JP3190088B2 - Control device for direct drive robot - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control device for a direct drive robot (hereinafter abbreviated as DD robot).

[0002]

2. Description of the Related Art A schematic configuration of a horizontal DD robot is shown in FIG.
Shown in In FIG. 2, 1 is one axis and 2 is two axes. FIG. 8 shows a conventional position control system when each of the axes 1 and 2 in such a DD robot is controlled independently.

In FIG. 8, 3 is a PID controller for a position loop, 4 is a PID controller for a speed loop, and 5
Is a current amplifier composed of a PWM inverter, 6 is a transfer function G (S) of a DD robot single axis, 7 is a velocity filter composed of a low-pass filter, θ * is a position command, θ is a position, ω * is a velocity command, ω Is the speed.

In such a control system, the position / velocity loop is qualitatively controlled by a PID controller without considering the dynamics (transfer function) of the DD robot as a load.

[0005]

However, in a qualitatively constructed control system as shown in FIG. 8, the servo parameters (control gain and cutoff frequency of the speed filter) are quantitatively determined. It was difficult, and humans have made adjustments based on experience and intuition. However, in such an adjustment method, it takes a long time to perform the adjustment, and the adjustment varies, and it is doubtful that the adjustment is optimal. Also,
The DD robot cannot immediately respond to a change in the moment of inertia due to a load.

[0006] In view of the above-mentioned conventional problems, the present invention provides a DD
It is an object of the present invention to provide a control device for a DD robot capable of automatically adjusting a servo parameter of the robot.

[0007]

According to the present invention, there is provided a DD robot control device comprising a set of a control gain calculated by an optimal regulator theory and a speed filter which does not narrow a frequency band of a transfer function from an input to an output determined by the gain. Are stored in a table, and when the robot makes a step response to each set, a set having a wide frequency band of the transfer function is adopted as an optimum servo parameter among sets whose maximum deviation after a predetermined time does not exceed a predetermined value. Means are provided.

[0008] Preferably, there is provided means for obtaining information on the moment of inertia necessary for obtaining a control gain using the optimal regulator theory by driving the robot with a constant amplitude current while the posture of the robot is fixed.

[0009]

According to the present invention, the optimum servo parameters of the DD robot can be automatically determined by the above-mentioned configuration, and if the means for obtaining the above-mentioned preferable moment of inertia is provided, it can be calculated without complicated calculations. Obtainable.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A control device for a DD robot according to an embodiment of the present invention will be described below with reference to FIGS.

The mechanical schematic configuration of the DD robot is the same as that of the conventional example shown in FIG. 2, and the description is omitted here and the description is omitted. The motor used in this DD robot is a rotating magnetic field type synchronous machine whose rotor is made up of permanent magnets. The equation of motion of this motor is as follows, assuming that the position is θ (t), the angular velocity is ω (t), and the amplitude (torque current) of the three-phase symmetric current is i (t).

[0012]

(Equation 1)

The target position value is θ * (t) = a (= constant), the position deviation is e (t) = θ * (t) −θ (t), and the load torque T ₁ = 0. Expressed as follows:

[0014]

(Equation 2)

Here, the evaluation function H is represented by q ₁ , q ₂ , q ₃ , r
Is defined as a weight as follows.

[0016]

(Equation 3)

At this time, the Riccati equation corresponding to (Equation 3) is as follows.

[0018]

(Equation 4)

Using the semi-definite solution of the Riccati equation, the optimal control v * (t) is determined as follows.

[0020]

(Equation 5)

From equation (5), the optimum control input i * (t) is determined as follows.

[0022]

(Equation 6)

FIG. 3 shows (Equation 6) in a block diagram. In FIG. 3, θ * is a position command, θ is a position, e is a position deviation, i * is a control output (current command), ω is an angular velocity, 11 is an integral element, 12 is a proportional element, 13 is a differential element, and 14 is The transfer function of the single axis of the DD robot, 15 is a low-pass filter.

In the control system described above, the weights q ₂ , q
_3, r is fixed, when gradually increasing the weight q ₁ position deviation e, the gain is increased, the control band is widened. Here, the control band indicates a point where the gain of the transfer function W from the input to the output decreases by -3 db as shown in FIG. FIG. 4 shows a case where there is no speed filter. However, actually, the control band is narrowed as shown in FIG. 5 by applying the speed filter. The allowable range of this narrowing size is determined in advance,
The lowest cutoff frequency in the range is set as the cutoff frequency of the speed filter,
Make a table.

Next, the set of the gain and the filter is applied to an actual robot to cause a step response. If the maximum deviation after a predetermined time t1 is smaller than a predetermined value ε as shown in FIG. And a set of filters, and if it is equal to or larger than ε as shown in FIG. 7, a set that narrows the control band is selected.

By repeating the above operations, the convergence is made to the optimal combination of gain and filter. FIG. 1 is a flowchart showing this state. In FIG. 1, step # 1 is a tabulation process of a gain and a cutoff frequency of a filter, step # 2 is a process of causing the robot to make a step response, and step # 3 is whether the steady-state deviation is larger or smaller than a predetermined value ε. Step # 4 is a process of selecting a combination of a gain and a filter for expanding the control band when the steady-state error is smaller than ε. Step # 5 is a gain for narrowing the control band when the steady-state error is larger than ε. This is a process for selecting a set of filters.

First, the adjustment of the first axis is performed according to FIG. 1 with the second axis 2 servo-off, and then the second axis 2 is adjusted with the first axis 1 servo-on. When adjusting the two axes 2, the fixed deviation of the one axis 1 is also taken into consideration in selecting the combination of the gain and the filter.

In the present embodiment, the target value θ * is set in a step shape at the time of designing the optimal regulator. However, in order to improve the target value follow-up characteristic, the target value is set in a ramp shape (in this case, the control system is shown in FIG. It goes without saying that the present invention can be similarly applied to other shapes such as a PID control system instead of the I-PD control system shown). In addition, in the case where feedforward compensation is applied to improve the target tracking characteristic, the present invention can be similarly applied if a table of gain and filter is created without adding feedforward compensation.

(Embodiment 2) In the Riccati equation of (Equation 4), d / J and K _t / J are used as robot parameters.
There are two. Therefore, if the posture is mechanically fixed so that the moment of inertia does not fluctuate and the following assumption is made in (Equation 1), 1 / J can be obtained by (Equation 7).

The current offset ioff and the load torque T
_i is zero Viscosity coefficient d and torque constant _Kt are known and constant regardless of moment of inertia J

[0031]

(Equation 7)

When 1 / J of the two axes 2 is obtained, the one axis 1 is mechanically fixed to the robot base, and a three-phase symmetric current having a constant amplitude of the two axes 2 flows, and the angular acceleration α around the two axes 2 is obtained. (T) and the angular velocity ω (t) may be measured. Similarly, 1 / axis 1
When J is obtained, a three-phase symmetric current having a constant amplitude is applied to one axis 1 by mechanically fixing the one axis 1 and the two axes 2 at a predetermined relative angle, and an angular velocity α (t) about one axis 1 is obtained. And the angular velocity ω (t) may be measured. If 1 / J is obtained in this manner, the viscosity coefficient d and the torque constant _Kt are known, and thus _Kt /
J and d / J can be determined.

[0033]

(Equation 8)

In the first and second embodiments, the robot having two horizontal axes has been described symmetrically, but it goes without saying that the present invention can be similarly applied to three or more axes.

[0035]

According to the present invention, a set of a control gain calculated by the optimal regulator theory as described above and a speed filter that does not narrow the frequency band of the transfer function from input to output determined by the gain is tabulated. When the robot makes a step response to each set, by adopting the set having the widest frequency band of the transfer function as the optimum servo parameter among the sets whose maximum deviation after a predetermined time does not exceed the predetermined value, It can realize automatic tuning of the robot.

Further, information on the moment of inertia necessary for obtaining the control gain using the optimal regulator theory can be obtained without any complicated calculation by driving the robot with a constant amplitude current while the posture of the robot is fixed. be able to.

[Brief description of the drawings]

FIG. 1 is a flowchart of auto tuning of a DD robot according to an embodiment of the present invention.

FIG. 2 is a perspective view showing a schematic configuration of a horizontal DD robot in the embodiment.

FIG. 3 is a block diagram of one control system in the embodiment.

FIG. 4 is a Bode diagram of a transfer function without a speed filter in the embodiment.

FIG. 5 is a Bode diagram of a transfer function in the embodiment with a speed filter.

FIG. 6 is an explanatory diagram of a deviation in a step response in the embodiment.

FIG. 7 is an explanatory diagram of a deviation in a step response in the embodiment.

FIG. 8 is a block diagram of one control system in a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Integral element 12 Proportional element 13 Differential element 14 Transfer function 15 Low-pass filter

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁷ identification code FI G05B 11/36 503 G05B 11/36 503C (58) Investigated field (Int.Cl. ⁷ , DB name) G05B 13/02 B25J 13 / 00 G05B 11/36 G05B 11/36 501 G05B 11/36 503 JICST file (JOIS)

Claims (2)

(57) [Claims] 1. A table of a set of a control gain calculated by the optimal regulator theory and a velocity filter which does not narrow a frequency band of a transfer function from an input to an output determined by the gain, and a step response to the robot is given to each set. Direct drive robot control, comprising means for adopting a set having a wide frequency band of a transfer function as an optimum servo parameter in a set in which a maximum deviation after a predetermined time does not exceed a predetermined value when the control is performed. apparatus. 2. Information on the moment of inertia necessary for obtaining a control gain using the optimal regulator theory is
2. The control device for a direct drive robot according to claim 1, further comprising means for obtaining a value obtained by driving the robot with a constant amplitude current in a state where the posture of the robot is fixed.