by, for example, a CPU (not shown) reading and executing a program from a storage unit (not shown) such as a ROM. In the present embodiment, the controller 20 is provided as an example of the vibration suppression device.

The functional configuration of the machine system 10 is the same as that described in the first embodiment, and therefore, the description thereof is omitted.

Next, a functional configuration of the controller 20 will be explained.

As shown in the figure, the controller 20 includes a PID control portion 21, an inertia compensation portion 22, and a friction compensation portion 23. These configurations are also the same as those described in the first embodiment, and therefore, the description thereof is omitted.

Further, the controller 20 includes an oscillating torque estimating unit 40. The vibration torque estimation unit 40 obtains the motor angular velocity d θ output from the machine system 10, as in the vibration torque estimation unit 30 of the first embodiment_MDt and an estimated value d of vibration torque generated by the elastic body 3 and acting on the motor 2_ωAnd a. Specifically, the vibration torque estimation unit 40 includes a calculation unit 41 and a least square estimation unit 42.

The computing unit 41 obtains the motor angular velocity d θ output from the machine system 10_MAnd/dt. Then, the motor angular velocity d theta is adjusted_MDt times arm inertia J_AAnd motor inertia J_MThe result of the sum of the two is differentiated to output an inertial force.

The least square estimation unit 42 obtains the disturbance torque d that is output from the friction compensation unit 23 and the inertial force output from the calculation unit 41The sum of the two torques based on the friction model (2) is obtained by subtracting the command torque τ to the motor 2. Then, the disturbance torque d is subjected to weighted least square estimation and forgetting factor represented by the following equation to calculate the vibration torque estimated value d_ωAnd outputs the estimated value d of the vibration torque obtained by the calculation_ω^。

[ number 3]

Here, g (t) is a weight at time t, ρ is a forgetting coefficient, and d (t) is an initial interference value at time t.

In the present embodiment, the estimated value d of the vibration torque is used as an example of the measurement related to the mechanical system_ωAn oscillation torque estimating unit 40 is provided as an example of an estimating means for estimating the measurement amount.

The controller 20 includes an arithmetic unit 43. The calculation unit 43 obtains the estimated value d of the vibration torque output from the vibration torque estimation unit 40_ωAnd a. Then, the value d is estimated by estimating the vibration torque_ωMultiplying feedback gain G_dThe vibration suppression torque is calculated, and the calculated vibration suppression torque is output.

Thus, the feedforward torque based on the inertia model output by the inertia compensation unit 22 is added to the feedback torque output by the PID control unit 21, and the vibration suppression torque output by the calculation unit 43 is added to the result. In this sense, the arithmetic unit 43 is an example of a correction mechanism that corrects the drive signal based on the measurement. Then, the feedforward torque based on the friction model output by the friction compensation unit 23 is added to the result, and the result becomes the command torque τ for the motor 2.

However, in the second embodiment, the controller 20 includes the weight changing unit 44 in addition to these configurations. Weight changing unit 44 and target angular velocity d θ instructed to controller 20_DThe weight g (t) of the formula used in the least square estimation unit 42 is changed in accordance with the value/dt, thereby obtaining the same result as the first embodiment. Specifically, when the target angular velocity d θ_DIn dt hours, the weight g (t) is reduced. In the second embodiment, the same graphs as those of fig. 5 to 8 can be obtained, but in these graphs, the target angular velocity d θ is set to be equal to_DWhen the absolute value of/dt is 0.01rad/s or less, the influence of modeling error is eliminated by setting the weight g (t) to 0. In the present embodiment, the target angular velocity d θ is used as an example of a period during which the modeling error of the mechanical system increases_DWhile the time period/dt is reduced, the weight g (t) is used as an example of the gain used in the estimation means, and the weight changing unit 44 is provided as an example of changing means for changing the gain so that the influence of the increase in the modeling error is reduced while the modeling error of the mechanical system is increased.

Fig. 11 is a flowchart showing an example of the operation of the weight changing unit 44, the least square estimating unit 42, the calculating unit 43, and the like of the controller 20 according to the second embodiment.

As shown in the figure, in the controller 20, the weight changing unit 44 determines the target angular velocity d θ instructed to the controller 20_DWhether the absolute value of/dt is below a threshold (e.g., 0.01rad/s) (step 201). If it is determined that the target angular velocity d θ is reached_DWhen the absolute value of/dt is equal to or less than the threshold, the weight changing unit 44 changes the weight g (t) in the least square estimating unit 42 (step 202).

Next, the least square estimation unit 42 calculates the vibration torque estimation value d by performing least square estimation of the interference torque d using a forgetting factor and weighting_ωAnd (step 203). That is, when the weight g (t) is changed in step 202, the vibration torque estimated value d is calculated using the changed weight g (t)_ωAnd a. On the other hand, if the weight g (t) is not changed, the weight of the predetermined value is usedG (t) to calculate an estimated value d of vibration torque_ω^。

Next, the calculation unit 43 estimates the value d based on the vibration torque calculated in step 203_ωAnd ^ calculating vibration suppression torque (step 204).

As a result, the controller 20 outputs the command torque τ corrected based on the vibration suppression torque calculated in step 204 to the motor 2 (step 205).

[ third embodiment ]

In the first and second embodiments, the vibration is suppressed by the resonance ratio control that is controlled so as to be weaker than the estimated vibration torque, but in the third embodiment, the same vibration suppression effect is obtained by negatively feeding back the differential value of the vibration torque. In this case, the estimated value of the vibration torque, which is the output of the periodic disturbance observer, may be differentiated and fed back, or the estimated value of the vibration torque differential value (hereinafter referred to as "estimated value of the vibration torque differential value") may be output to the periodic disturbance speed observer and fed back.

Fig. 12 is a block diagram showing a configuration example of a case where a periodic disturbance speed observer is configured in the machine control system 300 according to the third embodiment. As shown, machine control system 300 includes a machine system 10 and a controller 20 that controls machine system 10. The controller 20 realizes each function by reading a program from a storage unit (not shown) such as a ROM and executing the program, for example, by a CPU (not shown). In the present embodiment, the controller 20 is provided as an example of the vibration suppression device.

The functional configuration of the machine system 10 is the same as that described in the first and second embodiments, and therefore, the description thereof is omitted.

Next, a functional configuration of the controller 20 will be explained.

As shown in the figure, the controller 20 includes a PID control portion 21, an inertia compensation portion 22, and a friction compensation portion 23. These configurations are also the same as those described in the first and second embodiments, and therefore, the description thereof is omitted.

Further, the controller 20 includes an oscillation torque differential value estimating unit 50. Oscillation torque differential value estimation unit 50 obtains motor angular velocity d θ output from machine system 10_MAnd/dt, and outputs an estimated value of differential value of vibration torque (dd), which is an estimated value of differential value of vibration torque generated in the elastic body 3 and acting on the motor 2_ωDt). Specifically, the vibration torque differential value estimation unit 50 includes an arithmetic unit 51 and a periodic disturbance speed observer 52.

The computing unit 41 obtains the motor angular velocity d θ output from the machine system 10_MAnd/dt. Then, the motor angular velocity d theta is adjusted_MDt times arm inertia J_AAnd motor inertia J_MThe result of the sum of the two is differentiated to output an inertial force.

The periodic disturbance speed observer 52 obtains a disturbance torque d obtained by subtracting the command torque τ to the motor 2 from the sum of the inertia force output from the arithmetic unit 51 and the torque based on the friction model output from the friction compensation unit 23. Then, a filtering process for extracting a vibration velocity component of the natural frequency ω represented by the following equation is performed on the disturbance torque d to calculate an estimated value (dd) of a differential value of the vibration torque_ωDt ^ and outputs the calculated vibration torque differential value estimation value (dd)_ω/dt)^。

[ number 4]

In the present embodiment, as an example of the measurement related to the mechanical system, the estimated value of the differential value of the vibration torque (dd) is used_ωAnd/dt) ^ is provided with an oscillation torque differential value estimation unit 50 as an example of an estimation means for estimating the measurement amount.

The controller 20 includes a calculation unit 53. Operation unit 53 obtains the estimated value of the differential value of the vibration torque (dd) outputted from vibration torque differential value estimation unit 50_ωDt). Then, the vibration torque differential value is estimated as (dd)_ω/dt) multiplied by feedback gain G_dvThereby carryingThe vibration suppression torque is calculated, and the calculated vibration suppression torque is output.

Thus, the feedforward torque based on the inertia model output by the inertia compensation unit 22 is added to the feedback torque output by the PID control unit 21, and the vibration suppression torque output by the calculation unit 53 is subtracted from the result. In this sense, the arithmetic unit 53 is an example of a correction mechanism that corrects the drive signal based on the measurement. Then, the feedforward torque based on the friction model output by the friction compensation unit 23 is added to the result, and the result becomes the command torque τ for the motor 2.

However, in the third embodiment, the controller 20 includes the convergence gain changing unit 54 in addition to these configurations. The convergence gain changing unit 54 compares the target angular velocity d θ instructed to the controller 20 with the target angular velocity d θ_DThe value of ζ corresponding to the convergence gain of the filter used in the periodic disturbance speed observer 52 is corrected accordingly, thereby preventing the influence of the turning-back time when the modeling error including the high-frequency component increases. Specifically, when the target angular velocity d θ_DAnd when the/dt is small, the convergence gain zeta is reduced. In the third embodiment, the same graphs as those of fig. 6 to 8 can be obtained, but in these graphs, the target angular velocity d θ is set to be equal to_DWhen the absolute value of/dt is 0.01rad/s or less, the convergence gain ζ is set to 0, and the influence of the modeling error is eliminated. In the present embodiment, the target angular velocity d θ is used as an example of a period during which the modeling error of the mechanical system increases_DWhile dt is decreasing, a convergent gain ζ is used as an example of a gain used in the estimating means, and a convergent gain changing unit 54 is provided as an example of a changing means that changes a gain so that an influence of an increase in a modeling error of the mechanical system is decreased while the modeling error is increasing.

Fig. 13 is a flowchart showing an operation example of the convergence gain changing unit 54, the periodic disturbance speed observer 52, the calculating unit 53, and the like of the controller 20 according to the third embodiment.

As shown in the drawing, in the controller 20, the convergence gain change unit 54 determines the target angular velocity d θ instructed to the controller 20_DWhether the absolute value of/dt is a threshold value(e.g., 0.01rad/s) or less (step 301). If it is determined that the target angular velocity d θ is reached_DWhen the absolute value of/dt is equal to or less than the threshold value, the convergence gain changer 54 changes the convergence gain ζ in the periodic disturbance speed observer 52 (step 302).

Next, the periodic disturbance speed observer 52 performs a filtering process for extracting only the vibration speed component of the natural frequency ω from the disturbance torque d to calculate the vibration torque differential value estimation value (dd)_ωDt ^ (step 303). That is, when the convergence gain ζ is changed in step 302, the vibration torque differential value estimation value (dd) is calculated using the changed convergence gain ζ_ωDt). On the other hand, if the convergence gain ζ is not changed, the estimated value (dd) of the differential value of the vibration torque is calculated using the convergence gain ζ of a predetermined value_ω/dt)^。

Next, the computing unit 53 estimates a value (dd) based on the differential value of the vibration torque calculated in step 303_ωDt) and calculates a vibration suppression torque (step 304).

As a result, the controller 20 outputs the command torque τ corrected based on the vibration suppression torque calculated in step 304 to the motor 2 (step 305).

[ fourth embodiment ]

The fourth embodiment is an embodiment in which, when least square estimation is used as in the second embodiment, a differential value of an estimated value of vibration torque (hereinafter referred to as "estimated value differential value of vibration torque") is negatively fed back to obtain the same vibration suppression effect.

Fig. 14 is a block diagram showing a configuration example of a machine control system 400 according to the fourth embodiment. As shown, machine control system 400 includes a machine system 10 and a controller 20 that controls machine system 10. The controller 20 realizes each function by reading a program from a storage unit (not shown) such as a ROM and executing the program, for example, by a CPU (not shown). In the present embodiment, the controller 20 is provided as an example of the vibration suppression device.

The functional configuration of the machine system 10 is the same as that described in the first to third embodiments, and therefore, the description thereof is omitted.

Next, a functional configuration of the controller 20 will be explained.

As shown in the figure, the controller 20 includes a PID control portion 21, an inertia compensation portion 22, and a friction compensation portion 23. These configurations are also the same as those described in the first to third embodiments, and therefore, description thereof is omitted.

Further, the controller 20 includes an oscillating torque estimating unit 60. The vibration torque estimation unit 60 obtains the motor angular velocity d θ output from the machine system 10, as in the vibration torque estimation unit 40 of the second embodiment_MDt and an estimated value d of vibration torque generated by the elastic body 3 and acting on the motor 2_ωAnd a. Specifically, the vibration torque estimation unit 60 includes a calculation unit 61 and a least square estimation unit 62.

The computing unit 61 obtains the motor angular velocity d θ output from the machine system 10_MAnd/dt. Then, the motor angular velocity d theta is adjusted_MDt times arm inertia J_AAnd motor inertia J_MThe result of the sum of the two is differentiated to output an inertial force.

The least square estimation unit 62 obtains the disturbance torque d obtained by subtracting the command torque τ to the motor 2 from the sum of the inertia force output from the calculation unit 61 and the torque based on the friction model output from the friction compensation unit 23. Then, the disturbance torque d is subjected to weighted least square estimation and forgetting factor represented by the following equation to calculate the vibration torque estimated value d_ωAnd outputs the estimated value d of the vibration torque obtained by the calculation_ω^。

[ number 5]

Here, g (t) is a weight at time t, ρ is a forgetting coefficient, and d (t) is an initial interference value at time t.

The controller 20 includes an arithmetic unit 63 and an arithmetic unit 64.

The calculation unit 63 obtains the estimated value d of the vibration torque output from the vibration torque estimation unit 60_ωAnd a. Then, the value d is estimated by estimating the vibration torque_ωDifferentiating to calculate the differential value d (d) of the estimated value of the vibration torque_ωA/dt, and outputs a differential value d (d) of the estimated value of the vibration torque obtained by the calculation_ω^)/dt。

In the present embodiment, as an example of the measurement related to the mechanical system, the differential value d (d) of the estimated value of the vibration torque is used_ωThe vibration torque estimation unit 30 and the calculation unit 63 are provided as an example of an estimation means for estimating the measurement amount.

The computing unit 64 obtains the differential value d (d) of the estimated value of the vibration torque output from the computing unit 63_ωAnd a)/dt. Then, the vibration torque is estimated by differentiating the value d (d)_ω^)/dt multiplied by feedback gain G_dvThe vibration suppression torque is calculated, and the calculated vibration suppression torque is output.

Thus, the feedforward torque based on the inertia model output by the inertia compensation unit 22 is added to the feedback torque output by the PID control unit 21, and the vibration suppression torque output by the calculation unit 64 is subtracted from the result. In this sense, the arithmetic unit 64 is an example of a correction mechanism that corrects the drive signal based on the measurement. Then, the feedforward torque based on the friction model output by the friction compensation unit 23 is added to the result, and the result becomes the command torque τ for the motor 2.

However, in the fourth embodiment, the controller 20 includes the weight changing unit 65 in addition to these configurations. Weight changing unit 65 and target angular velocity d θ instructed to controller 20_DThe weight g (t) of the formula used by the least square estimation unit 62 is changed in accordance with the value/dt, thereby obtaining the same result as the third embodiment. Specifically, when the target angular velocity d θ_DThe weight g (t) is reduced in the dt hours. In the fourth embodiment, the same graphs as those of fig. 6 to 8 can be obtained, but in these graphs, the target angular velocity d θ is set to be equal to_DWhen the absolute value of/dt is 0.01rad/s or less, the influence of modeling error is eliminated by setting the weight g (t) to 0. In the present embodiment, the target angular velocity d θ is used as an example of a period during which the modeling error of the mechanical system increases_DWhile the time period/dt is reduced, the weight g (t) is used as an example of the gain used in the estimation means, and the weight changing unit 65 is provided as an example of changing means for changing the gain so that the influence of the increase in the modeling error is reduced while the modeling error of the mechanical system is increased.

Fig. 15 is a flowchart showing an example of operations of the weight changing unit 65, the least square estimating unit 62, the calculating unit 63, the calculating unit 64, and the like of the controller 20 according to the fourth embodiment.

As shown in the figure, in the controller 20, the weight changing unit 65 determines the target angular velocity d θ instructed to the controller 20_DWhether the absolute value of/dt is below a threshold (e.g., 0.01rad/s) (step 401). If it is determined that the target angular velocity d θ is reached_DIf the absolute value of/dt is equal to or less than the threshold, the weight changing unit 65 changes the weight g (t) in the least square estimating unit 62 (step 402).

Next, the least square estimation unit 62 calculates an estimated value d of the vibration torque by performing least square estimation of a forgetting factor and weighting on the disturbance torque d_ωAnd (step 403). That is, when the weight g (t) is changed in step 402, the vibration torque estimated value d is calculated using the changed weight g (t)_ωAnd a. On the other hand, if the weight g (t) is not changed, the estimated value d of the vibration torque is calculated using the weight g (t) of a predetermined value_ω^。

Then, the calculation unit 63 estimates the vibration torque estimated value d calculated in step 403_ωA is advancedDifferentiating to obtain an estimated differential d (d) of the vibration torque_ωA)/dt (step 404).

Next, the arithmetic unit 64 estimates the differential value d (d) based on the vibration torque calculated in step 404_ωAnd a vibration suppression torque is calculated according to the relationship between the values of the two frequencies (step 405).

As a result, the controller 20 outputs the command torque τ corrected based on the vibration suppression torque calculated in step 405 to the motor 2 (step 406).

[ fifth embodiment ]

Although the observer for estimating the disturbance is configured in the first to fourth embodiments, the arm angular velocity d θ is configured in the fifth embodiment_AAnd/dt is estimated.

Fig. 16 is a block diagram showing a configuration example of a machine control system 500 according to a fifth embodiment. As shown, machine control system 500 includes a machine system 10 and a controller 20 that controls machine system 10. The controller 20 realizes each function by reading a program from a storage unit (not shown) such as a ROM and executing the program, for example, by a CPU (not shown). In the present embodiment, the controller 20 is provided as an example of the vibration suppression device.

The functional configuration of the machine system 10 is the same as that described in the first to fourth embodiments, and therefore, the description thereof is omitted.

Next, a functional configuration of the controller 20 will be explained.

As shown in the figure, the controller 20 includes a PID control portion 21, an inertia compensation portion 22, and a friction compensation portion 23. These configurations are also the same as those described in the first to fourth embodiments, and therefore, description thereof is omitted.

Further, the controller 20 includes an arm angular velocity estimating section 70. The arm angular velocity estimating unit 70 obtains the motor angular velocity d θ output from the machine system 10_MDt and outputs an arm angular velocity estimated value (d theta)_ADt). Specifically, the arm angular velocity estimation unit 70 includes an arm angular velocity estimation observer 71.

The arm angular velocity estimation observer 71 acquires the output from the machine system 10Angular velocity d θ of motor_MAnd a torque τ M obtained by subtracting the torque based on the friction model output by the friction compensation unit 23 from the command torque τ for the motor 2. Then, the angular velocity d theta of the motor is adjusted_MThe arm angular velocity estimation value (d θ) is calculated by performing the following processing on the torque τ M and/dt_ADt ^ and outputs the calculated estimated arm angular velocity value (d θ)_A/dt)^。

FIG. 17 is a view showing a motor for varying an angular velocity d θ according to a motor_MCalculating an arm angular velocity estimate (d θ) from the torque τ M and/or the dt_ABlock diagram of/dt) ^ flow.

As shown in the figure, in the arm angular velocity estimation observer 71, the calculation unit 711 compares the motor angular velocity d θ_MDifferentiating the angular acceleration d of the motor²θ_M/dt². In addition, the calculation unit 712 multiplies the torque τ M by 1/J_MAnd τ M/J is calculated_M. The arithmetic unit 713 multiplies ^ epsilon calculated as described later by K/J_MTo operate K epsilon/J_M. Thus, the calculation unit 714 determines the angular acceleration d of the slave motor²θ_M/dt²Minus τ M/J_MAnd adding K epsilon/J to the result_MAnd the result is multiplied by the convergence gain L₁To calculate L₁×(d²θ_M/dt²-τM/J_M+Kε^/J_M)。

On the other hand, the arithmetic unit 715 determines the angular acceleration d of the slave motor²θ_M/dt²Minus τ M/J_MAnd adding K epsilon/J to the result_MAnd the result is multiplied by the convergence gain L₂To calculate L₂×(d²θ_M/dt²-τM/J_M+Kε^/J_M). The arithmetic unit 716 determines the angular velocity d θ of the motor_M(dt) plus L₂×(d²θ_M/dt²-τM/J_M+Kε^/J_M) And subtracting the arm angular velocity estimation value (d theta) from the result_AThe result of the operation is integrated to calculate ε ^ by the integration of the results obtained by the operation of/dt ^ with the integer. The operation unit 717 multiplies ε ^ by K/J_ATo operate K epsilon/J_A。

Thus, the arithmetic unit 718 compares L with L₁×(d²θ_M/dt²-τM/J_M+Kε^/J_M) Plus K epsilon ^/J_AThe obtained result is integrated to calculate an arm angular velocity estimation value (d theta)_A/dt)^。

In this way, the arm angular velocity estimation observer 71 is realized in the form of a minimum dimension observer. In addition, the state quantity d θ is not normally used_MDifferential value d of/dt²θ_M/dt²And is replaced with the output value τ M, f, but in this embodiment, the state quantity d θ is retained by the output value τ M, f without being replaced_MDifferential value d of/dt²θ_M/dt²In the form of (1). This is an example of estimating the measured quantity using a differential value of the state quantity in the state equation of the mechanical system. In addition, such a method can be applied not only to the fifth embodiment but also to the first to fourth embodiments.

In the present embodiment, the arm angular velocity estimation value (d θ) is used as an example of the measurement related to the mechanical system_AAnd dt) as an example of an estimation mechanism for estimating the measurement amount, an arm angular velocity estimation unit 70 is provided.

Referring again to fig. 16, the controller 20 includes an arithmetic section 72. The computing unit 72 obtains the arm angular velocity estimation value (d θ) output from the arm angular velocity estimating unit 70_ADt). Then, the arm angular velocity is estimated by the estimated value (d θ)_AThe feedback gain Gv is multiplied by/dt ^ to calculate vibration suppression torque, and the calculated vibration suppression torque is output.

Thus, the feedforward torque based on the inertia model output by the inertia compensation unit 22 is added to the feedback torque output by the PID control unit 21, and the vibration suppression torque output by the calculation unit 72 is subtracted from the result. In this sense, the arithmetic unit 72 is an example of a correction mechanism that corrects the drive signal based on the measurement. Then, the feedforward torque based on the friction model output by the friction compensation unit 23 is added to the result, and the result becomes the command torque τ for the motor 2.

However, in the fifth embodiment, the controller 20 includes the convergence gain changing unit 73 in addition to these configurations. Convergence gain changing unit 73 and target angular velocity d θ instructed to controller 20_DThe convergence gain L used in the arm angular velocity estimation observer 71 is changed in accordance with/dt₁、L₂. Specifically, when the target angular velocity d θ_DReduction of convergence gain L in dt hours₁、L₂And (4) finishing. At the target angular velocity d θ_DWhen the absolute value of/dt is less than 0.01rad/s, the gain L is converged₁、L₂Set to 0 and exclude the effect of modeling errors. In the present embodiment, the target angular velocity d θ is used as an example of a period during which the modeling error of the mechanical system increases_DWhile the dt is decreasing, the convergence gain L is used as an example of the gain used in the estimation means₁、L₂The convergence gain changing unit 73 is provided as an example of a changing mechanism that changes the gain so as to reduce the influence of an increase in the modeling error while the modeling error of the mechanical system is increasing.

FIGS. 18 and 19 show the unpaired convergence gain L₁、L₂With target angular velocity d theta_DAnd/dt is changed accordingly. Fig. 18 shows the difference between the actual value of the arm angular velocity and the estimated value of the arm angular velocity. From this figure, it is understood that the arm angular velocity estimated value is deviated from the actual value of the arm angular velocity due to the influence of the modeling error. Fig. 19 shows the result of feedback using the arm angular velocity estimation value of fig. 18. As is apparent from this figure, the motor angular velocity is greatly deviated from the target angular velocity by feeding back the arm angular velocity estimated value deviated from the actual value of the arm angular velocity. From this, it is estimated that the arm angular velocity also greatly deviates from the target angular velocity of the arm 1.

On the other hand, fig. 20 and 21 show the convergence gain L₁、L₂With target angular velocity d theta_DAnd/dt is changed accordingly, and the simulation result is shown. Fig. 20 shows the actual value of the arm angular velocity and the estimated value of the arm angular velocityThe difference in (a). From this figure, it is understood that the estimated arm angular velocity value substantially matches the actual arm angular velocity value can be obtained. Fig. 21 shows the result of feedback using the arm angular velocity estimation value of fig. 20. As is clear from this figure, a good vibration suppression effect can be obtained by feeding back an arm angular velocity estimated value that substantially matches the actual value of the arm angular velocity.

Fig. 22 is a flowchart showing an operation example of the convergence gain changing unit 73, the arm angular velocity estimation observer 71, the computing unit 72, and the like of the controller 20 according to the fifth embodiment.

As shown in the figure, in the controller 20, the convergence gain change unit 73 determines the target angular velocity d θ instructed to the controller 20_DWhether the absolute value of/dt is below a threshold (e.g., 0.01rad/s) (step 501). If it is determined that the target angular velocity d θ is reached_DWhen the absolute value of/dt is equal to or less than the threshold value, the convergence gain changing unit 73 changes the convergence gain L in the arm angular velocity estimation observer 71₁、L₂(step 502).

Next, the arm angular velocity estimation observer 71 calculates an arm angular velocity estimation value (d θ) by performing the processing shown in fig. 17_ADt ^ (step 503). That is, if the gains L1, L converge in step 502₂Is changed, the changed convergence gain L is used₁、L₂To calculate an arm angular velocity estimation value (d theta)_ADt). On the other hand, if the gain L is converged₁、L₂Without being changed, a predetermined value of convergence gain L is used₁、L₂To calculate an arm angular velocity estimation value (d theta)_A/dt)^。

Next, the calculation unit 72 estimates the arm angular velocity (d θ) based on the arm angular velocity calculated in step 503_ADt) and calculates a vibration suppression torque (step 504).

Thus, the controller 20 outputs the command torque τ corrected based on the vibration suppression torque calculated in step 504 to the motor 2 (step 505).

[ modified examples ]

In the first to fifth embodiments, the value d is estimated based on the vibration torque_ω^ d, vibration torque differential value estimation value (dd)_ω/dt)A differential value d (d) of the estimated value of vibration torque_ωA/dt and an arm angular velocity estimation value (d theta)_AAnd/dt) to calculate the vibration suppressing torque, but is not limited thereto. For example, the vibration suppression torque may be calculated based on the estimated value (d ∈/dt) ^ of the elastic deformation speed of the elastic body 3.

Description of reference numerals:

10 a mechanical system;

11. 12, 13, 14, 15 arithmetic units;

20 a controller;

21 PID control part;

22 an inertia compensation unit;

23 a friction compensation part;

30. 40, 60 vibration torque estimation part;

31. 33, 41, 43, 51, 53, 61, 63, 64, 72 calculation units;

a 32-cycle disturbance observer;

34. 54, 73 convergence gain changing parts;

42. 62 a least squares estimation unit;

44. 65 weight changing part;

a 50 vibration torque differential value estimating unit;

a 52-cycle disturbance speed observer;

a 70-arm angular velocity estimating unit;

71 an arm angular velocity estimation observer;

100. 200, 300 mechanical control system.