JP2637302B2 - Control method and device - 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 method and an apparatus therefor, and more particularly to a control method and an apparatus suitable for controlling the same control object by using a plurality of different control methods.

[0002]

2. Description of the Related Art Japanese Patent Application Laid-Open Publication No. HEI 1-1 discloses a prior art which employs a plurality of different control methods to control one control target.
There is one described in 77604. According to this conventional technique, when the deviation between the target value and the control amount is large, control is performed using a control device having excellent responsiveness, and when the deviation is small, the control device is switched, and the control device has excellent stability. Control is performed. This prior art uses two different controllers, both of which belong to linear control.

[0003] As control techniques in recent years, Fuzzy control and neural network control suitable for non-linear control have been spotlighted. This non-linear control device is good at imitating the control operation of a skilled person by learning, but it is still established for situations where even a skilled person rarely encounters it because the amount of data to learn the state is insufficient. It cannot be said that it is the technology that was done. Therefore, it is used in combination with the established linear control, which is a conventional technique. For example, Japanese Patent Laid-Open No. 1-2
In the prior arts described in Japanese Patent No. 50103 and Japanese Patent Application Laid-Open No. Hei 2-8903, a Fuzzy control device and a linear control device are used in combination, and when one control target is stable, PID control, that is, linear control is performed. Fuzz when bad
y control is being performed.

[0004]

For example, when the rotational speed of a motor under load is controlled to a target value, the conventional technology described in Japanese Patent Application Laid-Open No. 1-177604 is applied, and the actual rotational speed and the target value are controlled. When the deviation is large, a control device with excellent responsiveness is used, and when the deviation is small, the control device is switched to a control device with excellent stability, thereby performing control that satisfies both responsiveness and stability. be able to. However, there is no guarantee that no overshoot will occur when the control device is switched. Further, in the above example, since the state quantity of the control target is only one “rotational speed deviation”, it is easy to configure a control device corresponding thereto. However, in a large-scale plant such as a chemical plant, a steel plant, or a power plant, there are many state quantities to be controlled, and controlling all of these state quantities with a linear controller complicates the setting of the controller. . Further, in a plant or the like, there are many non-linear controls in which a plant state quantity rapidly approaches a target value and is stably controlled to the target value when approaching the target value, such as at the time of starting the plant. Therefore, it is desirable to introduce Fuzzy control or neural network control that exerts its power in nonlinear control.

[0005] In the prior arts described in JP-A-1-250103 and JP-A-2-8903 to which Fuzzy control is introduced, the Fuzzy control device is used as a backup in case of emergency. However, in this case, the characteristic of the fuzzy control that exerts its power by the non-linear control such as at the time of startup is not used. In addition, it is difficult to collect a data amount for defining a membership function when an abnormality occurs in the plant and it is not known what will happen, and Fuzzy control for such a state is hardly a well-established technique. More importantly, it does not guarantee a transient response, ie, an amount of overshoot or undershoot, that falls within a predetermined allowable range when switching to backup.

An object of the present invention also emphasizes taking advantage moreover safety characteristics of the non-linear control apparatus such as a Fuzzy and neural network, always to provide a control method and equipment can guarantee transients within the allowable range It is in.

[0007]

The above object is achieved by a non-linear control.
To control the controlled object using the control means and the linear control means
In the state space using the state quantities of the closed loop system as coordinate axes,
Switching control from non-linear control means to linear control means
Area is set in advance so that the transient response when
In addition, the control which is controlled by the non-linear control means
It is foreseen that the state quantity of the control object goes out of the convex area.
Is switched from the non-linear control means to the linear control means.
This can be achieved by controlling the control object instead .

[0008]

[Function] Even if it is a nonlinear control means, the membership related
Able to collect enough data to determine numbers etc.
Control within the area is controlled with the same reliability as linear control means.
Control and suitable for the characteristics of nonlinear control means
ing. The control becomes unstable, and the state quantity of the
When out of range, the established linear
It is preferable to control by the control means from the viewpoint of safety. Only
At this point, the point at which control is switched
Is a problem. Switching from nonlinear control to linear control
In this case, the state quantity of the control
The initial value becomes the initial value.
That is, it is related to control performance. Therefore, in the present invention,
This criterion for switching from shape control to linear control is
A convex region where the transient response is within the allowable range is determined in advance,
When it is predicted to go out of this convex area,
By switching the configuration, control performance is guaranteed and control
To ensure safety when switching.

[0009]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of a control device according to one embodiment of the present invention. The control device according to this embodiment is a Fuzzy
Controller 11 (illustrated as a controller 1) such as a controller and a neural network, a linear controller (illustrated as a controller 2) 12 for performing conventional PID control, and a contribution rate determining means 13
And the outputs of both controllers 11 and 12 are
3 to a plant 14 to be controlled. The controllers 11 and 12 take in plant state quantities including the output (control amount) of the plant 14 and generate control outputs for controlling the plant 14 independently.
On the other hand, the contribution ratio determination means 13 monitors both the internal state quantities of the controllers 11 and 12 and the state quantity of the plant 14 while contributing the control outputs from the controllers 11 and 12 at an appropriate ratio. , Generate a control output value. This control output value is the operation amount actually input to the plant 14.

FIG. 2 is a block diagram of one embodiment of the contribution ratio determining means 13 described above. In the present embodiment, each controller 11,
12 is 0% or 100% by the switch 21
Is determined. That is, the control during the normal operation of the plant 14 is performed with the output of only the nonlinear controller 11, and as described in detail later, when the plant state quantity is going to exceed a certain range, only the linear controller 12 is controlled. The output 14 controls the plant 14. It should be noted that, instead of the contribution ratio of 0% or 100%, an intermediate contribution ratio may be switched stepwise. That is, if a large change in the control input is disliked, the configuration may be such that the control is switched stepwise from 0% to 100% as soon as possible. In any case, the timing of this switching is determined by the switching determining means 2
This means 22 monitors and determines each state quantity as described above. The basis of the switching decision made here and its principle will be described in detail below.

The basis of the switching decision is a transient response formed by a closed loop between the linear controller 12 and the plant 14 which becomes effective immediately after the switching. Therefore, the previous Fuz
The switching is executed regardless of what kind of control the nonlinear controller 11 such as zy is executing.
As a result, the linear controller 12 functions as an excellent backup means. The problem here is that
It is determined whether the transient response of the loop formed by the linear controller 12 and the plant 14 falls within a predetermined range of overshoot and undershoot. Generally, the transient response of a linear system depends on the initial value of the state quantity at the start of control. In this regard, the following is a knowledge related to the present invention.

In the following, the linear controller 12 and the plant 14
Are represented as x1, x2,..., And simply represented as X, these are assumed to be a column vector that is put together, and the space created by this vector is called a state space. The output of the linear controller 12, that is, the control input to the plant 14 after the switching is represented by u1, u2,... Or collectively represented by a column vector U.

The differential equation relating to the state quantity X of the plant, that is, the state equation is:

[0014]

(Equation 1)

Where A and B are system matrices characterizing the dynamic characteristics of the plant 14. On the other hand, the linear controller 12 generates an output U according to the following equation (2).

[0016]

(Equation 2)

Here, F is a feedback gain matrix. From these two formulas 1 and 2, the following formula 3 is obtained.

[0018]

(Equation 3)

Here, Ac is a system matrix corresponding to a closed loop system. When this equation 3 is solved under an initial value X (0), a transient response X expressed by the following equation 4 is obtained as a solution.
(T) is obtained.

[0020]

(Equation 4)

As shown in equation (4), the transient response of the state quantity X depends on its initial value X (0), and even if the closed-loop stability of the linear control system is guaranteed by the value of the matrix Ac. However, depending on the initial value, a large overshoot or undershoot may occur. Here, the initial value is a value of the state quantity at the moment when the control is switched from the nonlinear controller 11 to the linear controller 12. Therefore, in order to ensure that the transient response falls within the previously permitted range, attention must be paid to at least the condition that the state quantity at the time of switching must satisfy. If, when Yuku examine all state quantity conceivable to exhaustive, usually than a status quantity amounting to several tens of several, each combination is a huge amount. That is, it is necessary to check the transient response using Equation 4 with each point on the state space of several tens of dimensions as an initial value, which is extremely difficult to implement.

However, by utilizing the fact that the initial response of the transient response has a convexity, it is possible to determine the condition that the initial value must satisfy by a very simple procedure. This convexity is shown as follows. First, a point Xα on the state space
(0) as an initial value, a transient response Xα (t) (formula 5 below)
And Xβ (0) as the initial value.
Assume that β (t) (shown in Equation 6 below) falls within the allowed range.

[0023]

(Equation 5)

[0024]

(Equation 6)

Next, an intermediate point Xγ on a straight line connecting both initial values is obtained.
Consider the nature of the transient response Xγ (t) with (0) as the initial value. The midpoint Xγ (0) is defined by the following equation (7).

[0026]

(Equation 7)

From the above equation 7, as shown in the following equation 8,
The transient response Xγ (t) is also the two transient responses Xα (t) and Xβ
It can be seen that it has an intermediate property of (t).

[0028]

(Equation 8)

That is, neither the overshoot nor the undershoot exceeds the former two (Xα (t), Xβ (t)) and always falls within the allowable range. From this, it is immediately derived that all points on a straight line connecting two points that can be accepted as initial values can be accepted as initial values, and furthermore, such a set of acceptable points is defined as a convex region in the state space. It is obvious that it will. Assuming that the state quantities are X1 to X3, an image shown in FIG. 3 is obtained.

Actually, it is easy to implement the configuration as shown in FIG. That is, in FIG. 4, the convex region is represented by a polyhedron,
Points P1, P2, determined along coordinate axes on the state space
Are connected to each other to form a polyhedron. In order to obtain these points, the transient response is confirmed by simulation or the like. However, since it is sufficient to obtain at least twice as many points as the coordinate axes, it is sufficiently practical.

As a more specific example, a two-dimensional (two state quantities) plant represented by the following equation 9 will be described as an example. There are two types of state quantities, x1 and x2, and x2
The response of (t) is unrestricted, but the response of x1 (t) is assumed to have an allowable range of -1 to +1 on the basis of zero.

[0032]

(Equation 9)

The linear controller 12 performs feedback as in the following equation (10).

[0034]

(Equation 10)

The gain [100, 20] is previously selected such that the eigenvalue of the closed loop is a double root of -10.0. Therefore, although the closed-loop system is sufficiently stable, the transient response differs depending on the initial value. FIG.
A state space with coordinate axes 1, 1 and 2 is drawn, and the results of calculating the transient response of x1 (t) with some points on it as initial values are also shown.

As shown in FIG. 5, first, two points (1, 0) and (-1, 0) are allowed on the x1 coordinate axis.
1 to +1). On the other hand, on the x2 coordinate axis,
(0,28) and (0, -28) are acceptable limits.
When these four points are connected to form a quadrilateral, the perimeter and inside are all allowed. As an example, the result of newly selecting four points on the circumference and calculating the transient response is also shown in FIG.

The method described above may have some further variations. First, using a circle or an ellipsoid included in a polyhedron as a convex region in the state space is an effective device for speeding up the calculation process. That is, in the case of a convex polyhedron region, in order to determine whether or not a certain point is inside, a first-order inequality equal to the number of faces must be examined. All you have to do is check the inequalities, saving you the hassle of doing calculations.

When the sample value is controlled by the digital computer, the state quantity at the sampling point in the next one step is foreseen by a state equation or the like, and switching judgment is performed using the value of the foreseeable state quantity. It is effective to do.
Further, in addition to the above-described embodiment in which the determination of switching is automatically performed, the position of the state quantity in the state space is displayed to the operator, and when the operator is about to exceed the allowable area (when approaching the boundary, Embodiments such as generating an alarm are also conceivable.
Such a display and an alarm are also effective in a plant control panel using only a conventional linear controller, and the configuration of such an embodiment is shown in FIG.

In FIG. 6, the state quantity monitoring means 62 monitors the state quantities of the plant 14 and the controller 61, plots the values of the state quantities in the state space, and displays them on the display device 62. When approaching the boundary of the above-mentioned allowable area, the operator is notified using the alarm generator 63. In this embodiment, the controller 61 is a linear controller, and the allowable range is set in accordance with the controller 61. In the case of some kind of large disturbance, the state quantity moves large instantaneously,
If control is continued as it is, it is effective as a method for generating an alarm when overshoot or undershoot is expected to exceed an allowable range.

FIG. 7 shows still another embodiment applied to water supply control of a nuclear reactor. In FIG. 7, the plant 7
Reference numeral 1 denotes the dynamic characteristics of the water supply system including the pump. It contains three state variables (x1: water level deviation, x2: flow deviation, x3: flow deviation change). On the other hand, the control device includes a fuzzy controller 72 and a linear controller 73
These are switched and used by the state quantity monitoring and switching means 74. The switching means 74 includes a display means for clearly explaining the basis of the switching and the state of the state to the operator.
75 .

The linear controller 73 is a so-called PI controller, and has a proportional gain Kp and an integral gain Ki.
For this purpose, a state quantity x4: water level deviation integrated value is included therein. The state equation of the plant is expressed by the following equation 11, and the linear feedback rule is expressed by the following equation 12.

[0042]

[Equation 11]

[0043]

(Equation 12)

Equations 11 and 12 correspond to Equations 1 and 2, respectively. As in the method described above, x1 to x
A hexahedral region can be created in a state space with 4 as a coordinate axis. If the state of the plant is inside this area,
Neural network control and Fu that have learned the operation of the expert
Non-linear control is performed using zzy control.

In the case where a linear controller is used as an emergency backup means as in this embodiment, a normal Fuz
It is an essential requirement to have a so-called transparency to clearly inform the operator of the basis for switching from the zy controller. An embodiment of the display screen of the display means 75 for that purpose is shown in FIG. The polyhedron in the four-dimensional space is divided into two two-dimensional coordinate spaces and displayed. The combination of coordinate axes for this division is the same no matter how it is determined, but as an example,
x1, x2 and x3, x4 are respectively combined.
The bright spot displayed on each two-dimensional space indicates the value of the current state quantity, and the rectangular frame indicates a plane obtained by cutting the aforementioned polyhedron on a plane including the current state quantity. I have. That is, the quadrilaterals P1 to P4 are obtained by cutting a polyhedron on a plane parallel to the x1 and x2 axes such that x3 = (x3 value of current state quantity) and x4 = (x4 value of current state quantity). . Therefore, the size of this quadrilateral changes according to the value of the current state quantity. Similarly, the results of cutting the polyhedron on the planes parallel to the x3 and x4 axes are shown for the quadrilaterals P5 to P7. If the bright spot is inside the quadrilateral, the Fuzzy control can be used with confidence, and the operator can also confirm it.

As another display example, there is a three-dimensional graph shown in FIG. This is effective when the control law is determined on the assumption that the state quantity x4 always starts from zero immediately after switching to the linear control. in this case,
Since it is sufficient to consider a three-dimensional polyhedron obtained by cutting the above-mentioned four-dimensional polyhedron at a plane of x4 = 0, it is easy to visually display the three-dimensional polyhedron. If the internal state quantity of the plant is larger than that in the present embodiment, it is also effective to divide the display into a plurality of three-dimensional graphs and display them.

The present invention can be applied not only to the above-described water supply control system of the power plant, but also to general control including other nonlinear control. For example, the present invention can be applied to output control of a nuclear power plant. In this case, the state variables used in the control are preferably a main steam flow deviation, a core pressure deviation, a neutron flux deviation, a core flow deviation, a feedwater flow deviation, and the like. When the present invention is applied to the control of the rotation angle of the motor, the rotation angle deviation, the rotation angular velocity or the differential value of the rotation angle deviation, and the integral value of the rotation angle deviation are used as the state quantity. When applied to the control of the rotational angular velocity of the motor, the rotational angular velocity deviation and the integral value of the rotational angular velocity deviation are used as the state quantity. Furthermore, when applied to the active suspension control of an automobile, a vibration displacement, a vibration speed, and a vibration acceleration are set as state quantities. By switching between non-linear control and linear control with these state quantities, or by changing the contribution ratio, automatic control with high safety and imitating the control of a skilled worker can be performed. .

[0048]

According to the present invention, a transient response that is always within an allowable range can be guaranteed for any controller including a fuzzy or a neural network, and safety can be improved. .

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a control device according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a contribution rate determining unit shown in FIG. 1;

FIG. 3 is an explanatory diagram of the operation principle of the present invention.

FIG. 4 is an explanatory diagram of a specific implementation method of the present invention.

FIG. 5 is an explanatory diagram of a specific operation of the present invention.

FIG. 6 is a configuration diagram of a control device according to another embodiment of the present invention.

FIG. 7 is a configuration diagram of a control device according to still another embodiment of the present invention.

8 is a configuration diagram of a display screen in the control device shown in FIG.

FIG. 9 is a configuration diagram of a display screen different from that of FIG. 8;

[Explanation of symbols]

11: Nonlinear control controller, 12: Linear controller, 13: Contribution ratio determining means, 14: Plant, 62: State quantity monitoring means, 63: Display device, 64: Alarm generator.

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁶ Identification number Agency reference number FI Technical indication location G05B 23/02 301 0360-3H G05B 23/02 301Z G21C 17/00 GDB G21D 3/00 GDBA G21D 3 / 00 GDB 3/08 GDBQ 3/08 GDB G21C 17/00 GDBW (56) References JP-A-1-250103 (JP, A) JP-A-2-8903 (JP, A) JP-A 64-32301 ( JP, A) JP-A-1-293402 (JP, A) JP-A-3-164903 (JP, A) PROCEEDINGS OF THE IEEE, 78-3! (1990-3.) (U.S.A.), STEPHEN BOYD, CRAIG BAR RATT AND STEPHEN NORMAN, "LI NEAR CONTROLLER DE SIGN: LIMITS OF PER FORFORMANCE VIA CONVISION OPTIMATIVE IPS. Applied Division Magazine), 111-D ~ 3! (No. 3-3-20), Nobunori Matsunaga, Shigeyasu Kawaji, "Fuzzy Hybrid Control of DC Motor", 195− 200

Claims (6)

(57) [Claims] 1. A method using non-linear control means and linear control means.
In a control method for controlling a controlled object by
In the state space with the state quantity as the coordinate axis, the nonlinear control means
Response when switching control to the linear control means from
Is set in advance to be within the allowable range, and
The state quantity of the controlled object controlled by the linear control means is
The non-linear when it is foreseen to go out of the convex region
Switching from the shape control means to the linear control means
A control method characterized by controlling an object . 2. The non-linear control means according to claim 1, wherein
When the control is switched from the
A control method characterized by switching step contribution ratios in stages . 3. A nonlinear control means for controlling an object to be controlled.
And linear control means and the state quantities of the closed loop system as coordinate axes
In the state space from the nonlinear control means to the linear control means
Convex where the transient response when switching control is within the allowable range
The region is set in advance and is controlled by the nonlinear control means.
The state quantity of the controlled object that has been
From the non-linear control means when
Control for controlling the control object by switching to a shape control means
A control device comprising: means switching means . 4. The non-linear control means according to claim 3, wherein
When the control is switched from the
Equipped with a contribution ratio determining means for changing the contribution ratio of the step in stages
A control device characterized in that: 5. The method according to claim 3, wherein
When the state quantity of the control target goes out of the convex region or
Equipped with an alarm means for issuing an alarm when it is foreseen to leave
A control device characterized in that: 6. The method according to claim 3, wherein :
The state space is displayed as coordinates, and the
Display means for displaying the position of the state quantity of the control target
A control device characterized by the above-mentioned .