JPH06274201A - Control method and control device - Google Patents

Abstract

(57) [Summary] [Object] To provide a control method for stably controlling feedback control that is executed while changing a control gain for a control target having a strong nonlinearity. [Structure] The difference value Δx of the state quantity x output from the controlled object 1 from the previous time is obtained by the difference mechanism 4, and the feedback mechanism 2
Then, the difference value Δx is multiplied by the feedback coefficient Fx to obtain the deviation value Δu of the control command. Command deviation value Δu in the integration mechanism 5
Is integrated to determine a control command value (manipulation amount) u. The gain Fx of the feedback mechanism 2 is appropriately modified by the control parameter modification mechanism 3 during setpoint control.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control system having a state feedback system, and more particularly to stabilization of the control system due to changes in control parameters.

[0002]

2. Description of the Related Art In the current state feedback control, a theory for solving a differential equation to be controlled while remaining nonlinear has not been established. Therefore, when controlling an object to be controlled having a strong non-linearity, an operating point is determined, and feedback control is performed around the operating point, so that a control system is generally configured as a so-called regulator problem.

In the determination of the operating point in the control system, the change over time of the controlled object is ignored, and the steady solution of the differential equation is solved, which is called set point control or setup control. When the operating point is determined, the regulator is configured to linearly approximate the controlled object and settle the control at the operating point. This is because the modern control theory based on the state equation can be applied to the linear differential equation.

In a situation where the operating point of the controlled object changes every moment, at present, set point control is sometimes performed to determine the operating point, and the controlled object is linearly approximated in the vicinity of the operating point. The feedback coefficient is determined by using the modern control theory and the feedback control is implemented. For such set point control, for example, “Theory and Practice of Sheet Rolling (p283-29)
4) ”(published by The Iron and Steel Institute of Japan; s September 959).

[0005]

When the feedback coefficient (parameter) is suddenly changed by the set point control, the control command (the manipulated variable of the control effector) may suddenly change according to the change of the feedback coefficient. is there.
As a result, in a control system with a fast response, the state of the controlled object suddenly changes, which causes a situation where stability is lost.

For example, in the case of automatic welding by a robot, the arm expands and contracts according to the welding position, and the moment of inertia of the arm fluctuates greatly. That is, when the arm is extended, the moment of inertia is large, and when a position deviation appears at this time, a large force is required to restore it, so it is possible to design a large feedback coefficient to generate a large control command. . On the other hand, when the arm is contracted, the moment of inertia is small, so even a small force can move well.

Accordingly, when welding is performed while changing the state where the arm is extended to the state where the arm is extended, if a position deviation occurs in the state where the arm is contracted, a large control command based on a large feedback coefficient is given to the actuator. However, since the moment of inertia of the arm is small, it easily overshoots, and a large command is again generated to correct this overshooting amount in the next sampling cycle, and an oscillating motion is repeatedly caused.

As a result, the location where the welding should be performed linearly becomes a thick welding line width due to a fine sine wave vibration locus, resulting in deterioration of quality due to defective welding.

As described above, in a controlled object whose moment of inertia fluctuates, not only a large feedback coefficient, but also when the control target value suddenly changes due to setup, the same vibrational operation occurs, resulting in control accuracy and safety. There was a problem that deteriorated the sex.

An object of the present invention is to provide a control system for controlling feedback control executed at high speed with high accuracy and stability for a control object having a strong nonlinearity as described above.

[0011]

SUMMARY OF THE INVENTION The present invention realizes a control system capable of suppressing overcontrol even when there is a sudden change in control parameters or control commands.

That is, the control method of the present invention is a control method or a control device for determining a control command value while changing the control gain of the control model according to the state of the controlled object. The differential value of the state quantity or the difference value from the previous time is calculated, the differential value or difference value and the control gain at the control point are synergistically calculated to find the deviation of the control command value, and this deviation value is integrated or integrated. The control command value is determined by the above.

[0013]

According to the present invention, the control command value is obtained by temporarily differentiating the state quantity from the previous value and multiplying the differential value or the difference value with the control gain to obtain the deviation command value, which is integrated or the previous value. To obtain the control command value. This is because the differentiated (difference) is integrated (integrated) again, so there is no change in the control operation seen in the macro, but there is a large difference instantaneously. That is, since the change in the state quantity is the difference value from the previous value, the change quantity is small, which can reduce the influence of the change in the control gain and suppress the overcontrol, and thus the accuracy and stability of the control can be improved. It is possible to improve.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 shows an embodiment of an optimum regulator constructed in accordance with the present invention. The controlled object 1 is controlled by the feedback control mechanism 2 so as to maintain a constant state. The control model of the control mechanism 2 is identified by a linear differential equation, and each coefficient (control parameter) thereof is calculated by the control parameter correction mechanism 3. Correction is a predetermined cycle T (seconds)
According to the state of the controlled object detected every time, it is calculated by the well-known modern control theory such as the optimal regulator theory and the H∞ control theory. Regarding the optimum regulator, "Modern control engineering (Chapter 7, pages 141 to 158; Takeshi Tsuchiya, Masayoshi Egami; Sangyo Tosho Co., Ltd., April 26, 1991)"
And "System Control Theory (Chapter 6, pages 222-256; Masami Ito, published by Shokoidosha, May 20, 1973)". In this embodiment, the difference mechanism 4 and the integration mechanism 5 are added to the configuration of such an optimum regulator. Difference mechanism 4
Detects the state x (kT) of the controlled object 1 every sampling cycle T, and the state vector x ({k-1}) one cycle before is detected.
The difference state vector Δx (kT) which is the difference from T) is calculated by (Equation 1), and this Δx (kT) is output to the feedback control mechanism 2.

[0015]

[Equation 1]

The feedback control mechanism 2 calculates the manipulated variable of the controlled object 1 by the parameter-corrected control model. That is, according to (Equation 2), the input difference value Δ
x is multiplied by the modified parameter F to obtain the deviation value Δu of the control command value u. The integrating mechanism 5 calculates the control command value u by multiplying the deviation value Δu to the previous command value by (Equation 3), and outputs the control command value u to a control effector (not shown).

[0017]

[Equation 2]

[0018]

[Equation 3]

FIG. 2 shows an embodiment in which the optimum regulator of the present invention is applied to robot arm control. Normally, a robot performs position control, but here, in order to facilitate understanding of the essence of the invention, a mechanism is simplified and a speed control system of an electric motor for controlling a robot arm will be described. The original position control can be realized by introducing an integral system into the speed control system.

In FIG. 1, a robot mechanism 100 includes a load 101 corresponding to a robot arm whose arms extend and contract, a rotor 102 of a DC motor that drives the load 101 with a torque T _L , a commutator 103 that contacts the rotor, 104,
An amplifier 106 that supplies the armature current i _a to the commutator, a field circuit 107 that supplies the field current _if to the field winding, a current detector 109 that detects the armature current i _a, and rotation of the rotor 102. The rotation angular velocity detector 108 detects the angular velocity ω _r .

In the feedback control system of the present embodiment, first, the deviation i _a ′ between the armature current i _a and the armature current target value obtained by the set point control is obtained by the adder 110, and the rotational angular velocity ω _r is set. The deviation ω _r ′ from the rotational angular velocity target value, which is the point output, is obtained by the adder 111, and these are input to the difference mechanism 4. In the difference mechanism 4, each input value and i _a ((k−1) T) ′ and ω which are values one cycle before the input value
The difference with _r ((k-1) T) 'is obtained, and each difference value Δi
(kT) and Δω (kT) are output.

In the feedback control mechanism 2, the difference value Δi is calculated by the feedback coefficient reviewed every cycle T.
From (kT) and Δω (kT), the control deviation value Δu (kT) shown in (Equation 2) is calculated and output. The control parameter correction mechanism 3 uses the armature current i _a (kT) and the rotational angular velocity ω.
By inputting _r (kT), the correction coefficient F is corrected.

The control deviation value Δu is added to the integrated value up to then by the integration mechanism 5, and the new integrated value is added to the target value u by the adder 6.
_It is added to _ref to determine the control command value u. Command value u
Is input to the amplifier 106, and the armature current i _{a is changed} to the command value u.
Control to the value corresponding to.

With respect to the control system shown in FIG. 2, the neighborhood of the operating point within the range of linear approximation will be considered. That is, in the control system of FIG. 2, actual values are used for the armature current and the rotational angular velocity, but since the operating point can be determined at the set point, feedback control that makes the deviation from the vicinity of the operating point zero Should be considered. Focusing on this deviation, considering the outputs from the armature current detector 109 and the rotational angular velocity detector 108 in FIG. 2 as deviations from the target values, FIG. 3 is an equivalent modification of the control system in FIG. . Then, it is obvious that if the configuration of FIG. 3 is generalized, the configuration of FIG. 1 is obtained.

FIG. 4 shows the structure of the difference mechanism 4 shown in FIGS. The difference mechanism 4 includes a difference unit 120 (123) according to the number of inputs. This example has two inputs from the rotational angular velocity detector 108 and the armature current detector 109 of FIG. The difference unit 120 includes a dead time mechanism 121 and an adder 12
It is composed of 2 and subtracts the input (that is, the input one cycle before) passing through the dead time mechanism 121 from the input directly added to the adder 122, and outputs the difference value. The dead time of the dead time mechanism 121 is the sampling cycle T of the control system.
(Seconds).

As described above, the outputs of the rotational angular velocity detection mechanism 108 and the current detection mechanism 109 are the difference device 12 of the difference mechanism 4.
0, 123, and the state feedback mechanism 2 which feeds back the amount of change in the state quantity. In addition,
In this example, since the input is a discrete value of the period T, the difference mechanism is used, but when it is a continuous quantity, the differentiation mechanism is used.

FIG. 5 shows a model of the feedback mechanism 2. The rotational angular velocity difference value Δω and the armature current difference value Δi from the difference mechanism 7 are multiplied by the feedback coefficients f ₁₁ and f ₁₂ , respectively, and then added by the adder 132. Are added and output as a command deviation value Δu.

FIG. 6 shows the structure of the integrating mechanism 5. An adder corresponding to the input from the feedback mechanism 2 is provided, and input values discretized in the cycle T are integrated in the cycle T and output. In this example, since the number of inputs is 1, it is composed of one adder 125. The adder 125 is a dead time element 1 that outputs the input from the feedback mechanism 2 and the addition result up to that time.
The output of 27 is added by an adder 126, and the new addition result is used as a control command value (manipulation amount) u to make the amplifier 1 of the robot mechanism.
It outputs to 06. An integration mechanism is used for the mechanism 5 when the input is a continuous quantity.

Next, the differential equation used for designing the control system by linearly approximating the operation of the robot arm shown in FIG. 3 in the velocity system will be described.

First, the mechanical system of the armature and the robot arm directly connected to it can be expressed by (Equation 4) and (Equation 5) and the generated torque by (Equation 6) by Newton's equation of motion.

[0031]

[Equation 4]

[0032]

[Equation 5]

[0033]

[Equation 6]

However, K _T : torque constant, θ (t): rotation angle, ω _r (t): each rotation speed, i _a (t): armature current, J: moment of inertia, _if (t): Field current, B: damping coefficient, p: number of pole pairs, φ (t): field magnetic flux, M: mutual inductance between armature winding and field winding, T _g (t): generated torque, _TL ( t): Load torque. Here, the moment of inertia J has a change of 1 to 10 times in terms of the motor axis according to the position of the robot arm.

On the other hand, there are two electric systems, an armature circuit and a field circuit, which are represented by (Equation 7) and (Equation 8).
The back electromotive force can be expressed by (Equation 9).

[0036]

[Equation 7]

[0037]

[Equation 8]

[0038]

[Equation 9]

Where v _a (t): DC applied voltage, v
_f (t): field磁印applied voltage, R _a, L _a: total resistance and the total inductance of the armature winding circuit, R _f, L _f: total resistance and the total inductance of the field winding circuit, e _i (t ): It is a back electromotive force.

Further, by defining the state quantities x ₁ and x ₂ and the manipulated variable u of this control system as in (Equation 10), and rearranging the equations from (Equation 4) to (Equation 9), (Equation 11) ) Is obtained. Further, since the output (control amount) is the rotational angular velocity, the output equation of (Equation 12) is obtained.

[0041]

[Equation 10]

[0042]

[Equation 11]

[0043]

[Equation 12]

[0044]

[Table 1]

Here, the specifications shown in Table 1 are changed to the above (Equation 1
Substituting into 1) and (Equation 12), the differential equation of (Equation 13) is obtained. However, J = 4.89 × 10 ³ and the moment of inertia is the smallest.

Furthermore, (Equation 13) is set to the sampling period 2
When discretized with ms, the difference equation of (Equation 14) is obtained,
It can be seen that this corresponds to the general form of (Equation 3).

Next, when the robot extends its arm and the moment of inertia J is 4.98 × 10 to ² , the differential equation becomes (Equation 15). Similarly, when the sampling period is discretized at 2 ms, (Equation 16). Is obtained.

[0048]

[Equation 13]

[0049]

[Equation 14]

[0050]

[Equation 15]

[0051]

[Equation 16]

Comparing (Equation 14) and (Equation 16), the element in the first row and second column of the system matrix is 0.00 of (Equation 14).
From 0286 to 0.0000287 in (Equation 16), it becomes about 1/10, which is a change of 90%. Thus, when the moment of inertia is small (Equation 14) and when it is large (Equation 16), the coefficient of the state equation, and thus the parameter of the control model, changes greatly. This means that to make the operating state of (Equation 16) the same as the operating state of (Equation 14), x
Indicates that there is a need to increase ₂ (t) = i _a armature current is about 10 times.

Now, a load is applied to the robot arm,
Assuming the armature current has changed from 5A to 6A,
It is assumed that the robot arm is changed from the contracted state to the extended state. When the coefficients f ₁₁ and f ₁₂ of the feedback mechanism 2 of FIG. 6 at this time are solved by the well-known Riccati equation (detailed in the above “modern control engineering” etc.), the values shown in Table 2 are obtained.

[0054]

[Table 2]

Now, the optimum regulator of this embodiment (FIG. 3) will be compared with the conventional optimum regulator without the difference mechanism 4 and the integration mechanism 5. The feedback mechanism 2
There is no change in the basic configuration of.

When the control command value before and after the robot state changes from the above (Equation 14) to (Equation 16) and the feedback coefficient f ₁₂ is corrected (however, only for the component caused by the armature current), In the conventional case, the case of (Equation 14): u ₁₂ = −f ₁₂ × i = −0.0301
From × 5 = −0.151, in the case of (Expression 16): u ₁₂ = −f ₁₂ × i = −0.0096
A sudden change of about 70% is made to 1 × 6 = −0.0481. As a result, it will be apparent that the state deviation rapidly increases and the operation of the motor becomes unstable.

On the other hand, in the case of the present embodiment, the command value before the correction of the feedback coefficient is the same as in the prior art-0.151, but when the current changes from 5A to 6A, the output of the differential mechanism 123 becomes 1A. Therefore, the command deviation value is Δu ₁₂ = −f
₁₂ × Δi = −0.00961 × 1 = −0.00961 and is added to the previous command value 0.151 in the integration mechanism 5. As a result, the control command value u ₁₂ = −0.141. In this case, the change in the command value is as small as about 7%, and the operation of the electric motor is kept stable.

As described above, according to the optimum regulator of this embodiment, since the command value does not change suddenly, the control system becomes stable and the deviation falls within the allowable range. Further, since the integrating mechanism 5 is an integrating operation, the deviation is steadily eliminated and the optimum regulator is satisfied.

Next, an optimum servo control system according to the second embodiment of the present invention will be described. FIG. 7 shows a conventional optimum servo control mechanism. The first feedback control system that improves the responsiveness by the control mechanism 2 that proportionally controls the instantaneously changing state amount, and the control amount by the control mechanism 6 that controls the steady-state deviation between the target value and the control amount to zero. The second feedback control system that keeps constant, and the outputs of both systems are added to manipulate the manipulated variable u.
And a parameter correction mechanism 3 for correcting the control parameters of both systems during setpoint control.

As described above, the optimum servo is provided with the proportional control means for increasing the response and the integral control for making the steady deviation zero. The theory of optimum servo is described in "Modern Control Engineering" (Chapter 8, p159-p16).
2) ”, it is easily derived from the optimal regulator theory.

FIG. 8 shows an optimum servo control mechanism according to this embodiment. A differentiation mechanism 4'and an integration mechanism 5'are added to the first feedback control system so that the state quantity x from the differentiation mechanism 4'can be reduced. After inputting the variation (dx / dt) to the feedback mechanism 2, the output is integrated (∫dx · dt) to be a part of the manipulated variable. This is the same as the basic configuration of the optimum regulator according to the first embodiment shown in FIG.

The optimum servo of the present embodiment is such that in each of the first feedback control system and the second feedback control system,
The feature is that the integrating mechanism serving as the output mechanism is an integrating mechanism 5'combined as shown in FIG.

The differentiating mechanism 4'can be replaced by the difference mechanism 4 shown in FIG. 4, and the integrating mechanism 5'can be replaced by the integrating (adding) mechanism 5 shown in FIG. The differentiating mechanism and the difference mechanism, and the integrating mechanism and the integrating mechanism have the same configuration,
Only the gain is different. Therefore, in this embodiment, the differentiating mechanism and the difference mechanism, and the integrating mechanism and the integrating mechanism are synonymous.

This embodiment is constructed as described above,
When the operating point of the controlled object 1 changes, the control parameters of the feedback control mechanism 2 and the feedback control mechanism 6
The control parameter of is calculated and corrected by the control parameter correction mechanism 3.

The first feedback system feeds back the deviation value obtained by multiplying the deviation from the previous state quantity by the feedback coefficient, and adds the deviation command value added to the deviation value of the second feedback system to the previous command value. Then, the current control command value u is obtained. As a result, it is possible to avoid a sudden change of the control command value u due to the change of the control parameter, and it is possible to realize a stable optimum servo even for a control target having a strong nonlinearity.

FIG. 17 is an example in which the optimum servo according to the second embodiment described above is applied to the robot control mechanism of FIG. 3, and is a block diagram for performing speed control and position control of the robot.

In the figure, the difference from the optimum regulator of FIG. 3 is that a position detector 108 'consisting of an absolute encoder for detecting the rotation angle position of the robot arm,
A position command generating mechanism 7 for indicating an arm position, an adder 8 for obtaining an error with respect to a target position from both outputs,
It is provided with a second feedback system including a second feedback control mechanism 6 and an integration mechanism 5 for obtaining an error control command component for making the error from the adder 8 zero. This second feedback system is a position control control system.

The output of the second feedback control mechanism 6 is added by the adding mechanism 9 with the state command deviation that makes the difference of the state quantity from the first feedback mechanism 2 zero, and then,
The integrating mechanism 5 integrates the previous value. The error control command component from the feedback mechanism 6 becomes a control command for making the error (position) zero by this integration, and is output until the error becomes zero.

On the other hand, the first feedback system (speed control system) consisting of the differential mechanism 4 and the first feedback control mechanism 2 has the same structure as the optimum regulator of FIG. 3, and it depends on the large fluctuation of the feedback coefficient and the state quantity. Instability factors are reduced.

The integration mechanism 5 of this embodiment is shared by the first feedback system and the second feedback system, and the configuration is simplified. Of course, they may be provided separately.

According to this embodiment, the speed control loop (optimal regulator configuration) of the robot according to the first embodiment is
It is configured as an optimum servo system with an added position control loop. Therefore, it not only shares the responsiveness of the optimum regulator and the good control accuracy of the optimum servo, but the control command does not change suddenly even if the feedback coefficient is changed according to the moment of inertia, so high response and high accuracy are achieved. Highly stable control can be realized.

Next, the optimum servo of the second embodiment is
An example applied to the control of the sizing press will be described. Figure 9
~ Fig. 11 illustrates the operating principle of the sizing press.
The sizing press uses the thick plate slab 30 as an anvil 31,
It is compressed by 32 and molded into a predetermined size, and is used in the pre-process of the rolling mill.

In FIG. 9, the slab 30 entered in the direction of the arrow.
Is narrowed by being compressed in the width direction from the anvils 31 and 32. At this time, the anvils 31 and 32 are interlocked with the progress of the slab and, as shown in FIG. 10, swing from the state of (a) to (b) → (c) → (d) in the forward / backward direction and left / right direction of the slab. To do. (A) and (d) of the same figure show the same state, and the slab diagonal line part 30-1 of (c) shows the state crushed from the (a) state.

FIG. 11 shows a drive mechanism for the anvil 32 (31). Screw 33 on anvil 32
The structure 34 attached through the rocks together with the anvil. Anvil 32 (31) in structure 34
A main crank mechanism composed of a main connecting rod 36 and a main crank 37, which are arranged in the slab width direction, is attached to the main crank 37, and a shaft 38 of the main crank 37 is rotated by an electric motor 39.

On the other hand, in the structural member 34, the swing main crank 41 and the swing connecting rod 42, which interlock with the main crank 37, give the anvil 32 (31) an elliptical motion in the front-rear, left-right directions.
A swing crank mechanism including a swing sub crank 43 is attached. The shaft 44 of the swing main crank is rotated by the swing motor 45.

Due to the combined operation of the main crank mechanism and the swing crank mechanism, the anvil 32 (31) operates in the elliptical locus shown in FIG. 11 and executes one cycle of FIGS. 10 (a) to 10 (d). In this locus, the thick line portion from a to b is the slab pressurization period.

FIG. 12 shows the system configuration including the swinging part of the sizing press and the control device. The electric motor 39 is controlled by the control device 50, and its angle ρ, angular velocity ω ₁ ,
The angular acceleration ω ₁ 'is operated by a pulse generator (PG) 4
Detected by 0. The swing motor 45 is controlled by the control device 51, and the swing motor 45 has an angle θ, an angular velocity ω,
The angular acceleration ω ′ is detected by the PG 46.

The command generator 52 generates the target values ρ _ref and θ _ref of the main crank angle ρ and the swing sub crank angle θ in accordance with the functional relationship that gives the anvils 31 and 32 the above-mentioned elliptical locus, and at every control cycle. To the control devices 50 and 51.

FIG. 13 shows a physical model of the main crank mechanism. The connecting rod 36 having a length of l ₀ makes a movement of the crank angle ρ by the rotation of the main crank 37, so that the structure 34
(Therefore, the anvils 31, 32) are moved back and forth in the slab width direction according to the change in the distance y. This physical model is
It is shown in formulas (17-1) to (17-5) of 7).

[0080]

[Equation 17]

FIG. 14 shows a physical model of the swing crank mechanism. The distance x between the main swing crank 41 and the sub swing crank 43, which is parallel to the slab advancing direction, is determined by the swing (equation 1).
It changes like 8). Here, the radius of the main swing crank 41 is r ₁ , the main swing crank angle is θ, the radius of the sub swing crank 42 is r ₂ , and the sub swing crank angle is (2θ−90).
°), the length 1 of the swing connecting rod 42, and the inclination φ.

FIG. 15 shows the swing relationship of the entire main crank mechanism and swing crank mechanism, and the physical model is shown in equations (19-1) to (19-5) of (Equation 19).

[0083]

[Equation 18]

[0084]

[Formula 19]

In the above-mentioned physical model of the entire swing relation, the equation (17-4) representing the load torque T _M applied to the electric motor 39 of the main crank mechanism and the electric motor 45 of the swing crank mechanism.
(19-3) expressing the torque T _S of
The magnitudes of the values of the term and the second term are almost the same.

The second term in the above equation indicating the equivalent GD ² applied to each electric motor greatly varies depending on each crank angle ρ or θ. In one cycle, ρ is 0 to tan to ¹ (r ₁ +
r ₂ ) / x, θ changes from 0 to 2π, and the control command is output hundreds to thousands of times during this period.

As described above, since the equivalent GD ² applied to each electric motor varies greatly depending on the position of the sizing press, a fixed control model, that is, an invariable feedback coefficient causes an extreme decrease or uncontrollability of control accuracy. .

Next, rearranging (Equation 17) to (Equation 19),
The differential equations shown in Expressions (20-1) to (20-4) of (Equation 20) are derived. Formula for differentiation of angular acceleration (20-
In 4), the coefficient Js is the moment of inertia multiplied by the rotational angular velocity x ₃ of the swing motor 45, Tc is the temporary delay time constant of the motor 45, and the others are constants.

When the differential equation of (Equation 20) is displayed in a matrix, the state equation of (Equation 21) is obtained. Note that (Equation 2
The matrix and variable vector of each term in 1) are shown in (Equation 22).

[0090]

[Equation 20]

[0091]

[Equation 21]

[0092]

[Equation 22]

In the equation of state of (Equation 21), Ts of the third term
Is the disturbance vector. The swing torque Ts is (19-
It is a function of Vx in the equation (1), and its value greatly varies sinusoidally as described above. The control parameter correction mechanism 3 corrects the feedback coefficient according to the position (angle) of the sizing press using the control model at the operating point corresponding to the torque fluctuation.

For this reason, the control device for the sizing press needs to have a structure for suppressing a sudden change in the control command due to the correction of the feedback coefficient, as in the case of the robot control described above.

FIG. 16 shows a control device 51 of the swing electric motor 45 which drives the swing crank mechanism. This control device 51
Is a DDC control device configured by an optimum servo system and configured to reduce the influence of fluctuations in the swing torque Ts.

PG directly connected to the rotary shaft of the swing motor 45
From 46, the state quantities of the main swing crank angle θ, the angular velocity ω, and the angular acceleration ω ′ are sampled in a predetermined cycle. In addition,
PG 46 is a digital pulse generator, which is composed of a pulse counter and a differential calculator, and outputs the state quantities x _i of θ, ω, and ω ′. The state of the kth step to be controlled is represented by (Equation 23).

[0097]

[Equation 23]

In this embodiment, since the deviation value system is used, the state quantity x _i is input to the adder 201 and is set as the setup value x _{s of the} state quantity given from the command generator 52 or the upper setup computer (not shown). Deviation value Δx _i is obtained.

This ΔX _i (k) is the difference mechanism 202-
1 to 202-3, and the previous sampling (k-
With respect to the state deviation value Δx _i (k−1) in 1), the state difference value ∇x _i (k) is obtained by (Equation 24) and (Equation 25). The state difference value ▽ x _i (k) in this example is ▽
θ (k), ▽ ω (k), and ▽ ω '(k).

[0100]

[Equation 24]

[0101]

[Equation 25]

Each state difference value ∇x _i (kT) is input to the state feedback mechanism 203 and is multiplied by each feedback coefficient (proportional gain) Fx _i . The feedback coefficient Fx _i is appropriately corrected by the control parameter correction mechanism 3 according to the state of the controlled object. Fx
The modification of _i is given by (Equation 26) when the well-known Riccati equation is solved by considering the expansion matrix of (Equation 28).
This solution is detailed in "Modern Control Theory" by Takeshi Tsuchiya mentioned above, for example.

[0103]

[Equation 26]

The state feedback mechanism 203 multiplies the state difference value ∇x _i (kT) by the feedback coefficient Fx _i, and obtains the manipulated variable deviation value ΔU ₁ as an angle command component from (Equation 27) and (Equation 28). . ΔU ₁ according to the state difference
Is output to the adder 204, and the above constitutes a first feedback control system for state feedback.

[0105]

[Equation 27]

[0106]

[Equation 28]

On the other hand, the swing crank angle θ, which is also the control amount y in this example, and the target value θ _ref given from the command generator 52.
Is subtracted by the adder 205, and the error ε is (Equation 2
9).

[0108]

[Equation 29]

This error ε is determined by the error feedback mechanism 2
At 06, a feedback coefficient (proportional gain) Fe = fe ₁₁
And the manipulated variable deviation value ΔU ₂ is calculated by (Equation 30). ΔU ₂ corresponding to the error is output to the adder 204. The above constitutes the second feedback control system for error feedback.

[0110]

[Equation 30]

[0111] The manipulated variable deviation △ U ₁ of the first feedback control system, the manipulated variable deviation value by the second feedback control system △ U ₂ is the addition is made the amount of operation of the adder 204 (the number 31) The deviation value ΔU is calculated, and ∫d
The operation amount U is determined by performing the calculation of (Equation 32) in the integration mechanism 207 having the calculation function of t.

[0112]

[Equation 31]

[0113]

[Equation 32]

The integrating mechanism 207, like the integrating mechanism 5 of FIG. 6, integrates the current manipulated variable deviation value ΔU (k) with the previous manipulated variable U (k−1) to obtain the current manipulated variable U ( k) is calculated.
The operation amount u thus obtained is the main swing crank 41
Control command crank angle θp.

In order to control the crank angle θ of the main swing crank 41 to this θp, the electric motor current I for speed controlling the electric motor 45 to the rotational angular velocity ωp which is the differential value of θp is determined. Basically, it is the same as the example of FIG.

Similarly, a control command ρp for controlling the crank angle ρ of the main crank 37 is obtained and given to the control device 50, but details thereof will be omitted.

As described above, the sizing press control apparatus of this embodiment corrects the feedback coefficient according to the fluctuation of the swing angle in order to cope with the equivalent GD ² that greatly changes in the cycle. The optimum servo is composed of a state feedback system having a difference mechanism for suppressing the sudden change of the command and an error feedback system for making the position error zero. As a result, unstable operation of the sizing press due to a sudden change in motor torque can be prevented, and it is not necessary to keep the control gain low in advance.

The present invention is not limited to the robot and the sizing press described above. For example, it is needless to say that the present invention can be applied to control in a wide range of fields such as trains and automobiles whose inertial performance rate varies greatly, and rolling systems.

[0119]

According to the present invention, it is possible to suppress over-control due to a sudden change of a control command due to a change of a control parameter in a control object having a strong non-linearity in which a moment of inertia or the like changes greatly, so that highly accurate control can be achieved. There is an effect that can be carried out stably.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an optimum regulator according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of a speed control system of a robot showing an application example of FIG.

FIG. 3 is a block diagram of the speed control system of FIG. 2 based on linear approximation.

FIG. 4 is a configuration diagram of a difference mechanism.

FIG. 5 is a configuration diagram of a feedback control mechanism.

FIG. 6 is a configuration diagram of an integration mechanism.

FIG. 7 is a configuration diagram of a conventional optimum servo control mechanism.

FIG. 8 is a configuration diagram of an optimum servo control mechanism according to a second embodiment of the present invention.

FIG. 9 is an explanatory diagram showing the principle of a sizing press to which the second embodiment is applied.

FIG. 10 is an explanatory diagram illustrating an operation cycle of the sizing press.

FIG. 11 is an explanatory diagram illustrating a drive mechanism of the sizing press.

FIG. 12 is a schematic system configuration diagram illustrating a control system of a sizing press.

FIG. 13 is an explanatory diagram of a physical model of a main crank mechanism of the sizing press.

FIG. 14 is an explanatory diagram of a physical model of the swing crank mechanism of the sizing press.

FIG. 15 is an explanatory diagram of a physical model of the swing relation of the entire sizing press.

FIG. 16 is a functional block diagram showing a configuration of a control device of the sizing press.

FIG. 17 is a configuration diagram of a robot control mechanism that performs speed control and position control.

[Explanation of symbols]

1 ... Control object, 2 ... First feedback mechanism, 3 ... Control parameter correction mechanism, 4 ... Difference mechanism (differential mechanism), 5
... integration mechanism (integration mechanism), 6 ... second feedback mechanism, 7 ... command generation mechanism, 52 ... command generation device, 201 ...
Addition mechanism, 202 ... Difference mechanism, 203 ... State feedback mechanism, 204, 205 ... Addition mechanism, 206 ... Error feedback mechanism, 207 ... Integration mechanism.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yutaka Saito 5-2-1 Omika-cho, Hitachi-shi, Ibaraki Hitachi Ltd. Omika factory

Claims (10)

[Claims] 1. A control method of a state feedback control system in which a feedback coefficient is changed during operation, wherein control is performed so as to suppress a change in a feedback amount when the feedback coefficient is changed. 2. A control method of a state feedback control system in which a control parameter including a feedback coefficient is changed during operation, a command deviation value obtained by differentiating a predetermined state quantity and multiplying this differential value by the feedback coefficient. A control method characterized by integrating to obtain a control command value. 3. The differentiation is performed by a difference between a previous value and a current value of the predetermined state quantity when the predetermined state quantity is a discrete value which is periodically detected. The described control method. 4. The integration is performed by adding a previous control command value and the command deviation value when the predetermined state quantity is a discrete value detected periodically.
The described control method. 5. A control device of a state feedback control system configured such that a control parameter including a feedback coefficient can be changed during operation, wherein the state feedback control system includes a previous value of a predetermined state quantity detected periodically. And a feedback mechanism for obtaining a command deviation value by multiplying the difference value and the feedback coefficient, and an integration mechanism for adding the command deviation value and the previous control command value. Characteristic control device. 6. A control device of a state feedback control system, wherein a control parameter including a feedback coefficient is configured to be changeable during operation, wherein the state feedback control system comprises a preset set value of a predetermined state quantity and a cycle. The addition mechanism for obtaining the deviation of the detected value of the predetermined state quantity that is detected automatically, the difference mechanism for obtaining the difference between the previous value and the current value of the state deviation value from this addition mechanism, and the state from this difference mechanism. A control device comprising: a feedback mechanism for multiplying a difference value by the feedback coefficient to obtain a command deviation value; and an integration mechanism for adding a command deviation value from the feedback mechanism and a previous control command value. 7. The control device according to claim 6, wherein the state feedback control system is configured as an optimum regulator that determines a current control command value so as to zero the state deviation value. 8. A first feedback control system for controlling the state quantity to be constant, a second feedback control system for controlling the control quantity to approach a target value, and feedback coefficients of these feedback control systems (including control parameters In a control device that configures an optimum servo by a control parameter correction mechanism that corrects and a target value generator that sets the target value according to the state of the controlled object, the first feedback control system periodically It has a difference mechanism for obtaining a difference value between the previous value and the current value of the detected predetermined state quantity, and a first feedback control mechanism for obtaining a first command deviation value by multiplying the difference value and the first feedback coefficient. The second feedback control system adds the target value set by the target value generator and the deviation value of the control amount detected periodically. And a second feedback control mechanism for obtaining a second command deviation value by multiplying the deviation value of the controlled variable by a second feedback coefficient, the first command deviation value and the second command deviation. A control device comprising a control command value calculating means for adding and integrating values and outputting a control command value. 9. A first feedback control system for controlling the state quantity to be constant, a second feedback control system for controlling the control quantity to approach a target value, and a control parameter including feedback coefficients of these feedback control systems. In a control device that configures an optimum servo by a control parameter correction mechanism that corrects and a set value generator that sets a target value of the control amount and a set value of the state amount, the first feedback control system is A first addition mechanism that obtains a deviation value between a predetermined state amount detected by the target value and a target value of the state amount set from the target value generator;
A difference mechanism for obtaining the difference between the previous value and the current value of the state quantity deviation value from this addition mechanism, and the state quantity deviation value to zero by multiplying the state quantity difference value from this difference mechanism and the first feedback coefficient. To determine the first command deviation value so that
The second feedback control system, wherein the second feedback control system obtains a deviation value between the target value of the control amount set by the set value generator and the control amount periodically detected. And a second feedback control mechanism for obtaining a second command deviation value so as to make the control quantity deviation value zero by multiplying the control quantity deviation value from the adding mechanism by a second feedback coefficient, A control device comprising a control command value calculating means for adding a first command deviation value and the second command deviation value, integrating the result, and outputting a control command value. 10. A sensor that detects a state quantity of a controlled object, a control effector that operates according to a control command to control the controlled object to a desired state, and a control parameter is changed and a control command is determined according to the state of the controlled object. In a control device of a state feedback control system including a computer device, the computer device, a difference value calculation means for obtaining a difference value from a predetermined state amount periodically detected by the sensor and its previous value, and the difference value. It includes a command deviation calculating means for multiplying a feedback coefficient to obtain a command deviation value, an integrating means for adding the command deviation value and a previous control command value to obtain a control command value to be output to the control effector, and the feedback coefficient. A status parameter characterized by comprising control parameter modification means for modifying the control parameter based on a change in the control model operating point. Control device for feedback control system.