JP2018173702A - Parameter identification device, drive system, parameter identification method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a parameter identification device, a drive system, a parameter identification method, and a program that can obtain a model which accurately imitates a mechanical system including a motor and a load.SOLUTION: A parameter identification device comprises: a data obtaining unit that obtains a torque command, an angle and angular velocity of a motor, an angle and angular velocity of a load; a phase plane map generating unit that generates a phase plane map on the basis of the torque command, the angle and the angular velocity of the motor, the angle and the angular velocity of the load; and a parameter identifying unit that identifies inertia, viscosity, and friction of the motor, inertia, viscosity, and friction of the load, rigidity of a connection member, and a model parameter showing a dead zone width of the connection member on the basis of the phase plane map. The parameter identifying unit identifies the model parameter on the basis of both of the phase plane map obtained in a small amplitude operation of the load and the phase plane map obtained in a large amplitude operation of the load.SELECTED DRAWING: Figure 6

Description

  The present invention relates to a parameter identification device, a drive system, a parameter identification method, and a program.

In designing a drive system that drives a load using a motor, there is a limit in feedback control in order to improve control performance, and it is necessary to perform feedforward control with high response characteristics to a command signal. In order to realize feedforward control, the control side needs to have an inverse model of a model imitating a controlled object (that is, a mechanical system having a motor and a load). If there is an error with the actual characteristics of the control target, the control performance cannot be improved.
Therefore, in order to realize feedforward control with high control performance, it is necessary to obtain a model that accurately simulates the control target.

  Patent Document 1 discloses a method of performing feedforward control based on an inverse model characteristic of a two-inertia system model (two-mass system model). The 2-inertia model is a model imitating a mechanical system having, for example, a motor, a load, and a connecting member (ball screw, gear, etc.) that connects the motor and the load. By constructing such a two-inertia model, the characteristics of a mechanical system having a load and a motor can be imitated more accurately.

JP 2002-023852 A

The conventional two-inertia system model includes only linear elements such as inertia and viscosity, and often does not include non-linear elements such as motor-side friction, load-side friction, and backlash generated in the connecting member. The current situation is that non-linear elements are dealt with separately by empirical adjustment.
Such a two-inertia system model that does not include a nonlinear element cannot be accurately imitated by a mechanical system having a motor and a load. Therefore, it is not possible to obtain a model that accurately simulates the object to be controlled, and it is therefore impossible to realize a drive system with high control performance.

  An object of the present invention has been made in view of the above problems, and provides a parameter identification device, a drive system, a parameter identification method, and a program capable of accurately obtaining a model imitating a mechanical system having a motor and a load. There is to do.

  According to the first aspect of the present invention, a parameter identification device for identifying a model parameter of a two-inertia system model imitating a mechanical system in which a motor and a load are coupled by a coupling member includes: a torque command for the motor; A measured value of the angle and angular velocity of the motor, a measured value of the angle and angular velocity of the load, a data acquisition unit that acquires the acquired torque command, and a measured value of the angle and angular velocity of the motor, A phase plane generator that generates a plurality of phase planes based on the measured values of the load angle and angular velocity, and the inertia, viscosity, and friction of the motor based on the generated phase planes A parameter identification unit that identifies model parameters indicating inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member. The parameter identification unit identifies the model parameter based on both the phase plane acquired during the small amplitude operation of the load and the phase plane acquired during the large amplitude operation of the load.

  According to the second aspect of the present invention, the parameter identification unit is configured to show at least the friction of the motor and the friction of the load based on the phase plane acquired during the small amplitude operation of the load. A parameter is identified, and model parameters indicating at least the inertia and viscosity of the motor, the inertia of the load, and the viscosity are identified based on the phase plane acquired during the large amplitude operation of the load.

  Further, according to the third aspect of the present invention, the phase plane generation unit includes at least a phase plane showing a relationship between the load angle and the torque command, the angular velocity of the motor or the load, and the A phase diagram showing the relationship with the torque command, a phase diagram showing the relationship between the angle of the load and the twist angle of the connecting member, and the relationship between the angular velocity of the motor or the load and the twist angle of the connecting member And a phase plane diagram showing

  According to the fourth aspect of the present invention, the phase plane diagram generation unit further generates a phase plane diagram showing a relationship between the torque command and a twist angle of the connecting member.

  According to the fifth aspect of the present invention, the drive system outputs the torque command based on the parameter identification device, the mechanical system, and a deviation of a current angle with respect to a target angle of the load. A feedback control unit for calculating, and a feedforward control unit for calculating the torque command based on the model parameter identified by the parameter identification device.

  According to a sixth aspect of the present invention, a parameter identification method is a method for identifying model parameters of a two-inertia system model simulating a mechanical system in which a motor and a load are coupled by a coupling member. A data acquisition step for acquiring a torque command for the motor, an actual value of the angle and angular velocity of the motor, and an actual value of the load angle and angular velocity; the acquired torque command; and the angle of the motor And a phase plane diagram generating step for generating a plurality of phase plane diagrams based on the measured values of the angular velocity and the measured values of the angle and angular velocity of the load, and based on the generated plurality of the phase plane diagrams, Parameter identification step for identifying model parameters indicating motor inertia, viscosity, friction, inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member And the parameter identifying step determines the model parameter based on both the phase plane acquired during the small amplitude operation of the load and the phase plane acquired during the large amplitude operation of the load. Identifying.

  Further, according to the seventh aspect of the present invention, the program stores a computer of a parameter identification device that identifies a model parameter of a two-inertia system model imitating a mechanical system in which a motor and a load are coupled by a coupling member. A data acquisition unit for acquiring a torque command for the motor, an actual value of the angle and angular velocity of the motor, and an actual value of the load angle and angular velocity, the acquired torque command, the angle of the motor, and A phase plane diagram generating unit that generates a plurality of phase plane diagrams based on the measured values of angular velocities and the actual values of the load angles and angular velocities, and based on the generated plurality of phase plane diagrams, As a parameter identification unit for identifying model parameters indicating inertia, viscosity, friction, inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member, The parameter identification unit is capacity, and the phase plane obtained at the time of a small amplitude motion of the load, based on both said phase plane obtained when a large amplitude operation of the load, to identify the model parameters.

  According to the above-described parameter identification device, drive system, parameter identification method and program, a model imitating a mechanical system having a motor and a load can be obtained with high accuracy.

It is a figure showing the whole drive system composition concerning a 1st embodiment. It is a block diagram which shows the transfer characteristic of the mechanical system which concerns on 1st Embodiment. It is a figure explaining the friction characteristic of the motor side and load side of the mechanical system which concerns on 1st Embodiment. It is a figure explaining the characteristic of a dead zone in the connection member of the mechanical system concerning a 1st embodiment. It is a figure which shows the function structure of the drive system which concerns on 1st Embodiment. It is a figure which shows the function structure of the parameter identification apparatus which concerns on 1st Embodiment. It is a figure which shows the processing flow of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 1 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 2 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 3 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 4 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 5 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 6 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 7 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 8 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 9 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment. It is FIG. 10 for demonstrating the process of the parameter identification apparatus which concerns on 1st Embodiment.

<First Embodiment>
Hereinafter, the drive system according to the first embodiment will be described in detail with reference to FIGS.

(overall structure)
FIG. 1 is a diagram illustrating an overall configuration of a drive system according to the first embodiment.
As shown in FIG. 1, the drive system 9 according to the first embodiment includes a control device 1 and a mechanical system 2 controlled by the control device 1.
The mechanical system 2 includes a motor 20, a load 21, and a connecting member 22 that mechanically connects the motor 20 and the load 21. The mechanical system 2 is a drive mechanism in which the load 21 is carried to a desired position (angle) when the motor 20 is driven to rotate, and is a machine tool, for example. The motor 20 is a servo motor that has a rotation detector (encoder) capable of detecting a rotation angle and a rotation speed therein and can realize precise positioning. The connecting member 22 is a member such as a gear, a ball screw, or a ball bearing.

As shown in FIG. 1, the control device 1 receives a load angle command θ _REF indicating a target angle of the load 21 from a host device (not shown). The control device 1 calculates a torque τ to be generated in the motor 20 according to the received load angle command θ _REF , and outputs a torque command τ indicating the torque τ to the motor 20. The motor 20 is rotationally driven with the torque τ indicated by the received torque command τ, and the torque τ is transmitted from the motor 20 side to the load 21 side through the connecting member 22. As a result, the load 21 is rotationally driven according to the torque τ ′ transmitted to the load 21. In this way, the control device 1 performs control such that the rotation angle of the load 21 (hereinafter also simply referred to as “angle”) matches the desired target angle θ _REF (load angle command θ _REF ). .
In the present embodiment, for simplification of description, the motor 20 is described as being capable of outputting the torque τ according to the torque command τ when the torque command τ is received from the control device 1. .

The control device 1 according to the first embodiment is a so-called full-closed system, and performs feedback control based on an actual measurement value (detection value) of the angular velocity ω _{M of the} motor 20 and the angle θ _{L of the} load 21.
Here, as described above, a rotation detector (encoder) (not shown) is installed in each of the motor 20 and the load 21 constituting the mechanical system 2. The control device 1 detects the angle θ _{M of the} motor 20 and the angle θ _{L of the} load 21 through the rotation detector. Further, the control device 1 can detect the angular velocity ω _M of the motor 20 by differentiating the acquired detected value of the angle θ _M of the motor 20 with respect to time. Moreover, the control apparatus 1 makes it possible to detect the angular velocity ω _L of the load 21 by time-differentiating the acquired detection value of the angle θ _L of the load 21 internally.
In the following description, the detected value of the angular velocity of the motor 20 is also referred to as “motor angular velocity ω _M ”, and the detected value of the angle of the motor 20 is also referred to as “motor angle θ _M ”. The detected value of the angular velocity of the load 21 is also referred to as “load angular velocity ω _L ”, and the detected value of the angle of the load 21 is also referred to as “load angle θ _L ”.

The control device 1 further has a two-inertia system model (described later) imitating the mechanical system 2 as a two-inertia system in advance, and performs feedforward control using the two-inertia system model.
When the load 21 in the mechanical system 2 is not a rotating system but a linear motion system, the parameters to be controlled for the load 21 are not strictly “angle” and “angular velocity” but “position” and “speed”. . However, in this case, the control device 1 treats the “position” and “velocity” of the load as motor shaft conversion values as a two-inertia system in order to handle the “angle” and “angular velocity” of the motor 20 in the same dimension. Various control is performed by sequentially converting to “angle” and “angular velocity”.

When the mechanical system 2 is regarded as a two-inertia system, the parameter group representing the characteristics unique to the mechanical system 2 includes a parameter indicating the characteristics on the motor 20 side, a parameter indicating the characteristics on the load 21 side, and the motor 20 and the load. 21 is roughly divided into parameters indicating the characteristics of the connecting member 22 that mechanically connects 21 and transmits power. Specifically, as parameters indicating characteristics on the motor 20 side, there are a motor moment of inertia J _M , a motor viscosity coefficient D _M , and a motor friction torque τ _fM (motor friction). Further, as parameters indicating characteristics on the load 21 side, there are a load moment of inertia J _L , a load viscosity coefficient D _L , and a load friction torque τ _fL (load friction). Further, parameters indicating the characteristics of the connecting member 22 include a torsional stiffness coefficient K _R , a torsional viscosity coefficient D _R , and a dead band width BL.

(Mechanical transmission characteristics)
FIG. 2 is a block diagram showing transfer characteristics of the mechanical system according to the first embodiment.
By regarding the mechanical system 2 (see FIG. 1) according to the present embodiment as a two-inertia machine composed of the motor 20, the load 21, and the connecting member 22, the relationship between the input and output in the mechanical system 2 is shown in FIG. Can be represented by a block diagram using a transfer function as shown in FIG.

As shown in FIG. 2, the motor 20, the controller 1 inputs the torque τ which it was generated based on the torque command τ accepted (FIG. 1), and outputs the motor angle theta _M. The torque τ generated by the motor 20 is subtracted from the motor friction torque τ _fM , which is friction generated in the motor 20, and the torque τ ′ transmitted to the load 21 through the connecting member 22, and then the transmission element 1 / (J _M The motor angular velocity ω _M is converted through s + D _M ), and the motor angle θ _M is further converted through the transmission element 1 / s (integration element).
Here, the motor friction torque τ _fM is not only the Coulomb friction component whose sign is reversed when the speed of the motor 20 is reversed (when the sign of the motor angular speed ω _M is reversed), but also the displacement θ (motor It includes a nonlinear friction component that varies nonlinearly depending on the angular velocity ω _M (integral value from zero). Motor friction torque tau _{fM [theta]} defines a modeled nonlinear friction component, obtained through the motor friction characteristic function G _M to an input variable motor angular velocity omega _M. For more information about the motor friction characteristic function G _M will be described later.

Similarly, the load 21 receives the torque τ ′ (= (K _R + D _R s) (θ _M −θ _L −BKLS)) transmitted from the motor 20 through the connecting member 22, and outputs the load angle θ _L as an output. To do. The torque τ ′ input to the load 21 is converted to the load angular velocity ω _L through the transmission element 1 / (J _L s + D _L ) after subtracting the load friction torque τ _fL that is the friction generated in the load 21 and further transmitted. It is converted to the load angle theta _L through the element 1 / s.
Here, the load friction torque τ _fL similarly includes a non-linear friction component that changes non-linearly depending on the displacement θ after the speed reversal of the load 21 (the integrated value of the load angular speed ω _L from zero). Load friction torque tau _{fL [theta]} defines a modeled nonlinear friction component, obtained through the load friction characteristic function G _L to an input variable load angular velocity omega _L. Details of the load friction characteristic function _GL will be described later.

As shown in FIG. 2, the connecting member 22 receives a deviation between the motor angle θ _M and the load angle θ _L (hereinafter also referred to as “twist angle (θ _M −θ _L )”) and twists. This is a transmission system that outputs torque τ ′ corresponding to an angle (θ _M −θ _L ). Input twist angle (theta _M - [theta] _L) is dead band (hereinafter, also referred to as "backlash".) And dead zone characteristic function F representing the nonlinearity in the transfer element K _R and transmission elements D _{R s,} through , Converted to torque τ ′ applied to the motor 20 and the load 21.
Here, the dead band characteristic function F is a non-linear function defined by the dead band width BL. Dead zone characteristic function F is twist angle of the coupling member 22 a (θ _M -θ _L) as an input variable, the displacement from the twist angle (θ _M -θ _L) backlash (hereinafter, also referred to as "rattle displacement BKLS". ) Minus the angle. This dead band characteristic function F indicates that the backlash BKLS portion of the torsion angle (θ _M −θ _L ) of the connecting member 22 does not contribute to torque transmission from the motor 20 to the load 21. Details of the dead band characteristic function F will be described later.
Further, the torsional rigidity coefficient K _R, a parameter indicating the degree of rigidity of the torsional direction of the connecting member 22, corresponding to the spring constant of the connecting member 22. That is, the torsional rigidity coefficient K _R, of the torque tau 'applied to the load 21, providing a component proportional to the twist angle of the coupling member 22 (θ _M -θ _L).
Further, the twist viscosity coefficient D _R, a parameter indicating the degree of stickiness of the connecting member 22, of the torque tau 'applied to the load 21, the motor angular velocity omega _M in the connecting member 22 and the load angular velocity omega _L Component (hereinafter also referred to as “twist angular velocity (ω _M −ω _L )”).

As described above, the mechanical system 2 to be controlled by the control device 1 is characterized as a two-inertia system model by a plurality of parameter groups and a nonlinear function.
Here, the model parameter group includes the motor inertia moment J _M , the motor viscosity coefficient D _M , the load inertia moment J _L , the load viscosity coefficient D _L , the torsional stiffness coefficient K _R , the torsional viscosity coefficient D _R , and the dead zone. The width is BL.
The non-linear function is a motor friction characteristic function G _M that gives a motor friction torque τ _fM [θ] corresponding to an integral value (displacement θ) of the motor angular velocity ω _{M and} an integral value (displacement θ) of the load angular velocity ω _L. The load friction characteristic function G _L giving the corresponding load friction torque τ _fL [θ] and the torsion angle (θ _M −θ _L ) are input and the backlash displacement BKLS [θ is calculated from the torsion angle (θ _M −θ _L ). _This is a dead band characteristic function F that outputs an angle obtained by subtracting _M −θ _L ] (described later).
Further, the torque τ that is an input to the mechanical system 2 is a parameter that can be observed by the control device 1 itself from the torque command τ. The motor angle θ _M and the load angle θ _{L that} are outputs (detected values) from the mechanical system 2 are parameters that can be observed by the control device 1 through the above-described rotation detector.

(Friction characteristics)
FIG. 3 is a diagram illustrating the friction characteristics on the motor side and the load side of the mechanical system according to the first embodiment.
FIG. 3A is a graph showing nonlinear friction characteristics generated in each of the motor friction torque τ _fM [θ] and the load friction torque τ _fL [θ].
The well-known Coulomb friction changes only in the direction (positive or negative sign) depending on the direction (positive or negative sign) of the object speed (motor angular speed ω _M , load angular speed ω _L ), The amount is known not to vary with respect to speed (ω) and displacement (θ). However, when the motor 20 and the load 21 include rolling elements such as ball bearings, it is necessary to consider “rolling friction” having characteristics different from those of normal Coulomb friction.
Here, rolling friction is considered by dividing it into a “fine movement area” where some of the rolling elements do not roll effectively immediately after the speed reversal and a “coarse movement area” where all rolling elements roll effectively. The
In a “fine movement region” where rolling does not occur effectively in at least some rolling elements, it appears according to the degree of displacement (integral value from zero of motor angular velocity ω _M and load angular velocity ω _L ) after speed reversal. Has a non-linear spring characteristic in which the spring constant changes dynamically, and draws a hysteresis curve as shown in FIG.
Further, in the “coarse motion region” where all the rolling elements roll effectively, the behavior due to Coulomb friction becomes dominant and shows a static characteristic with respect to speed. That is, as shown in FIG. 3A, the friction of the motor 20 (motor friction torque τ _fM [θ]) is equal to the motor coulomb friction torque τ _fMc when the displacement θ after the speed reversal becomes a predetermined value or more. Saturates. Further, the friction of the load 21 (load friction torque τ _fL [θ]) is saturated at the load coulomb friction torque τ _fLc when the displacement θ after the speed reversal becomes a predetermined value or more. The motor Coulomb friction torque tau _FMC and load Coulomb friction torque tau _FLC is a parameter that defines a motor friction characteristic function G _M and the load friction characteristic function G _L.

In this embodiment, motor friction characteristic function G _M and the load friction characteristic function G _L is defined on the basis of the non-linear friction properties obtained by the polygonal line approximating the GMS (Generalized Maxwell Slip) model shown in FIG. 3 (a) . Here, the GMS model imitates a state in which N blocks having different characteristics (rolling elements) and a spring are connected in parallel. Each of the plurality of straight lines constituting the graph (polygonal line) shown in FIG. 3B represents how many of the N blocks are rolled (effectively rolling). ing.
As shown in FIG. 3 (b), the motor friction characteristic function _{G M} is displaced _{0~θ 1, θ 1 ~θ 2,} θ 2 ~θ 3, the slope of the straight line to be employed in each of the theta ₃ through? ₄ K ₁ , K ₂ , K ₃ , K ₄ , and offset (intercept at displacement 0) τ _fM1 (= _{−τ fMc} ), τ _fM2 , τ _fM3 , τ _fM4 , τ _fM5 (= τ _fMc ) .
The displacement θ on the horizontal axis is given as an integral value from the time of reversing the motor angular speed ω _M (ω _M = 0).
Although not shown, the load friction characteristic function G _L is also defined similarly to the motor friction characteristic function G _M shown in FIG. 3 (b). In other words, the load friction characteristic function _GL includes the slope of a straight line corresponding to the displacement θ (the integrated value from the speed reversal of the load angular velocity ω _L (ω _L = 0)) and the offset ( _{−τ fLc} to + τ _fLc ). Range).

(Characteristics of dead zone)
FIG. 4 is a diagram for explaining the characteristics of the dead zone in the connection member of the mechanical system according to the first embodiment.

A dead band characteristic function F shown in FIG. 4A represents a transfer characteristic based on a dead band width BL indicating a width of a backlash (dead band) of the connecting member 22 provided between the motor 20 and the load 21.
The dead band characteristic function F is a nonlinear characteristic defined by the dead band width BL, and the torsion angle (θ _M −θ _L ) of the connecting member 22 is an input variable. As shown in FIG. 4A, when the absolute value of the twist angle (θ _M −θ _L ) of the connecting member 22 is equal to or less than the dead band width BL (−BL ≦ θ _M −θ _L ≦ + BL), the backlash output is zero. It becomes. That is, in this case, since the twist angle (θ _M −θ _L ) is within the width of the backlash, torque is not transmitted from the motor 20 to the load 21. On the other hand, when the absolute value of the twist angle (θ _M −θ _L ) of the connecting member 22 is larger than the dead band width BL (θ _M −θ _L <−BL, θ _M −θ _L > + BL), the backlash output is a twist. A value (θ _M −θ _L + BL (θ _M −θ _L <−BL) or θ _M −θ _L −BL (θ _M −θ _L > + BL) smaller than the angle (θ _M −θ _L ) by the dead band width BL. )). That is, the component that contributes to the torque transmission through the characteristics (torsional stiffness coefficient K _R , torsional viscosity coefficient D _R ) of the connecting member 22 is only the portion where the dead band width BL is removed from the entire torsion angle (θ _M −θ _L ). It is expressed that.

Also, the backlash displacement function F ′ shown in FIG. 4B has a relationship between the backlash displacement BKLS [θ _M −θ _L ] and the twist angle (θ _M −θ _L ).
The backlash displacement BKLS [θ _M −θ _L ] is a parameter indicating the amount of displacement in the dead zone having a width of the dead zone width BL (−BL to + BL). FIG. 4 shows the twist angle (θ _M −θ _L ). It has the characteristics shown in (b). That is, the backlash output shown in FIG. 4A can be expressed as “θ _M −θ _L −BKLS” using the backlash displacement BKLS [θ _M −θ _L ].
As shown in FIG. 4B, the backlash displacement BKLS [θ _M −θ _L ] is a backlash displacement BKLS [θ in a range where the twist angle (θ _M −θ _L ) is smaller than the minimum value of the dead zone (−BL). θ _M −θ _L ] takes a minimum value (−BL), and in a range where the twist angle (θ _M −θ _L ) is larger than the maximum value (+ BL) of the dead zone, the backlash displacement BKLS [θ _M −θ _L ] is Take the maximum value (+ BL). In the range where the twist angle (θ _M −θ _L ) is not less than the minimum value (−BL) and not more than the maximum value (+ BL) of the dead zone, the backlash displacement BKLS [θ _M −θ _L ] is equal to the twist angle (θ _M The characteristic has the same value as −θ _L ).

(Functional configuration of drive system)
FIG. 5 is a diagram illustrating a functional configuration of the drive system according to the first embodiment.
As shown in FIG. 5, the drive system 9 according to the first embodiment includes a control device 1, a mechanical system 2, and a parameter identification device 3.

  The control device 1 includes a feedback control unit 10 and a feedforward control unit 11.

Feedback control unit 10, a motor angular velocity omega _M observed through the rotation detector, the load angle command theta target angle of the load 21 which is designated by the _REF theta _REF and the current load angle theta _L deviations observed (theta _REF −θ _L ) and the motor 20 is controlled.
Specifically, the feedback control unit 10 calculates a torque for setting the deviation (θ _REF −θ _L ) between the target angle θ _REF and the load angle θ _L to zero, and outputs a torque command indicating the calculation result. To do. At this time, the feedback control unit 10 refers to the motor angular velocity ω _M observed through the rotation detector and calculates a torque that allows appropriate and quick feedback control.

The feedforward control unit 11 has a two-inertia model MOD imitating the mechanical system 2 inside. This two-inertia system model MOD is composed of a plurality of model parameter groups, and is configured to be an inverse model of the mechanical system 2 shown in FIG.
Specifically, the plurality of model parameter groups are: model motor inertia moment J _M0 , model motor viscosity coefficient D _M0 , model motor friction torque τ _fM0 , model load inertia moment J _L0 , model load viscosity coefficient D _L0 , model load The friction torque τ _fL0 , the model dead zone width BL ₀ , and the model torsional stiffness coefficient K _R0 .
In the present embodiment, the connecting the twist viscosity coefficient D _R of the member 22, it is assumed motion effects is sufficiently small, and excluded from the modeling.

The two-inertia system model MOD is an inverse model of the model constructed as shown in FIG. 2 simulating the mechanical system 2, and when the target angle θ _REF of the load 21 is input, the load 21 is transmitted through the mechanical system 2. The torque τ to be applied to be positioned at the target angle θ _REF is calculated and output. The above-described model parameter group (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ) forming the two-inertia system model MOD is a parameter group that represents the actual characteristics of the mechanical system 2 ( J _M , D _M , τ _fM , J _L , D _L , τ _fL , BL, K _R ).

Here, the model parameter group (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ) and the parameter group indicating the actual characteristics of the mechanical system 2 (J _M , D _M , Τ _fM , J _L , D _L , τ _fL , BL, K _R ), the two-inertia system model MOD completely matches the inverse model of the actual mechanical system 2. It will be a thing. Then, as a result of driving the feedforward controller 11 by outputting the torque calculated based on the target angle θ _REF and the two-inertia system model MOD to the mechanical system 2, the load angle θ _L actually obtained through the mechanical system 2 is It should completely match the target angle θ _REF (θ _L = θ _REF ). In this way, the feedforward control unit 11 calculates torque while incorporating characteristics unique to the mechanical system 2 on the basis of the two-inertia model MOD that is preliminarily defined by imitating the mechanical system 2. In feed-forward control with high responsiveness, more accurate control can be realized.

  As shown in FIG. 5, the control device 1 is configured such that the torque τ (torque command τ) obtained by adding the torque calculated by the feedback control unit 10 and the torque calculated by the feedforward control unit 11. Is output to the mechanical system 2 (motor 20 (FIGS. 1 and 2)).

As described above, the control device 1 according to the present embodiment includes a plurality of model parameter groups (J _M0 , D) that form a two-inertia model MOD that imitates the mechanical system 2 including the motor 20, the load 21, and the connecting member 22. _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ), a torque τ to be generated by the motor 20 is calculated, and a torque command τ is output.

In the feedforward control unit 11, model parameter groups (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ) are assumed in advance as parameters representing the characteristics of the mechanical system 2. It is specified by the value. However, actually, the model parameter group (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ) and the parameter group specific to the mechanical system 2 (J _M , D _M , If there is an error between τ _fM , J _L , D _L , τ _fL , BL, K _R ), highly accurate feedforward control cannot be realized.
Therefore, the drive system 9 according to the present embodiment includes the parameter identification device 3 that can accurately identify the plurality of model parameter groups described above.

As shown in FIG. 5, the parameter identification device 3 inputs measured values of the motor angle θ _M , the motor angular velocity ω _M , the load angle θ _L and the load angular velocity ω _{L in} the mechanical system 2. The parameter identification device 3 also receives a torque command τ output from the control device 1 toward the mechanical system 2. The parameter identification device 3 uses the above-described model parameter group (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K _R0 ) based on the four actually measured values and the torque command τ. Is identified.
Hereinafter, the function of the parameter identification device 3 will be described in detail.

(Functional configuration of parameter identification device)
FIG. 6 is a diagram illustrating a functional configuration of the parameter identification device according to the first embodiment.
As shown in FIG. 6, the parameter identification device 3 includes a data acquisition unit 30, a phase plane generation unit 31, a parameter identification unit 32, and an angle command output unit 33.

The data acquisition unit 30 measures the torque command τ for the motor 20, the measured values of the motor 20 angle (motor angle θ _M ) and angular velocity (motor angular velocity ω _M ), the load 21 angle (load angle θ _L ) and angular velocity ( And an actual measurement value of the load angular velocity ω _L ).

The phase plane diagram generation unit 31 includes a plurality of phase planes based on the torque command τ acquired by the data acquisition unit 30, the measured values of the angle and angular velocity of the motor 20, and the measured values of the angle and angular velocity of the load 21. Generate a diagram.
Here, as shown in FIG. 6, the phase plane diagram generator 31 is a “torsion angle-torque command” phase plane diagram showing the relationship between the twist angle (θ _M −θ _L ) of the connecting member 22 and the torque command τ. P0a and P0b are generated.
Further, the phase plane diagram generation unit 31 generates “angle-torque command” phase plane diagrams P1a and P1b indicating the relationship between the angle θ of the motor 20 and the load 21 and the torque command τ.
Further, the phase plane diagram generation unit 31 generates “angular velocity-torque command” phase plane diagrams P2a and P2b indicating the relationship between the angular velocity ω of the motor 20 and the load 21 and the torque command τ.
In addition, the phase plane diagram generation unit 31 displays “angle-twist angle” phase plane diagrams P3a and P3b indicating the relationship between the angle θ of the motor 20 and the load 21 and the twist angle (θ _M −θ _L ) of the connecting member 22. Generate.
In addition, the phase plane diagram generation unit 31 displays “angular velocity-twist angle” phase plane diagrams P4a and P4b indicating the relationship between the angular velocity ω of the motor 20 and the load 21 and the twist angle (θ _M −θ _L ) of the connecting member 22. Generate.

The parameter identification unit 32 is configured based on a plurality of phase plane diagrams generated by the phase plane generation unit 31 based on the model parameter group (J _M0 , D _M0 , τ _fM0 , J _L0 , D of the two-inertia system model MOD). _{L0, τ fL0, BL 0,} K R0) to identify.

The angle command output unit 33 outputs a load angle command θ _REF for parameter identification to the control device 1. Specifically, the angle command output unit 33 outputs a load angle command θ _REF1 for operating the load 21 with a small amplitude and a load angle command θ _REF2 for operating the load 21 with a large amplitude. Here, the load angle command θ _REF1 is a command for causing the load 21 to perform a sinusoidal motion with an amplitude small enough to ignore the effects of inertia and viscosity of the motor 20 and the load 21 in the motion of the mechanical system 2. On the other hand, the load angle command θ _REF2 is a command for causing the load 21 to perform a sinusoidal motion with a large amplitude such that the effects of the inertia and viscosity of the motor 20 and the load 21 are sufficiently expressed in the movement of the mechanical system 2.

(Processing flow of parameter identification device)
FIG. 7 is a diagram illustrating a processing flow of the parameter identification device according to the first embodiment.
FIGS. 8 to 17 are FIGS. 1 to 10 for explaining the processing of the parameter identification device according to the first embodiment, respectively.
The processing flow shown in FIG. 7 is, for example, when the parameter identification device 3 (FIGS. 5 and 6) performs parameter identification for the two-inertia system model MOD (FIG. 5) before the actual operation of the mechanical system 2 is started. To be executed.

First, the angle command output unit 33 of the parameter identification device 3 outputs the load angle command θ _REF1 for small amplitude operation, so that the angle of the load 21 (load angle θ _L ) is sinusoidal motion with small amplitude and low frequency. (Step S01).

Next, the data acquisition unit 30 of the parameter identification device 3 receives the torque command τ, the motor angular velocity ω _M , the motor angle θ _M , the load angular velocity ω _L, and the load from the mechanical system 2 that is sine wave motion with a small amplitude and low frequency. acquiring the measured value of the angle theta _L (step S02).

  Next, the phase diagram generator 31 of the parameter identification device 3 generates a plurality of phase diagrams from the actual measurement values acquired in step S02 (step S03).

Next, the parameter identification unit 32 of the parameter identification device 3 uses the model motor friction torque τ _fM0 , the model load among the various model parameters forming the two-inertia model MOD based on the various phase planes generated in step S03. The friction torque τ _fL0 , the model dead zone width BL ₀ and the model torsional stiffness coefficient K _R0 are identified (step S04).

Next, the angle command output unit 33 of the parameter identification device 3 outputs the load angle command θ _REF2 for large amplitude operation, thereby causing the angle of the load 21 (load angle θ _L ) to sine-wave with a large amplitude ( Step S05).

Next, the data acquisition unit 30 of the parameter identification device 3 receives the torque command τ, the motor angular velocity ω _M , the motor angle θ _M , the load angular velocity ω _L, and the load angle θ _L from the mechanical system 2 that is in a sinusoidal motion with a large amplitude. Is obtained (step S06).

  Next, the phase plane diagram generation unit 31 of the parameter identification device 3 generates a plurality of phase plane diagrams from the actual measurement values acquired in step S06 (step S07).

Next, the parameter identification unit 32 of the parameter identification device 3 includes the model motor moment of inertia J _M0 , the model motor among the various model parameters forming the two-inertia system model MOD, based on the various phase planes generated in step S07. The viscosity coefficient D _M0 , the model load inertia moment J _L0 , and the model load viscosity coefficient D _L0 are identified (step S08).

  Here, based on the block diagram shown in FIG. 2, Equation (1) that is the equation of motion for the motor 20 and Equation (2) that is the equation of motion for the load 21 can be obtained.

From Expressions (1) and (2), Expression (3) regarding the torque command τ and Expression (4) regarding the twist angle (θ _M −θ _L ) of the connecting member 22 can be obtained.

In the formula (3), the overall moment of inertia J is the sum of the motor inertia _{J M} and the load inertia _{J L (J = J M +} J L). The overall viscosity coefficient D is the sum of the motor viscosity coefficient _DM and the load viscosity coefficient D _L (D = D _M + D _L ). In the equations (3) and (4), the angle θ = θ _M ≈θ _L.
Since the load angle command θ _REF1 (and θ _REF2 ) is a sine wave, when the angle θ is “θ = θ ₀ · sin (2πf · t)”, the angular velocity sθ (= ω) is “sθ = _{(2πf) θ 0 · cos (} 2πf · t) ", is the angular acceleration ^{s 2 θ," s 2 θ} = - is _{^{(2πf) 2 θ 0 · sin}} (2πf · t) ". Therefore, the angular velocity sθ (= ω) is expressed as the equation (5) as a function of the angle θ.

Further, the angular acceleration s ² θ is expressed as a function of the angle θ as shown in Expression (6), and is expressed as a function of the angular velocity ω as shown in Expression (7).

  Using the equations (5) to (7) and the above equation (3), the equation (8) indicating the relationship between the torque command τ and the angle θ and the equation indicating the relationship between the torque command τ and the angular velocity ω. (9) is obtained.

In Equation (9), “ω ₀ = θ ₀ · 2πf” is set.

Further, using the formulas (5) to (7) and the above-described formula (4), the formula (10) showing the relationship between the twist angle (θ _M −θ _L ) and the angle θ and the twist angle (θ Equation (11) indicating the relationship between _M −θ _L ) and angular velocity ω is obtained.

  Next, referring to FIGS. 8 to 12 and the above formulas (8) to (11), the phase plane diagram generated by the phase plane diagram generation unit 31 during the small amplitude operation and the parameter identification unit The parameter identification process (step S03, step S04) performed by the process 32 will be described in detail.

In step S03, the phase diagram generator 31 associates the measured value of the torsion angle (θ _M −θ _L ) acquired in step S02 with the torque command τ to associate the torsion angle (θ _M −θ _L ). Is plotted on a phase plane with the horizontal axis representing the torque command τ and the vertical axis representing the torque command τ. Here, the “actual value of the twist angle (θ _M −θ _L )” is obtained by subtracting the actual value of the load angle θ _L from the actual value of the motor angle θ _M. As a result, the phase plane diagram generation unit 31 has a twist angle-torque command phase that is a phase plane diagram with the torsion angle (θ _M −θ _L ) on the horizontal axis and the torque τ on the vertical axis as shown in FIG. A plane view P0a is generated. The torsion angle-torque command phase diagram P0a is a phase diagram generated based on the actual measurement value acquired during the small amplitude operation.
The phase view generator 31, a measured value obtained from the positive direction of the load angle theta _L when moving in the negative direction, the measured value acquired when moving from a negative direction to the positive direction, For each of The actual measurement values are linked. In the thus generated torsion angle-torque command phase diagram P0a, a hysteresis curve caused mainly by the nonlinear friction characteristic (see FIG. 3A) of the mechanical system 2 and the dead band characteristic function F appears.

In addition, in each operating point and operating region (T1 to T5) in the curve shown in FIG. 8 (twist angle-torque command phase diagram P0a), the following states are obtained.
Operating point T1 ... A state where the gear (dead zone) is in contact and being driven. Friction is also maximum.
Operating region T2: A state in which the driving force (torque τ) decreases and the torsion of the shaft (connecting member 22) decreases.
Operating region T3... Only the twist of the shaft is changed while the driving force remains unchanged. That is, the dead zone (backlash) of the connecting member is moving.
Operating region T4: A state in which the driving force in the opposite direction increases and the twist of the shaft increases. Based on the load friction characteristic function _GL , the friction changes (increases) nonlinearly.
Operating point T5: A state where the gear is in contact and being driven. Friction is also maximum.

In the case of a sine wave motion with a small amplitude and a low frequency, the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) can be ignored. Therefore, from the equation (8), the torque command τ at the operating points T1 and T5 is the total friction component (τ _fMc + τ _fLc ).
In addition, during play movement (operation region T3), since it is not affected by the load 21 (only the characteristics of the motor 20 can be seen), the torque command τ in the region is a motor coulomb that is a saturation value of the motor friction torque _τfM. The friction torque τ _fMc .
The relationship between the torque command τ and twist angle (θ _M -θ _L), since as indicated by the torsional stiffness coefficient K _R, torsional rigidity coefficient K _R is specified from the slope in the operating region T2.
Further, since the amount of change in the twist angle (θ _M −θ _L ) while moving through the dead zone corresponds to the total dead zone width (length from end to end of backlash = 2BL), the twist angle (θ in the operation region T3) _The dead zone width BL is specified from the magnitude of _M −θ _L ).
Thus, parameter identification unit 32, twist angle - based on the torque command phase view P0a, identifiable motor Coulomb friction torque tau _FMC, load Coulomb friction torque tau _FLC, the dead zone width BL, the torsional rigidity coefficient _{K R.}

Further, the phase plane diagram generation unit 31 generates an angle-torque command phase plane diagram P1a (see FIG. 9) indicating the relationship between the angle θ and the torque command τ based on various actual measurement values acquired during the small amplitude operation. To do.
In the equation (8), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are ignored, the torque command τ is the total friction torque (motor friction torque τ _fM [θ] and The sum of the load friction torque τ _{fL and} [θ]. Here, the motor friction torque τ _fM [θ] and the load friction torque τ _fL [θ] are characteristics defined by the motor friction characteristic function G _M and the load friction characteristic function _GL , respectively (see FIG. 3). ).
As shown in FIG. 9, the parameter identification unit 32 identifies “τ _fMc + τ _fL [θ]” based on the curve of the angle-torque command phase diagram P1a. Here, the parameter identification unit 32 assumes that the degree of change in the motor friction torque τ _fM [θ] with respect to the angle (displacement) θ is sufficiently smaller than the load friction torque τ _fL [θ].
In addition, the parameter identification unit 32 identifies “τ _fMc + τ _fLc ” at the saturation point of τ _fL [θ].

Further, the phase plane diagram generation unit 31 generates an angular velocity-torque command phase plane diagram P2a (see FIG. 10) indicating the relationship between the angular velocity ω and the torque command τ based on various actually measured values acquired during the small amplitude operation. To do.
In the equation (9), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are ignored, the torque command τ is the total friction torque (motor friction torque τ _fM [θ] and The sum of the load friction torque τ _{fL and} [θ]. Here, the characteristics of the motor friction torque τ _fM [θ] and the load friction torque τ _fL [θ] with respect to the angular velocity ω are respectively the motor friction characteristic function G _M and the load friction characteristic function G _L (function of the angle θ). ) Differential characteristics.
As shown in FIG. 10, the parameter identification unit 32 identifies “τ _fMc ” and “τ _fMc + τ _fLc ” based on the curve of the angular velocity-torque command phase diagram P2a.

Furthermore, the phase plane diagram generation unit 31 is an angle-twist angle phase plane diagram P3a (representing the relationship between the angle θ and the twist angle (θ _M −θ _L ) based on various actual measurement values acquired during the small amplitude operation. 11) is generated.
In the equation (10), when the effects of the moment of inertia (J _M , J _L ) and the viscosity coefficients (D _M , D _L ) are ignored, the torsion angle (θ _M −θ _L ) is determined by the backlash displacement component (BKLS [θ _M −θ _L ]) and the load friction torque component (τ _fL [θ] / K _R ).
Since the influence of inertia and viscosity is small during the small amplitude operation, the timing at which the torsion angle (θ _M −θ _L ) moves the backlash BKLS is the time when the angular velocity ω is reversed (that is, the maximum value of the angle θ and (Minimum value).
As shown in FIG. 11, the parameter identification unit 32 determines “(τ _fL [θ] / K _R ) + BL”, “(τ _fLc / K _R ) + BL based on the curve of the angle-torsion angle phase diagram P3a. Is specified.

Furthermore, the phase plane diagram generation unit 31 is based on various measured values acquired during the small amplitude operation, and the angular velocity-torsion angle phase diagram P4a (representing the relationship between the angular velocity ω and the twist angle (θ _M −θ _L )). (See FIG. 12).
In the equation (11), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are ignored, the torsion angle (θ _M −θ _L ) is determined by the backlash displacement component (BKLS [θ _M −θ _L ]) and the load friction torque component (τ _fL [θ] / K _R ).
Since the influence of inertia and viscosity is small during the small amplitude operation, the timing at which the twist angle (θ _M −θ _L ) moves the backlash BKLS coincides with the reversal of the angular velocity ω (ω = 0).
As shown in FIG. 12, the parameter identification unit 32 identifies “(τ _fLc / K _R ) + BL” based on the curve of the angular velocity-torsion angle phase diagram _P4a .
Further, the angular velocity omega angular as the motor angular speed omega _M - when generating the twist angle phase side view P4a, during rattling displacement BKLS movement, only the motor angular velocity omega _M operation region appears to be increased while the load angular velocity omega _L unchanged (Arrow Q shown in FIG. 12). The width of the torsion angle (θ _M −θ _L ) of the operation region indicated by the arrow Q corresponds to the total dead zone width (2BL). Therefore, the parameter identification unit 32 further identifies “2BL” based on the angular velocity-torsion angle phase diagram P4a.

As described above, the parameter identification unit 32 generates “τ _fMc + τ _fL [θ]” and “τ _fMc + τ _fLc ” based on the four phase plane diagrams P1a to P4a generated by various actual measurement values during the small amplitude operation. , “Τ _fMc ”, “(τ _fL [θ] / K _R ) + BL”, “(τ _fLc / K _R ) + BL”, “2BL”. Parameter identification unit 32, by solving simultaneous equations of these numbers groups identified, the motor friction torque tau _fM, the load friction torque tau _fL, dead zone width BL, identifies the torsional rigidity coefficient K _R.

  Next, referring to FIGS. 13 to 17 and the above formulas (8) to (11), the phase plane diagram generated by the phase plane diagram generation unit 31 during the large amplitude operation and the parameter identification unit The parameter identification process (steps S07 and S08) performed by the process 32 will be described in detail.

In step S07, the phase diagram generator 31 plots the measured value of the torsion angle (θ _M −θ _L ) acquired in step S06 and the torque command τ in association with each other, thereby producing a torsion angle-torque command. A phase diagram P0b is generated. The torsion angle-torque command phase diagram P0b is a phase diagram generated based on the actual measurement value acquired during the large amplitude operation.

Even during large-amplitude sinusoidal motion, the amount of change in the twist angle (θ _M −θ _L ) while moving through the dead band corresponds to the total dead band width (length from end to end of backlash = 2BL). . Therefore, the parameter identification unit 32 is based on the twist angle-torque command phase diagram P0b, and the operation region (the operation region in FIG. 8) in which the twist angle (θ _M −θ _L ) is changed without changing the torque command τ. The dead zone width BL can be specified from the magnitude of the twist angle (θ _M −θ _L ) (corresponding to T3).

Further, the phase plane diagram generation unit 31 generates an angle-torque command phase plane diagram P1b (see FIG. 14) indicating the relationship between the angle θ and the torque command τ based on various actual measurement values acquired during the large amplitude operation. To do.
In the equation (8), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are taken into consideration, the torque command τ is the total friction torque component (motor friction torque τ _fM [θ] And the load friction torque τ _fL [θ]), the total inertia component (−J (2πf) ² θ), and the total viscosity component (± D (2πf) (θ ₀ ² −θ ² ) ^1/2 ) It is expressed as the sum of
Here, during a large amplitude operation, the motor friction torque τ _fM [θ] and the load friction torque τ _fL [θ] are predominantly based on Coulomb friction. Therefore, the total friction torque component on the angle-torque command phase diagram P1b is the total Coulomb friction torque component (τ _fMc + τ _fLc ) at the timing when the angular velocity ω is reversed (that is, the maximum value and the minimum value of the angle θ). The behavior is reversed.
On the angle-torque command phase diagram P1b, the overall inertia component (-J (2πf) ² θ) has a linear characteristic with a slope of −J (2πf) ² .
On the angle-torque command phase diagram P1b, the overall viscosity component (± D (2πf) (θ ₀ ² −θ ² ) ^1/2 ) has an elliptic characteristic with a degree of swelling of D (2πf). .
As shown in FIG. 14, the parameter identification unit 32 calculates “2 (τ _fMc + τ _fLc )”, “J (2πf) ² ”, “D (2πf)” based on the angle-torque command phase diagram P1b. Identify. In particular, in the example shown in FIG. 14, since the elliptic component is hardly seen in the angle-torque command phase diagram P1b, it can be read that the overall viscosity coefficient D is almost zero.

Further, the phase plane diagram generation unit 31 generates an angular velocity-torque command phase plane diagram P2b (see FIG. 15) indicating the relationship between the angular velocity ω and the torque command τ based on various actually measured values acquired during the large amplitude operation. To do.
In the equation (9), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are taken into consideration, the torque command τ is the total friction torque component (motor friction torque τ _fM [θ] And the load friction torque τ _fL [θ]), the total inertia component (± J (2πf) (ω ₀ ² −ω ² ) ^1/2 ), and the total viscosity component (Dω). The
Here, during a large amplitude operation, the motor friction torque τ _fM [θ] and the load friction torque τ _fL [θ] are predominantly based on Coulomb friction. Therefore, the total friction torque component on the angle-torque command phase diagram P1b has a behavior in which the sign of the total Coulomb friction torque component (τ _fMc + τ _fLc ) is reversed at the timing when the angular velocity ω is reversed (ω = 0). Become.
In addition, on the angular velocity-torque command phase diagram P2b, the overall inertia component (± J (2πf) (ω ₀ ² −ω ² ) ^1/2 ) has an elliptic characteristic in which the degree of swelling is J (2πf). .
Further, on the angular velocity-torque command phase diagram P2b, the overall viscosity component (Dω) has a linear characteristic with a slope of D.
As shown in FIG. 15, the parameter identification unit 32 identifies “2 (τ _fMc + τ _fLc )”, “J (2πf)”, “D” based on the angular velocity-torque command phase diagram P2b. In particular, in the example shown in FIG. 15, since almost no linear component is seen in the angular velocity-torque command phase diagram P2b, it can be read that the overall viscosity coefficient D is almost zero.

Furthermore, the phase plane diagram generation unit 31 is an angle-twist angle phase plane diagram P3b (P3b) showing the relationship between the angle θ and the twist angle (θ _M −θ _L ) based on various actual measurement values acquired during the large amplitude operation. 16).
In the equation (10), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are taken into consideration, the torsion angle (θ _M −θ _L ) is determined by the backlash displacement component (BKLS [θ _M− θ _L ]), load friction torque component (τ _fL [θ] / K _R ), load inertia component (− (J _L / K _R ) (2πf) ² θ), and load viscosity component (± ( D _L / K _R ) (2πf) (θ ₀ ² −θ ² ) ^1/2 ).
Here, during large-amplitude operation, the load friction torque τ _fL [θ] has a dominant behavior based on Coulomb friction. Therefore, the load friction torque component on the angle-torsion angle phase diagram P3b is the sign of the load Coulomb friction torque component (τ _fLc ) at the timing when the angular velocity ω is reversed (that is, the maximum value and the minimum value of the angle θ). Will be reversed.
In addition, since the influence of inertia and viscosity is large during large amplitude operation, the timing at which the torsional angle (θ _M −θ _L ) moves the backlash BKLS is adjusted while the angle θ is moving from the minimum value to the maximum value. This coincides with the timing when the command τ becomes zero (see the angle-torque command phase diagram P1b in FIG. 14).
Further, on the angle-torsion angle phase diagram P3b, the load inertia component (-(J _L / K _R ) (2πf) ² θ) has a linear characteristic with an inclination of-(J _L / K _R ) (2πf) ² . It becomes.
On the angle-twist angle phase diagram P3b, the load viscosity component (± (D _L / K _R ) (2πf) (θ ₀ ² −θ ² ) ^1/2 ) has a degree of swelling of (D _L / K _R ) (2πf).
As illustrated in FIG. 16, the parameter identification unit 32 performs “2τ _fLc / K _R ”, “2BL”, “− (J _L / K _R ) (2πf) ² based on the angle-twist angle phase diagram P3b. ], “(D _L / K _R ) (2πf)”. In particular, in the example shown in FIG. 16, since the elliptic component is hardly seen in the angle-torsion angle phase diagram P3b, it can be read that the load viscosity coefficient _DL is almost zero.

Further, the phase plane diagram generation unit 31 is based on various measured values acquired during the large amplitude operation, and the angular velocity-torsion angle phase diagram P4b (representing the relationship between the angular velocity ω and the torsion angle (θ _M −θ _L )). 17).
In the equation (11), when the influence of the moment of inertia (J _M , J _L ) and the viscosity coefficient (D _M , D _L ) are taken into consideration, the torsion angle (θ _M −θ _L ) is determined by the backlash displacement component (BKLS [θ _M− θ _L ]), load friction torque component (τ _fL [θ] / K _R ), and load inertia component (± (J _L / K _R ) (2πf) (ω ₀ ² −ω ² ) ^1/2 ) And the load viscosity component ((D _L / K _R ) ω).
Here, during large-amplitude operation, the load friction torque τ _fL [θ] has a dominant behavior based on Coulomb friction. Accordingly, the load friction torque component on the angular velocity-torsion angle phase plane P4b has a behavior in which the sign of the load Coulomb friction torque component (τ _fLc ) is reversed at the timing when the angular velocity ω is reversed (ω = 0).
In addition, since the influence of inertia and viscosity is large during large amplitude operation, the timing at which the torsional angle (θ _M −θ _L ) moves the backlash BKLS is adjusted while the angle θ is moving from the minimum value to the maximum value. This coincides with the timing when the command τ becomes zero (see the angular velocity-torque command phase diagram P2b in FIG. 15).
In addition, on the angular velocity-torsion angle phase plane P4b, the load inertia component (± (J _L / K _R ) (2πf) (ω ₀ ² −ω ² ) ^1/2 ) has a degree of bulge ((J _L / K _R ) (2πf)).
In addition, on the angular velocity-twist angle phase plane diagram P4b, the load viscosity component ((D _L / K _R ) ω) has a linear characteristic with an inclination of (D _L / K _R ).
As illustrated in FIG. 17, the parameter identification unit 32 performs “2τ _fLc / K _R ”, “2BL”, “(J _L / K _R ) (2πf)” based on the angular velocity-torsion angle phase diagram P4b. Specify “D _L / K _R ”. In particular, in the example shown in FIG. 17, the angular velocity - for linear component is hardly seen in the twist angle phase side view P4b, it is read load viscosity coefficient D _L is approximately zero.

As described above, the parameter identification unit 32 performs “J (2πf) ² ”, “D (2πf)”, “D (2πf)”, “D” based on the four phase plane diagrams P1b to P4b generated by various actual measurement values during large amplitude operation. _{- (J L / K R)} (2πf) 2 "," _{(D L / K R) (} 2πf) "," _{(J L / K R) (} 2πf) "," identify the _D L / _{K R} ", etc. To do. The parameter identification unit 32 solves these numerical groups simultaneously (including the parameters specified in step S04), so that the motor inertia moment J _M , the load inertia moment J _L , the motor viscosity coefficient _DM, and the load viscosity coefficient Specify _DL .

The parameter identification unit 32 _converts the various parameters identified as described above into various model parameter groups (J _M0 , D _M0 , τ _fM0 , J _L0 , D _L0 , τ _fL0 , BL ₀ , K) of the two-inertia system model MOD. _R0 ).

(Function, effect)
As described above, the parameter identification device 3 according to the first embodiment is configured to include the above-described data acquisition unit 30, the phase plane generation unit 31, and the parameter identification unit 32.
In this way, the various parameters of the mechanical system 2 can be accurately and excessively insufficient based on both the phase diagram acquired during the small amplitude operation and the phase diagram acquired during the large amplitude operation. Can be identified.
That is, in the state of small amplitude and low frequency, the influence of the moment of inertia and the viscosity coefficient is small, and the characteristics of the motor 20, the friction of the load 21, the rigidity of the connecting member 22, and the dead band width can be easily captured. On the other hand, in the case of a large amplitude state, the effects of the friction of the motor 20 and the load 21, the rigidity of the connecting member 22 and the dead band width are relatively reduced, and the characteristics of the moment of inertia and the viscosity coefficient are easily captured. Therefore, by adopting a mode in which the above-described two-stage steps are performed, all parameters constituting the two-inertia system model MOD can be identified with high accuracy.
Therefore, it is possible to obtain a two-inertia model MOD that accurately simulates the mechanical system 2 having the motor 20 and the load 21.

  As described above, the drive system 9 and the parameter identification device 3 according to the first embodiment have been described in detail. However, specific modes of the drive system 9 and the parameter identification device 3 are not limited to those described above. Various design changes and the like can be added without departing from the scope of the invention.

For example, the parameter identification unit 32 according to another embodiment includes a plurality of model parameters that can be identified from a plurality of phase plane diagrams (P0a, P0b, P1a, P1b,..., P4a, P4b). When street values are identified, the average value may be used as a model parameter. For example, the parameter identification unit 32 includes the dead zone width BL identified from the torsion angle-torque command phase diagram P0a acquired at the time of the small amplitude and the dead zone width BL identified from the angular velocity-torsion angle phase diagram P4a. may be employed average value to model dead zone width BL ₀ of 2-inertia system model MOD.

Further, the phase plane diagram generation unit 31 according to another embodiment may generate only one phase plane diagram (twist angle-torque command phase plane diagram P0a) at the time of small amplitude. In this case, the parameter identification unit 32 determines the model motor coulomb friction torque τ _fMc0 , the model load coulomb friction torque τ _fLc0 , the model torsional stiffness coefficient K _R0, and the model dead zone based on the generated torsion angle-torque command phase diagram P0a. The width BL ₀ is identified.
By doing in this way, the identification process of each model parameter can be simplified.

In addition, the phase plane diagram generation unit 31 according to another embodiment repeats the acquisition of actual measurement data for each repetitive operation of the mechanical system 2 and generates each phase plane diagram based on the averaged result. Also good.
By doing so, the variation error of the measured data is reduced, so that the identification accuracy of the model parameters can be further increased.

Further, the parameter identification device 3 according to the first embodiment can be used not only for improving the feedforward performance in the control device 1 but also for initial shipment inspection, abnormality diagnosis, machine adjustment and the like of the drive system 9. it can.
Further, the parameter of the inertia moment J identified as described above can be used not only for improving the feedforward performance but also for adjusting (changing) the gain of feedback control.

Further, in the first embodiment, the parameter identification device 3 has been described as performing the above-described various processes with the torque command input from the host device as “torque τ”. It is not limited to.
For example, the parameter identification device 3 according to another embodiment performs the above-described various processes by using not “torque command” but “measured torque” obtained from the detected current value flowing in the motor 20 as “torque τ”. It may be.

In the first embodiment, the control device 1 has been described as detecting the motor angular velocity ω _M by differentiating the motor angle θ _M detected from the rotation detector with respect to time. It is not limited to an aspect. For example, the control device 1 may be a mode for detecting a motor angle theta _M by integrating motor angular velocity omega _M detected directly from the rotation detector.
The same applies to the load angle θ _L and the load angular velocity ω _L.

Moreover, in 1st Embodiment, although demonstrated as what is a full closed system which makes load angle (theta) _L the control object of the position loop in the feedback control part 10, in other embodiment, it is not limited to this aspect. In other embodiments, the control target position loop in the feedback control unit 10 may be a semi-closed system with motor angle theta _M.

Further, when there is a steady disturbance such as gravity, a constant offset is applied to the torque τ and the twist angle (θ _M −θ _L ). Therefore, the parameter identification device 3 according to another embodiment further has a function of obtaining an average of the torque τ and the torsion angle (θ _M −θ _L ) to remove the offset, and a function of taking the average of the break points positively and negatively. You may have.

In each of the above-described embodiments, a program for realizing the function of the parameter identification device 3 in the drive system 9 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is stored in the computer system. Each step is performed by reading and executing. Here, each process of the parameter identification device 3 described above is stored in a computer-readable recording medium in the form of a program, and the above-described various processes are performed by the computer reading and executing the program. Here, the computer-readable recording medium refers to a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, and the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.
Moreover, the aspect with which the function of the parameter identification apparatus 3 is comprised over the some apparatus connected with a network may be sufficient.

  As mentioned above, although some embodiment of this invention was described, these embodiment is shown as an example and is not intending limiting the range of invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the invention described in the claims and equivalents thereof, as long as they are included in the scope and gist of the invention.

DESCRIPTION OF SYMBOLS 1 Control apparatus 10 Feedback control part 11 Feedforward control part 2 Mechanical system 20 Motor 21 Load 22 Connecting member 3 Parameter identification apparatus 30 Data acquisition part 31 Phase plane generation part 32 Parameter identification part 33 Angle command output part 9 Drive system (theta) _REF Target angle τ Torque command (torque)
θ _M motor angle (motor angle)
ω _M motor angular velocity (motor angular velocity)
J _M motor moment of inertia D _M motor viscosity coefficient τ _fM motor friction torque τ _fMc motor coulomb friction torque θ _L load angle (load angle)
ω _L Load angular velocity (Load angular velocity)
J _L Load moment of inertia D _L Load viscosity coefficient τ _fL Load friction torque τ _fLc Load Coulomb friction torque K _R Torsional stiffness coefficient D _R Torsional viscosity coefficient BL _Dead band MOD Two inertia system model J _M0 model motor inertia moment D _M0 model motor viscosity Coefficient τ _fM0 model motor friction torque τ _fMc0 model motor coulomb friction torque J _L0 model load inertia moment D _L0 model load viscosity coefficient τ _fL0 model load friction torque τ _fLc0 model load coulomb friction torque BL ₀ model dead band width K _R0 model torsional stiffness coefficient F deadband characteristic function F 'backlash displacement function G _M motor friction characteristic function G _L load friction characteristic function

Claims (7)

A parameter identification device for identifying a model parameter of a two-inertia system model imitating a mechanical system in which a motor and a load are coupled by a coupling member,
A data acquisition unit for acquiring a torque command for the motor, measured values of the angle and angular velocity of the motor, and measured values of the angle and angular velocity of the load;
A phase plane generator that generates a plurality of phase planes based on the acquired torque command, measured values of the angle and angular velocity of the motor, and measured values of the angle and angular velocity of the load;
Based on the generated plurality of phase diagrams, model parameters indicating inertia of the motor, viscosity, friction, inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member are set. A parameter identification unit to identify;
With
The parameter identification unit
A parameter identification device that identifies the model parameter based on both the phase plane acquired during the small amplitude operation of the load and the phase plane acquired during the large amplitude operation of the load. The parameter identification unit
Based on the phase diagram acquired during the small amplitude operation of the load, at least model parameters indicating the friction of the motor and the friction of the load are identified,
The parameter identification device according to claim 1, wherein at least model parameters indicating the inertia and viscosity of the motor, the inertia of the load, and the viscosity are identified based on the phase plane acquired during the large amplitude operation of the load. The phase plane generator is at least
A phase diagram showing the relationship between the angle of the load and the torque command;
A phase plane showing the relationship between the angular velocity of the motor or the load and the torque command;
A phase diagram showing the relationship between the load angle and the twist angle of the connecting member;
A phase plane showing a relationship between an angular velocity of the motor or the load and a twist angle of the connecting member;
The parameter identification device according to claim 1 or 2, wherein: The phase plane generator further includes:
The parameter identification device according to claim 3, wherein a phase plane diagram showing a relationship between the torque command and a twist angle of the connecting member is generated. The parameter identification device according to any one of claims 1 to 4,
The mechanical system;
A feedback control unit that calculates the torque command based on a deviation of a current angle with respect to a target angle of the load;
A feedforward control unit that calculates the torque command based on the model parameter identified by the parameter identification device;
A drive system comprising: A method for identifying model parameters of a two-inertia system model simulating a mechanical system in which a motor and a load are connected by a connecting member,
A data acquisition step for acquiring a torque command for the motor, an actual measurement value of the motor angle and angular velocity, and an actual measurement value of the load angle and angular velocity;
A phase plane diagram generating step for generating a plurality of phase plane diagrams based on the acquired torque command, measured values of the motor angle and angular velocity, and measured values of the load angle and angular velocity;
Based on the generated plurality of phase diagrams, model parameters indicating inertia of the motor, viscosity, friction, inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member are set. A parameter identification step to identify;
Have
The parameter identification step includes:
A parameter identification method comprising the step of identifying the model parameter based on both the phase plane acquired during the small amplitude operation of the load and the phase plane acquired during the large amplitude operation of the load. A computer of a parameter identification device for identifying a model parameter of a two-inertia system model imitating a mechanical system in which a motor and a load are coupled by a coupling member;
A data acquisition unit for acquiring a torque command for the motor, an actual value of the angle and angular velocity of the motor, and an actual value of the angle and angular velocity of the load;
A phase plane generator that generates a plurality of phase planes based on the acquired torque command, measured values of the angle and angular velocity of the motor, and measured values of the angle and angular velocity of the load;
Based on the generated plurality of phase diagrams, model parameters indicating inertia of the motor, viscosity, friction, inertia of the load, viscosity, friction, rigidity of the connecting member, and dead band width of the connecting member are set. A parameter identification unit to identify,
The parameter identification unit
A program for identifying the model parameter based on both the phase plane acquired during the small amplitude operation of the load and the phase plane acquired during the large amplitude operation of the load.

Images