JP2005168166A - Load characteristic operation device and motor control device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a load characteristic operation device that can exactly measure load characteristics even if a load has a large offset load, and to provide a motor control device that can precisely control a motor without receiving an influence of the offset load. <P>SOLUTION: The load characteristic operation device comprises: a torque detection means that detects torque generated by the motor; a speed detection means 30 that detects the rotational speed of the motor; and a load characteristic operation means 70 that operates at least one of a viscosity friction coefficient, kinetic friction and the offset load on the basis of the detected torque (T*) and the speed ω. The load characteristic operation means 70 performs the operation on the basis of the torque and the detected value of the speed when at least one of the magnitude of the detected torque (T*), the magnitude of the detected speed ω and the magnitude of the time variation of the detected speed becomes a preset prescribed value or larger with respect to each of the magnitudes. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a device for calculating load characteristics of a motor. The present invention also relates to a motor control device that controls a motor based on a position command, a speed command, or a torque command while considering the load characteristics of the motor.

  As a conventional motor control device, for example, Patent Document 1 describes one that identifies dynamic friction, viscous friction, and inertia (also called inertia load) using three sets of torque and speed information.

In FIG. 5, the block diagram of the motor control apparatus of patent document 1 is shown. The host controller 904 periodically loads the output torque T from the servo controller while the motor 901 is loaded, the angular acceleration a obtained by a known method such as the angular velocity ω, the difference from the position feedback, and the temporary storage buffer 941. Remember. Then, the data for three samples in which the torque T, the dynamic friction torque D, the viscous friction coefficient R, and the inertia load J of the following formula (1) are uniquely obtained in the temporary storage buffer 941 are input to the load calculator 942. Then, inertial load J, dynamic friction torque D, and viscous friction coefficient R are obtained.
T−D−R · ω = J · a (1)

More specifically, in FIG. 5, data (T0, ω0, a0), (T1, ω1, a1), and (T2, ω2, a2) for three samples at times t0, t1, and t2 are expressed by the formula ( Substituting each into 1), a simultaneous linear equation like the following formula (2) is created. Then, the inertial load J, the dynamic friction torque D, and the viscous friction coefficient R are obtained by solving for J, D, and R in Equation (2).
T0−D−R · ω0 = J · a0 (2)
T1−D−R · ω1 = J · a1
T2−D−R · ω2 = J · a2

JP-A-6-225564

  The conventional motor control device has a problem that it cannot correctly measure the load characteristics of a load having a large partial load.

  FIG. 6 is a schematic diagram of the load. 6A and 6B are schematic diagrams of loads when there is no partial load. An example of such a load is an electric slider placed on a horizontal plane with respect to the ground. The load characteristics of such a load can be correctly measured using a conventional motor control device. That is, when the platform moves to the right as shown in FIG. 6A, the dynamic friction torque D and the viscous friction torque R · ω act to the left. Also, as shown in FIG. 6B, when the table moves to the left, the dynamic friction torque D and the viscous friction torque R · ω act to the right. Thus, the external forces acting on the load are only the dynamic friction torque D and the viscous friction torque R · ω, and the equation (1) is established (the direction of rotation is positive). Therefore, the conventional motor control device can measure the correct inertia load J, dynamic friction torque D, and viscous friction coefficient R in a load without partial load, and therefore can correctly measure the load characteristics.

FIGS. 6C and 6D are schematic views of the load when there is a partial load. An example of such a load is an electric slider placed on a vertical plane with respect to the ground. The load characteristics of such a load cannot be measured correctly using a conventional motor control device. That is, when the table moves downward as shown in FIG. 6C, the dynamic friction torque D and the viscous friction torque R · ω act upward. Further, gravity m · g acts in the opposite downward direction. On the other hand, when the platform moves upward as shown in FIG. 6D, the dynamic friction torque D and the viscous friction torque R · ω act downward. Similarly, gravity m · g acts downward. Thus, not only the dynamic friction torque D and the viscous friction torque R · ω act on the load, but also the weight m · g. The following equation (3) holds when the table moves downward (the direction of rotation, that is, downward), and the following equation (4) holds when the table moves upward (the direction of rotation, ie, upward) And). Thus, when the direction of rotation changes, the direction of gravity m · g with respect to the dynamic friction D changes. Therefore, the conventional motor control device cannot measure the correct dynamic friction torque D and the viscous friction torque R · ω, and cannot accurately measure the load characteristics, under a load with a large unbalanced load.
T− (D−m · g) −R · ω = J · a (3)
T− (D + m · g) −R · ω = J · a (4)

  The present invention solves the above-described problems, and an object of the present invention is to provide a load characteristic calculation device capable of correctly measuring load characteristics even in a load with a large bias. It is another object of the present invention to provide a motor control device that can be accurately controlled without being affected by partial weighting.

The load characteristic calculation device of the present invention includes torque detection means for detecting torque generated by the motor,
Speed detection means for detecting the speed of the motor, and calculation means for calculating and outputting at least one of viscous friction coefficient, dynamic friction, and partial load as load characteristics based on the detected torque and speed.

  The calculating means calculates torque when at least one of the detected torque magnitude, the detected speed magnitude, and the detected temporal change magnitude is equal to or greater than a predetermined value set for each. The calculation may be executed based on the detected value of the speed.

Further, the calculation means includes an average viscous friction coefficient obtained by averaging the viscous friction coefficient calculated during the normal rotation of the motor and the viscous friction coefficient calculated during the reverse rotation of the motor, and the dynamic friction calculated during the normal rotation of the motor. And at least one of the average dynamic friction obtained by averaging the dynamic friction calculated during the reverse rotation of the motor may be obtained.
The computing means may compute at least one of a viscous friction coefficient, dynamic friction, and partial weight using a sequential least square method based on detected values of speed and torque.

  A motor control device according to the present invention has a speed gain and a speed integration time constant based on at least one of the load characteristic calculation device and the viscous friction coefficient, dynamic friction, and partial load output from the load characteristic calculation device. A speed control unit that generates and outputs a torque command that is an output torque command generated by the motor, and a drive unit that generates and outputs a motor drive voltage according to the torque command.

  In addition, the motor control device includes a torque command correction unit that corrects the torque command output from the speed control unit based on at least one of the viscous friction coefficient, the dynamic friction, and the partial load output from the load characteristic calculation device. Further, it may be provided. At this time, the drive means generates and outputs a motor drive voltage in accordance with the corrected torque command.

  In addition, the motor control device generates a position command as a proportional constant and generates and outputs a speed command, a gain ratio indicating a ratio of the position gain to the speed gain based on at least one of the viscous friction coefficient and the dynamic friction, Gain changing means for changing at least one of the speed integration time constants may be further included.

  The gain changing means may change the gain ratio so that the gain ratio increases in accordance with at least one increase in the viscous friction coefficient and the dynamic friction.

  The gain changing means changes the speed integration time constant so that the speed integration time constant becomes smaller according to at least one increase in the viscous friction coefficient and the dynamic friction.

  The torque command correcting unit and the gain changing unit may perform the above correction and change based on the average viscous friction coefficient and the average dynamic friction.

  According to the present invention, the load characteristic of the motor can be calculated more accurately by calculating at least one of a viscous friction coefficient, dynamic friction, and partial load based on torque and speed.

  In addition, since the load characteristics are calculated based on the torque and speed when at least one of the magnitude of the speed, the magnitude of the torque, and the magnitude of the temporal change in the speed is equal to or greater than the corresponding value, the influence of noise Small information can be used, and load characteristics with high accuracy can be calculated.

  In addition, by correcting the torque command, which is a command for the output torque generated by the motor, based on at least one of the viscous friction coefficient, dynamic friction, and partial load, the torque command can be corrected with a simple calculation in a device with large partial load. Even if the rotation direction changes, the same characteristics can be realized with the same gain.

  Also, an optimum control coefficient for friction is set by changing at least one of a gain ratio indicating a ratio of a position gain to a speed gain and a speed integration time constant based on at least one of a viscous friction coefficient and a dynamic friction. And optimum characteristics (for example, characteristics such that the settling time is shortened) can be obtained.

  Further, by changing the gain ratio so that the gain ratio increases as at least one of the viscous friction coefficient and the dynamic friction increases, it is possible to set an optimal gain for friction and to set an optimal characteristic (for example, settling time). (Characteristics such that is shortened).

  Further, by changing the integration time constant so that the integration time constant becomes smaller as at least one of the viscous friction coefficient and the dynamic friction increases, it is possible to set an optimal integration time constant for friction, and to set an optimum characteristic (for example, The characteristic that the settling time is shortened) is obtained.

  Further, by averaging the load characteristics obtained at the time of forward rotation and reverse rotation of the motor rotation, the influence of noise can be reduced, and highly accurate load characteristics can be calculated.

  In addition, a configuration in which at least one of a viscous friction coefficient, a dynamic friction, and a partial load is calculated using a sequential least square method based on speed and torque, and load characteristics are calculated in real time, and control according to the load characteristics is performed. By doing so, control with quick response can be realized.

  Hereinafter, embodiments of a load characteristic calculation device and a motor control device according to the present invention will be described in detail with reference to the accompanying drawings.

(Overall configuration of motor control device)
FIG. 1 is a configuration diagram of a motor control device of the present invention. The motor control apparatus 1 includes a rotary encoder 4, a command generator 5, a position command receiver 6, a position receiver 7, a driver 8, a position deviation calculator 10, a position controller 20, a speed detector 30, and a speed deviation calculator. 40, a speed controller 50, a torque command corrector 60, and a load characteristic calculator 70.

  The functions of the position deviation calculator 10, the position controller 20, the speed detector 30, the speed deviation calculator 40, the speed controller 50, the torque command corrector 60, and the load characteristic calculator 70 are controlled by a microcomputer. It can be realized by executing.

  A motor 2 is connected to the motor control device 1, and the rotation of the rotor of the motor 2 is controlled by the motor control device 1. A load 3 is connected to the motor 2 via a coupling or the like, and performs a predetermined operation according to the rotation of the rotor of the motor 2.

  The rotary encoder 4 is connected to the rotor shaft of the motor 2 and outputs a signal indicating the position of the rotor of the motor 2 to the motor control device 1.

  The command generator 5 outputs a signal indicating the position command of the rotor of the motor 2 to the motor control device 1.

  The position command receiver 6 receives a signal indicating the position command generated by the command generator 5 and creates a position command θ * which is a numerical value indicating the position command.

  The position receiver 7 receives a signal indicating the position generated by the rotary encoder 4 and creates a detection position θ that is a numerical value indicating the position of the rotor of the motor 2.

  The position deviation calculator 10, the position controller 20, the speed detector 30, the speed deviation calculator 40, the speed controller 50, the torque command corrector 60, and the load characteristic calculator 70 are used to calculate the position command θ * and the position θ. An input is performed by the microcomputer for every sampling cycle Ts to create a torque command T *, and the torque command T * is output to the driver 8.

  The driver 8 inputs the torque command T * and the detection position θ, and controls the terminal voltage of the motor 2 so that the torque output from the motor 2 is in accordance with the torque command T *. That is, the driver 8 generates and outputs a motor drive voltage according to the torque command T *.

The position deviation calculator 10 receives the position command θ * and the detected position θ, and creates and outputs a position deviation eθ as shown in the following equation (5). Here, eθ (i) is a position deviation in the calculation in the current cycle. Further, although not included in the equation (5), eθ (i−1) represents a position deviation in the calculation in the previous cycle. Hereinafter, the value of the current calculation is represented by (i), and the value of the previous calculation is represented by (i−1).
eθ (i) = θ * (i) −θ (i) (5)

The position controller 20 receives the position deviation eθ and the load characteristic vector ψ, creates a speed command ω * as shown in the following equation (6), and outputs the speed command ω *. Note that Kpp is a position proportional gain and changes depending on the load characteristic vector ψ, as will be described later.
ω * (i) = Kpp (i) · eθ (i) (6)
Kpp (i) = Kpp (ψ (i-1)) (7)

The speed detector 30 receives the detection position θ and differentiates it to obtain a speed (detection speed) ω. Here, the speed ω is obtained as shown in the following equation (8) by taking the difference instead of differentiating. Ts is a sampling period as described above.
ω (i) = {θ (i) −θ (i−1)} / Ts (8)

The speed deviation calculator 40 receives the speed command ω * and the detected speed ω, creates a speed deviation eω as shown in the following equation (9), and outputs the speed deviation eω.
eω (i) = ω * (i) −ω (i) (9)

The speed controller 50 receives the speed deviation eω and the load characteristic vector ψ, creates a pre-correction torque command Tp * as shown in the following equation (10), and outputs the pre-correction torque command Tp *. Kvp is a speed proportional gain. Further, τvi is a speed integration time constant, and changes depending on the load characteristic vector ψ, as will be described later. Further, Σ represents integration, and is the sum of speed deviation eθ (i) of the current calculation (indicated by i) from speed deviation eω (j) at the integration start time (indicated by j). Note that J (i-1) is inertia, and how to find it will be described later.
Tp * (i) = Kvp (i) · J (i−1) · {eω (i)
+ Σ (eω (i)) · Ts / τvi (i)} (10)
τvi (i) = τvi (ψ (i−1)) (11)

The torque command corrector 60 receives the pre-correction torque command Tp * and the load characteristic vector ψ, creates a torque command T * as shown in the following equation (12), and outputs the torque command T *. Th is a torque command correction amount and varies depending on the load characteristic vector ψ, as will be described later.
T * (i) = Tp * (i) + Th (i) (12)
Th (i) = Th (ψ (i-1)) (13)

  The load characteristic calculator 70 calculates the load characteristic and outputs it. The load characteristics include inertia J, viscous friction coefficient R, dynamic friction D, and partial load G. Specifically, the load characteristic calculator 70 receives the detected speed ω and the detected torque value, and creates and outputs a load vector ψ based on these values. In the present embodiment, the torque command value T * is used as the torque detection value instead of actually detecting the torque. Further, an estimated value of torque obtained by calculation using a motor model from a motor current or the like can also be used as the detected torque value. Here, the operation of the load characteristic calculator 70 will be described in detail.

(Load characteristic calculator)
As described above, FIG. 6 is a schematic diagram of the load, and FIGS. 6A and 6B are schematic diagrams of the load when there is no partial weighting. FIGS. ) Is a schematic diagram of the load when the partial load is large. Although the direction of partial weighting changes depending on how to take the positive direction of coordinates, the equation of motion is as shown in the following equation (14).
T = J · a + R · ω + Td (14)
T is a torque, J is an inertia, a is an acceleration, R is a viscous friction coefficient, ω is a speed as described above, and Td is a composite value of dynamic friction D and partial load G. 6C and 6D, the downward direction is defined as positive.

As shown in FIG. 6C, when the load moves downward, that is, in a positive direction (speed ω> 0), the direction of the dynamic friction D and the direction of the partial load G are opposite as in the following equation (15). is there. As shown in FIG. 6D, when the load moves upward, that is, in a negative direction (speed ω <0), the direction of the dynamic friction D and the direction of the partial load G are expressed by the following equation (16). Match. Thus, when the rotation direction changes, the direction of the partial load G with respect to the dynamic friction D changes, and the composite value Td changes.
T = J · a + R · ω + (+ DG) (when ω> 0) (15)
T = J · a + R · ω + (−D−G) (when ω <0) (16)

If the torque T, the acceleration a, and the speed ω are known and there is a sufficient set of information, the inertia J, the viscous friction coefficient R, and the composite value Td can be obtained using the sequential least square method. . Therefore, the inertia J, the viscous friction coefficient R, and the combined value Td are obtained for each of the forward rotation and the reverse rotation, and the dynamic friction D and the partial weight G are obtained from the combined value Td during the normal rotation and the reverse rotation. . This will be specifically described below.
The load characteristic calculator 70 first determines the state based on the speed ω, the acceleration a, and the torque command T. Specifically, the load characteristic calculator 70 sets a state flag (Flag) based on the following equation (17).
Flag = “F” (when ω> ω0, | a |> a0, | T * |> T0)
Flag = “R” (when ω <−ω0, | a |> a0, | T * |> T0)
Flag = “S” (other than above) (17)
Here, ω0, a0, and T0 are positive numbers indicating threshold values of the speed ω, the acceleration magnitude | a |, and the torque command magnitude | T * | used to determine the state, respectively.

  State when speed ω is greater than threshold value ω0, acceleration magnitude | a | is greater than threshold value a0, and torque command magnitude | T * | is greater than threshold value T0 (when all three conditions are satisfied) The flag Flag = “F”. In this case, the status flag indicates that a calculation for forward rotation is performed.

  When the speed ω is smaller than the threshold −ω0, the acceleration magnitude | a | is larger than the threshold a0, and the torque command magnitude | T * | is larger than the threshold T0 (when all three conditions are satisfied). The status flag Flag = “R”. In this case, the status flag indicates that a calculation for reverse rotation is performed.

  When the status flag Flag does not correspond to either “F” or “R”, the status flag Flag = “S”. In this case, the status flag indicates that no calculation is performed.

As described above, the present embodiment has the following three criteria for setting the status flag.
i) Whether the speed ω is greater than a predetermined value.
ii) Whether the magnitude of the acceleration a is larger than a predetermined value.
iii) Whether the magnitude of the torque T is larger than a predetermined value.

However, in setting the status flag, it is not always necessary to consider all of the above three criteria i) to iii), and the flag is set in consideration of at least one of the above three criteria i) to iii). It may be set. For example, the status flag may be set as follows.
Flag = “F” (when ω> ω0)
Flag = “R” (when ω <ω0)
Flag = “S” (other than the above) (17a)
Or
Flag = “F” (when ω> 0, | a |> a0)
Flag = “R” (when ω <0, | a |> a0)
Flag = “S” (other than above) (17b)
Or
Flag = “F” (when ω> 0, | T |> T0)
Flag = “R” (when ω <0, | T |> T0)
Flag = “S” (other than the above) (17c)

  In addition, in the above-described determination criterion for setting the state flag, the speed command ω * or the time derivative {d (θ *) / dt} of the position command θ * may be used instead of the detected speed ω.

When the set state flag Flag is “F”, the calculation for forward rotation is performed. The relationship among the torque command T *, the acceleration a, and the speed ω, the inertia Jf at the time of forward rotation, the viscous friction coefficient Rf, and the composite value Tdf is expressed by the following formula (18). Note that the driver 8 uses the torque command T * instead of the torque T in order to control the motor 2 to output a torque corresponding to the torque command T *. The acceleration a is a time derivative of speed, but can be obtained using the difference as shown in the following equation (19). Here, as described above, ω (i) is the speed of the current calculation, and ω (i−1) is the speed of the previous calculation. As described above, Ts is a sampling period.
T * = Jf · a + Rf · ω + Tdf (18)
a (i) = {ω (i) −ω (i−1)} / Ts (19)
Then, Jf, Rf, and Tdf in Equation (18) are estimated using the successive least square method.

The input matrix φ (i) is defined as the following formula (20).

Further, the parameter estimation value matrix ξf (i) at the time of forward rotation is defined as the following equation (21).

Then, the prediction error εf (i) at the time of forward rotation is expressed by the following equation (22), and the covariance matrix Pf (i) at the time of normal rotation is expressed by the following equation (23). Note that the covariance matrix Pf (i) during forward rotation is a matrix of 3 rows × 3 columns.

Then, using the prediction error εf (i) at the time of normal rotation and the covariance matrix Pf (i) at the time of normal rotation, a parameter estimated value matrix ξf (i) is obtained as in the following equation (24). Here, λ is a forgetting factor, and a value less than 1 is used.

  In this way, the inertia Jf during normal rotation, the viscous friction coefficient Rf during normal rotation, and the combined value Tdf during normal rotation are estimated.

Further, when the state Flag = “R”, the calculation for reverse rotation is performed. The relationship between the torque command T *, the acceleration a, and the speed ω, the inertia Jr at the time of reverse rotation, the viscous friction coefficient Rr, and the composite value Tdr is as shown in the following formula (25).
T * = Jr · a + Rr · ω + Tdr (25)

  As in the forward rotation, the driver 8 uses the torque command T * instead of the torque T in order to control the motor 2 to output the torque according to the torque command T *. Moreover, although acceleration a is a time derivative of speed, it calculates | requires using a difference like Formula (19).

Then, Jr, Rr, and Tdr in Equation (25) are estimated using the successive least square method. An input matrix φ (i) is defined as in Expression (20). Further, the parameter estimation value matrix ξr (i) at the time of reverse rotation is defined as the following equation (26).

Then, the prediction error εr (i) at the time of inversion becomes the following equation (27), and the covariance matrix Pr (i) at the time of inversion becomes the following equation (28).

The covariance matrix Pr (i) at the time of inversion is a 3 × 3 matrix. Then, using the prediction error εr (i) at the time of reverse rotation and the covariance matrix Pr (i) at the time of forward rotation, a parameter estimated value matrix ξr (i) is obtained as in the following equation (29). Here, λ is a forgetting factor, and a value less than 1 is used. In this way, the inertia Jr during reverse rotation, the viscous friction coefficient Rr during reverse rotation, and the combined value Tdr during reverse rotation are estimated.

Note that when the state flag Flag = “R” (a state in which the calculation for reverse rotation is performed) or “S” (a state in which the calculation is not performed), the inertia Jf at the time of forward rotation, the viscous friction coefficient Rf, and the composite value Tdf Holds the value of the previous cycle as it is as in the following formulas (30), (31), and (32).
Jf (i) = Jf (i-1) (30)
Rf (i) = Rf (i-1) (31)
Tdf (i) = Tdf (i-1) (32)

It should be noted that when the state flag Flag = “F” (state in which the forward rotation calculation is performed) or the state flag Flag = “S” (state in which the calculation is not performed), the inertia Jr, the viscous friction coefficient Rr, The synthesized value Tdr holds the value of the previous cycle as it is as in the following formulas (33), (34), and (35).
Jr (i) = Jr (i-1) (33)
Rr (i) = Rr (i-1) (34)
Tdr (i) = Tdr (i-1) (35)

And load characteristic (psi) is calculated like following formula (36). The load characteristic ψ is a 4-row × 1-column matrix having four elements of inertia J, viscous friction coefficient R, dynamic friction D, and partial load G, and is obtained as follows. Inertia J is an average of inertia Jf during forward rotation and inertia Jr during reverse rotation. The viscous load coefficient R is an average of the viscous load coefficient Rf during forward rotation and the viscous load coefficient Rr during reverse rotation. Then, using the relationship of the equations (14), (15), and (16), as shown in the equation (36), the dynamic friction D and the partial weight G are calculated from the combined value Tdf at the time of forward rotation and the combined value Tdr at the time of reverse rotation. Ask.

  As described above, based on the speed ω and the torque command T *, the parameter estimation matrices ξf and ξr at the time of forward rotation and reverse rotation are obtained by using the sequential least square method, and the load characteristic ψ is obtained. As described above, the load characteristic ψ can be calculated in real time by calculating the load characteristic ψ for each cycle using the sequential least square method. Then, by executing various corrections using this load characteristic, it is possible to realize control with quick response.

(Position proportional gain)
In the present embodiment, the position controller 20 changes the position proportional gain Kpp based on the load characteristics as shown in Expression (7). FIG. 2 is a diagram showing a change in the position proportional gain with respect to the viscous friction coefficient R in the position controller 20 in the present embodiment. As shown in FIG. 2, the position proportional gain Kpp is changed so that the position proportional gain Kpp increases as the viscous friction coefficient R increases. Specifically, it is changed as shown in the following formula (37).
When R (i-1) <R11, Kpp (i) = Kpp1
When R11 ≦ R (i−1) <R12, Kpp (i) =
Kpp1 + (R (i-1) -R11). (Kpp2-Kpp1) / (R12-R11)
When R (i−1) ≧ R12 Kpp (i) == Kpp2 (37)

  That is, the position proportional gain Kpp is changed so that the ratio (gain ratio) of the position proportional gain Kpp to the speed proportional gain Kvp increases as the viscous friction coefficient R increases. Thereby, the optimal gain (position proportional gain Kpp, speed proportional gain Kvp) with respect to friction can be set, and optimal characteristics (for example, characteristics in which settling time is shortened) can be obtained.

When the load is a rigid body, the load transfer function GL (s) is expressed by the following equation (38). Here, s is a Laplace variable.

Then, the transfer function G (s) from the position command θ * to the position θ is expressed by the following equation (39).

  When the viscous friction coefficient R is increased, damping is improved. Therefore, even if the position proportional gain Kpp is increased by that amount, vibration or the like does not occur. Since the position command gain Kpp can be increased by that much, the cut-off frequency of the transfer function is increased and the settling time can be shortened.

  Although the position proportional gain Kpp is changed according to the viscous friction coefficient R, it may be changed according to other load characteristics. In particular, since the viscous friction coefficient R and the dynamic friction D have similar characteristics, the position proportional gain Kpp may be changed according to the dynamic friction D. Further, the position proportional gain Kpp may be changed according to the viscosity friction coefficient R and the dynamic friction D added at a certain ratio. Further, the control for changing the position proportional gain Kpp according to the viscous friction coefficient R and the control for changing the position proportional gain Kpp according to the dynamic friction D may be switched according to the speed ω or the like. Further, similarly to the position proportional gain Kpp, the speed proportional gain Kvp may be changed according to the load characteristics. That is, the gist of the present invention is to change the ratio of the position proportional gain and the speed proportional gain according to the load characteristics such as the viscous friction coefficient R and the dynamic friction D, and various modifications are possible. Are also included within the scope of the present invention.

(Speed integration time constant)
In the present embodiment, the speed controller 50 changes the speed integration time constant τvi based on the load characteristics as shown in Expression (11). FIG. 3 is a relationship diagram of the speed integration time constant τvi with respect to the viscous friction coefficient R in the speed controller 50 in the present embodiment. As shown in FIG. 3, when the viscous friction coefficient R is increased, the speed integration time constant is changed to be decreased. Specifically, it is changed as in the following formula (40).
When R (i-1) <R21,
τvi (i) = τvi1
When R21 ≦ R (i−1) <R22,
τvi (i) =
.tau.vi1 + (R (i-1) -R21). (. tau.vi2-.tau.vi1) / (R22-R21)
When R (i-1) ≧ R22,
τvi (i) = τvi2 (40)

  As described above, by changing the speed integration time constant τvi based on the load characteristic ψ, the optimum speed integration time constant τvi for friction can be set, and the optimum characteristics (for example, settling time is shortened). Such characteristics) can be obtained.

  In general, static friction shows the same tendency as the viscous friction coefficient R and dynamic friction D, and the static friction increases as the viscous friction coefficient R and dynamic friction D increase. Further, since the static friction acts as a brake against the load at the time of stopping, if the static friction is large, the speed integration time constant τvi can be reduced by that much, and the speed integration can be accelerated. Since the static friction is settled by the speed integral being compensated (the position deviation eθ falls within the range of the static definite width which is a set value), the settling time (position command θ * becomes smaller as the speed integral time constant τvi is smaller. After no longer changing, the time until the position deviation eθ enters the settling width is shortened.

  The speed integration time constant τvi may be changed according to other load characteristics other than the viscous friction coefficient R. In particular, since the viscous friction coefficient R and the dynamic friction D have similar characteristics, the speed integration time constant τvi may be changed according to the dynamic friction D. Further, the speed integration time constant τvi may be changed according to a value obtained by multiplying the viscous friction coefficient R and the dynamic friction D by respective predetermined ratios and adding them. Further, the control for changing the speed integration time constant τvi according to the viscous friction coefficient R and the control for changing the speed integration time constant τvi according to the dynamic friction D may be switched according to the speed ω. Further, the ratio of the proportional to the integrator in the speed controller 50 may be changed instead of directly changing the speed integration time constant τvi.

(Torque command correction)
The torque command correction device 60 sets the torque command correction amount Th, which is obtained by changing the sign of the partial load G of the previous cycle, as a new torque command correction amount Th as shown in the following equation (41).
Th (i) = − G (i−1) (41)

  As described above, the motor control device according to the present embodiment corrects the torque command T * based on the partial load G (one of the load characteristics ψ), so that the torque command can be calculated with a simple calculation in a device with a large partial load G. By correcting T *, the same characteristics (for example, the same settling time) can be realized with the same gain even if the rotation direction changes. In order to obtain the same effect (the same gain and the same characteristic even if the rotation direction changes), a method of using a different gain depending on the rotation direction can be considered, but it is necessary to set two sets of gains, The time required for gain setting becomes longer. In this embodiment, the same gain may be used, and the time required for gain setting is short.

In the torque command corrector 60, the torque command correction amount Th is obtained by changing the sign of the partial weight G. However, the torque command may be corrected according to other load characteristics. For example, it may be corrected by the partial load G and the dynamic friction D as in the following formula (42).
Th (i)
= −G (i−1) + D (i−1) (when ω> 0)
= −G (i−1) −D (i−1) (when ω <0) (42)

Moreover, you may correct | amend by the biased weight G, the dynamic friction D, and the viscous friction coefficient R like following formula (43).
Th (i)
= −G (i−1) + D (i−1) + R (i−1) · ω (i) (when ω> 0)
= −G (i−1) −D (i−1) + R (i−1) · ω (i) (when ω <0)
... (43)

(Operation of motor controller)
The overall processing flow of the motor control device 1 in this embodiment will be described. FIG. 4 shows the position deviation calculator 10, the position controller 20, the speed detector 30, the speed deviation calculator 40, the speed controller 50, the torque command corrector 60, and the load characteristic calculator 70 in the motor control device 1. It is a flowchart which shows the flow of the process performed.

  A subroutine for performing the processing shown in FIG. 4 is activated by an interrupt that occurs every sampling time Ts. When the subroutine is activated, the position command θ * and the detected position θ are input (step S11). Next, the position deviation calculator 10 calculates the position deviation eθ based on the position command θ * and the detected position θ as shown in Expression (5) (step S2).

  Next, the position controller 20 calculates the speed command ω * using the equations (6) and (7) based on the position deviation eθ and the load characteristic ψ (step S13). At this time, the position proportional gain Kpp is calculated according to the equation (37).

  Next, the speed differentiator 30 calculates the speed ω according to the equation (8) based on the detected position θ (step S14).

  Next, the speed deviation calculator 40 calculates a speed deviation eω based on the speed command ω * and the speed ω as shown in Expression (9) (step S5).

  Next, based on the speed deviation eω and the load characteristic ψ, the speed controller 50 calculates a pre-correction torque command Tp * as shown in equations (10) and (11) (step S16). At this time, the speed integration time constant τvi is calculated as shown in Expression (38).

  Then, the torque command corrector 60 calculates the torque command T * as shown in the equations (12) and (13) based on the pre-correction torque command Tp * and the load characteristic ψ (step S17). At this time, the torque command correction amount Th is calculated according to the equation (41), and the pre-correction torque command is corrected by the torque command correction amount Th to obtain the corrected torque command T *. The corrected torque command T * is output (step S18).

  Thereafter, the state flag Flag is set according to the equation (17) based on the speed ω and the torque command T * (step S19). Thereafter, the value of the status flag is determined (step S20), and processing corresponding to the value is performed (steps S21 to S23).

  When the state flag Flag = “F” (a state in which a forward rotation calculation is performed), based on the speed ω and the torque command T *, the inertia Jf, The viscous friction coefficient Rf and the composite value Tdf are calculated, and the inertia Jr, the viscous friction coefficient Rr, and the composite value Tdr at the time of reverse rotation in the previous cycle are set as values of the current cycle according to the equations (33) to (35). Hold (step S21). Thereafter, step S24 is executed.

  When the state flag Flag = “R” (a state in which the operation at the time of reverse rotation is performed), based on the speed ω and the torque command T *, according to the equations (19) and (25) to (29), the inertia at the time of reverse rotation Jr, viscous friction coefficient Rr, and composite value Tdr are calculated, and according to equations (30) to (32), inertia Jf, viscous friction coefficient Rf, and composite value Tdf during forward rotation in the previous cycle are the current cycle. (Step S22). Thereafter, step S24 is executed.

  When the state flag Flag = “S” (state in which no calculation is performed), according to the equations (30) to (35), the inertia Jf, the viscous friction coefficient Rf, and the composite value Tdf at the time of forward rotation in the previous cycle, The inertia Jr at the time of reverse rotation, the viscous friction coefficient Rr, and the composite value Tdr are held as values of the current cycle (step S23). Thereafter, step S24 is executed.

  In step S24, the load characteristic ψ is calculated according to the equation (36) based on the inertia Jf, the viscous friction coefficient Rf and the combined value Tdf at the time of forward rotation, and the inertia Jr, the viscous friction coefficient Rr and the combined value Tdr at the time of reverse rotation. To do.

  The load characteristics obtained in this way are the load characteristics calculated based on the torque command T * and speed ω during forward rotation, and the load characteristics calculated based on the torque command T * and speed ω during reverse rotation. Average value. That is, the inertia J is an average of the inertia Jf at the time of forward rotation and the inertia Jr at the time of reverse rotation. The viscous friction coefficient R is an average of the viscous friction coefficient Rf during forward rotation and the viscous friction coefficient Rr during reverse rotation. Since the influence of noise can be reduced by obtaining the average value in this way, a highly accurate load characteristic ψ (inertia J and viscous friction coefficient R) can be obtained. Various corrections with high accuracy can be realized by using the load characteristic ψ with high accuracy.

  The above steps S19 to S24 are executed by the load calculator 70.

  The motor control device of the present embodiment configured as described above has a load characteristic (viscous friction coefficient R, dynamic friction D, partial load G) based on a detected torque value (using a torque command value as an example) and a detected speed value. ) Is calculated. Thereby, it is possible to describe a detailed load characteristic, and to perform various corrections using this load characteristic.

  In addition, the motor control device according to the present embodiment calculates the load characteristic ψ based on the detected torque value and the detected speed ω when the magnitude of the detected torque value (torque command T *) is larger than the threshold value. As a result, only torque values having a large S / N ratio (signal / noise ratio) can be used, and the influence of noise can be reduced. Furthermore, since various corrections can use the load characteristic ψ with high accuracy, there is an effect that various corrections with high accuracy can be performed.

  In addition, the motor control device of the present embodiment performs the calculation of the load characteristic ψ based on the detected torque value and the speed ω when the speed magnitude | ω | is larger than the threshold ω0. As a result, only the speed ω having a large S / N ratio (signal-to-noise ratio) can be used, and the influence of noise can be reduced. Therefore, there is an advantage that the load characteristic ψ with high accuracy can be calculated. Furthermore, since various corrections can use the load characteristic ψ with high accuracy, there is an effect that various corrections with high accuracy can be performed.

  In addition, the motor control device according to the present embodiment performs the calculation of the load characteristic ψ based on the torque value and the speed ω when the magnitude | a | of the acceleration (time change in speed) is larger than the threshold value a0. As a result, only the acceleration a having a large S / N ratio (signal / noise ratio) is used, and the influence of noise can be reduced. Furthermore, since various corrections can use the load characteristic ψ with high accuracy, there is an effect that various corrections with high accuracy can be performed.

  In the present embodiment, a signal indicating time differentiation of the position command may be used instead of the signal indicating the position command θ *.

  Further, the motor control device of the present embodiment controls the rotor position based on the position command θ *, but the idea related to the calculation of the load characteristic ψ of the present invention is a motor control device that inputs a speed command and controls the speed. Alternatively, the present invention can be applied to a motor control device that controls torque by inputting a torque command.

  The additional characteristic calculation device of the present invention can accurately measure the load characteristic even in a load with a large bias, and can be used for a device for measuring a motor load and a motor control device. The motor control device of the present invention is useful for a motor control device that controls a motor in accordance with a position command, and is also applicable to a motor control device that controls a motor in accordance with a speed command. Further, the present invention can be applied to a motor control device that controls a motor in accordance with a torque command.

Configuration diagram of motor control device of the present invention Relationship diagram of position proportional gain with respect to viscous friction coefficient in position controller of motor control device of present invention Relationship diagram of speed integral time constant to viscous friction coefficient in speed controller of motor control apparatus of present invention Flowchart of calculation of position deviation calculator, position controller, speed detector, speed deviation calculator, speed controller, torque command corrector, and load characteristic calculator of the motor controller of the present invention Configuration diagram of a conventional motor control device It is a schematic diagram of the load, (a), (b) is a schematic diagram of the load when there is no partial load, (c), (d) is a schematic diagram of the load when the partial load is large

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Motor controller 2 Motor 3 Load 4 Rotary encoder 5 Command generator 10 Position deviation calculator 20 Position controller 30 Speed detector 40 Speed deviation calculator 50 Speed controller 60 Torque command corrector 70 Driver 80 Load characteristic calculator

Claims (14)

Torque detecting means for detecting torque generated by the motor;
Speed detecting means for detecting the speed of the motor;
A load characteristic calculation apparatus comprising: a calculation unit that calculates and outputs at least one of a viscous friction coefficient, dynamic friction, and partial load as a load characteristic based on the detected torque and speed.   In the calculation means, at least one of the magnitude of the detected torque, the magnitude of the detected speed, and the magnitude of the time change of the detected speed is equal to or greater than a predetermined value set for each. The load characteristic calculation device according to claim 1, wherein the calculation is performed based on detected values of torque and speed.   The calculating means calculates an average viscous friction coefficient obtained by averaging the viscous friction coefficient calculated at the time of forward rotation of the motor and the viscous friction coefficient calculated at the time of reverse rotation of the motor, and calculated at the time of forward rotation of the motor. 3. The load characteristic calculation device according to claim 1, wherein at least one of the average dynamic friction obtained by averaging the dynamic friction and the dynamic friction calculated at the time of reverse rotation of the motor is obtained.   3. The calculating means according to claim 1, wherein the calculating means calculates at least one of a viscous friction coefficient, dynamic friction, and partial weight using a sequential least square method based on the detected values of the speed and torque. Load characteristic calculation device. The load characteristic calculation device according to claim 1;
Based on at least one of the viscous friction coefficient, dynamic friction, and partial load output from the load characteristic calculation device, a torque command that is an output torque command generated by the motor is generated using a speed gain and a speed integration time constant. Output speed control means;
A motor control device comprising drive means for generating and outputting a motor drive voltage in accordance with the torque command. Torque command correction means for correcting the torque command output from the speed control means based on at least one of the viscous friction coefficient, dynamic friction and partial load output from the load characteristic calculation device,
6. The motor control device according to claim 5, wherein the driving means generates and outputs a motor driving voltage in accordance with the corrected torque command. Position control means for generating and outputting a speed command with a position gain as a proportional constant; and
The apparatus further includes a gain ratio indicating a ratio of the position gain to the speed gain based on at least one of the viscous friction coefficient and the dynamic friction, and a gain changing unit that changes at least one of the speed integration time constant. The motor control device according to claim 5.   The motor control device according to claim 7, wherein the gain changing unit changes the gain ratio so that the gain ratio increases in accordance with at least one increase in the viscous friction coefficient and the dynamic friction.   The said gain change means changes the said speed integration time constant so that the said speed integration time constant becomes small according to at least one increase of the said viscous friction coefficient and the said dynamic friction. Motor control device. The load characteristic calculation device according to claim 3,
Based on at least one of the viscous friction coefficient, dynamic friction, and partial load output from the load characteristic calculation device, a torque command that is an output torque command generated by the motor is generated using a speed gain and a speed integration time constant. Output speed control means;
A motor control device comprising drive means for generating and outputting a motor drive voltage in accordance with the torque command. Torque command correction means for correcting the torque command output from the speed control means based on at least one of the average viscous friction coefficient, average dynamic friction, and partial load output from the load characteristic calculation device,
11. The motor control device according to claim 10, wherein the driving means generates and outputs a motor driving voltage in accordance with the corrected torque command. Position control means for generating and outputting a speed command with a position gain as a proportional constant; and
A gain ratio indicating a ratio of the position gain to the speed gain based on at least one of the average viscous friction coefficient and the average dynamic friction; and a gain changing means for changing at least one of the speed integration time constants. The motor control device according to claim 10, further comprising:   13. The motor control device according to claim 12, wherein the gain changing unit changes the gain ratio so that the gain ratio increases as at least one of the average viscous friction coefficient and the average dynamic friction increases. . 13. The gain changing unit changes the speed integration time constant so that the speed integration time constant becomes smaller as at least one of the average viscous friction coefficient and the average dynamic friction increases. Motor control device.

Images