JPH0746282B2 - Non-interference control method and apparatus - Google Patents

Description

TECHNICAL FIELD The present invention relates to a method and apparatus for controlling a controlled object having a large number of variables necessary for control.

[Conventional technology]

Variables required to control the controlled object (called control amount)
In a system in which a plurality of control variables exist, various control variables may be affected by various adjustments of the control variable operation amount (called the control variable) in order to determine one control variable. . This phenomenon is generally called control amount interference, and impairs control precision and stability.

For example, in the control system of a rolling mill, if the delivery side plate thickness and forward tension of the rolled material are used as control variables and the roll reduction command and roll speed command of the rolling mill are used as manipulated variables, the roll reduction process is performed in order to control the delivery side plate thickness to a predetermined value. When the command and roll speed command operations are executed, the front tension, which is another control amount, may be interfered with and subject to fluctuations.

Conventionally, in order to prevent such interference, in a multi-variable control system (control system of a rolling mill) having three inputs (operation amount) and three outputs (control amount), mutual interference between the control amounts is previously compensated as an input. A method using such a method has been proposed (Japanese Patent Laid-Open Nos. 59-179209 and 59-202108). In these conventional techniques, non-interference between the controlled variables is performed so that the relationship of one input and one output is finally achieved.

[Problems to be solved by the invention]

According to the above-described conventional example, mutual decoupling is performed for all control amounts. According to such a technique,
When the scale of the controlled object increases, that is, when the number of controlled variables increases, the amount of calculation required for decoupling becomes enormous. For example, in the matrix operation performed in the process of decoupling, if the number of rows or columns is n times, the number of individual calculations required for the matrix operation is n ² times. For this reason, there are inconveniences such as an increase in the size of the device required for control, a delay in control, and the inability to perform real-time control.

Further, since all the control amounts are made non-interfering uniformly without considering the individuality of the controlled object, there is a problem that it is difficult to perform efficient control without waste.

Further, in the above-mentioned conventional example, the n-input n-output system is exclusively decomposed into the 1-input 1-output system.
There is a problem that it is useless for the output (n ≠ m) system.

The object of the present invention is to solve the above-mentioned drawbacks of the conventional example, and to prevent the control amount from increasing without increasing the number of control devices and delaying the control time even for a large-scale controlled object having a large number of control amounts. It is an object of the present invention to provide a control method and device capable of achieving interference.

Another object of the present invention is to provide a non-interference control method and device that is efficient and efficient.

From another point of view of the object of the present invention, a non-interference control method and apparatus capable of effectively performing non-interference control even for control targets having different numbers of inputs (operation amount) and outputs (control amount). To provide.

[Means for solving problems]

The feature of the method of the present invention to achieve the above object is that the entire controlled object is divided into a plurality of partial control systems (hereinafter referred to as blocks), and at least the blocks are controlled so as not to interfere with each other. In point. Here, according to the broadest idea of the present invention, how to divide blocks is such that one block has at least one operation amount, and at least one block has two or more operation amounts or control amounts. It is optional. Preferably, in the block formation,
Focusing on the characteristics of the controlled object and the controlled variable, the controlled variables that are closely related to each other are grouped together and made into blocks. The control amount having a close relationship with the one control amount means a control amount having a relation easily affected by the change of the one control amount. For example, in the controlled object, it is a controlled amount in each of two parts that are physically, positionally, and temporally close to each other. In other words, when one controlled variable fluctuates, the control is blocked so that the controlled variable fluctuations in the other blocks are smaller than the other controlled variable fluctuations in the own block.

Then, a first compensation element that cancels the control amounts of the other blocks that affect the control amount of the one block is added, and a second compensation element that determines the operation amount for the control target is added to the control amount of the one block. In addition, in addition to the controlled variable added with the first compensating element and the controlled variable added with the second compensating element, a third compensating element for canceling the influence exerted on the one block system from other blocks is added, The point is that the controlled object is controlled by using the controlled variable compensated by the compensation element 3 as the manipulated variable of the controlled object.

A feature of the present invention is that the control means divides the controlled object into a plurality of blocks each having at least one operation amount, and calculates and determines the operation amount for the controlled object for each block. And a first non-interference control means for inputting a control amount of the block and computing a compensation element for the control amount so as to cancel the interference between the control amounts of the blocks concerned, and an operation between the preceding blocks. The calculation results of the control means and the first non-interference control means are input so as to cancel the interference between the quantities, the compensation element for the operation quantity of the control means is calculated, and the compensated operation quantity is applied to the controlled object. The second non-interference control means for outputting is provided.

[Action]

According to the method of the present invention, since the non-interference between the blocks is not a uniform non-interference for each control amount, even if the number of control amounts is increased, the non-interference can be achieved by selecting the number of blocks. The calculation for does not increase to the left. Further, according to a preferred aspect of the present invention, the control quantities having a close relationship are considered as one group (one block), rather than making the non-interference uniform for each control quantity, so that it is impossible within the block. Decoupling does not have to be performed. That is, it is possible to suppress waste and inefficiency such as an increase in the amount of control calculation for decoupling, a complicated device, and a need for a large control signal (and thus a large mechanical operation at the operating end). Furthermore, in the method of the present invention, the decoupling between blocks is not the decoupling in the form of one input and one output, so that the number of inputs (the number of manipulated variables) and the number of outputs (the number of controlled variables) are generally used. ) Are applicable to different controlled objects.

According to the device of the present invention, since the control means executes the control calculation in block units that have been made non-interfering, the control means can be subdivided even if the controlled object becomes large in scale. Yes, so-called parallel operation is possible. Therefore, high-speed control and real-time control are possible.

Further, according to the present invention, since feedback compensation is performed on the feedback element and feedforward compensation is performed on the feedforward element, the control systems can be handled in parallel and highly accurate control can be performed.

〔Example〕

 The present invention will be described in more detail below.

FIG. 1 shows a case where the present invention is applied to control of a tandem rolling mill. In the figure, the rolling mill is a pair of rolling rolls 101, 10
2, 103 is representatively shown, by which the rolled material 104 is rolled in the direction of the arrow in the figure. The state 105, 106, 107 of the rolling mill is the converter 1 that performs processing such as matching the unit systems.
08,109,110, non-interference control system 3, optimal control system 111,112,
113 and dead time elements 114 and 115, respectively. The dead time element is for considering the time required for the rolled material to move to the adjacent stand and reflecting this time in the control, and its output is input to the optimum control systems 112 and 113. The outputs of the optimum control systems 111, 112, 113 are input to the non-interference control system 3. The output of the non-interference control system 3 is input to the actuators 116, 117, 118. These actuators are composed of a roll speed control system, a rolling down control system, and the like, and by these, the rolled material 104 is controlled to a desired plate thickness with a desired accuracy.

In the present embodiment, a control system for such a tandem rolling mill is
It is divided into blocks and non-interference control is performed between the blocks. Therefore, first, a linear model of the rolling mill to be controlled is obtained. As shown in FIG. 2, consider the ith stand and the (i + 1) th stand counting from the entrance side of the tandem rolling mill.

In order to obtain a linear model, the strip thickness deviation, roll opening, rolling load deviation, advance rate deviation, tension deviation, roll speed deviation, mass flow, rolling material speed (sheet speed) in the rolling mill
Are quantified as follows.

The output side thickness deviation Δh _i can be calculated by the following equation (1) from the gauge meter equation.
It is represented by.

However, Δh _i : Outlet side plate thickness deviation ΔS _i : Roll opening deviation ΔP _i : Rolling load deviation K _i : Elastic coefficient deviation i: i-th stand The roll opening is represented as an output of the rolling reduction device.
For simplification, if the reduction control amount is a first-order lag system, it is expressed by the following equation (2).

However, ΔS _pi : Roll opening command ΔT _si : Time constant of the rolling down control device K _si : Gain of the rolling down control device For rolling load deviation, Taylor's expansion of Hill's rolling load approximation formula is used to obtain only the first derivative. By making it linear,
It is expressed by the following equation (3).

However, P _i : rolling load formula H _i : inlet side plate thickness ΔH _i : inlet side plate thickness deviation h _i : outlet side plate thickness T _fi (= T _i ): outlet side tension ΔT _fi (= ΔT _i ): outlet side tension deviation T _bi (= T _i-1 ): Tension on the entry side ΔT _bi : Tension deviation on the entry side The deviation Δf _i of the advance rate is calculated from the advance rate equation f (H _i , h _i , T _fi , T _bi ), and the following (4 ) Get the expression From the tension formula, the following formula (5) is obtained.

However, E: Young's modulus of rolled material b: Strip width of rolled material L: Distance between the i-th stand and the (i + 1) th stand Δv _{ei + 1} : The i + _1th stand-in plate speed v _0i : The i-th stand plate speed When the speed control system is a first-order lag system, its differential equation is the following equation (6).

However, Δv _Ri : Roll speed deviation T _Vi : Time constant K _Vi : Gain Δv _Pi : Roll speed command deviation The following equation (7) is obtained from the mass flow equation.

The following formula (8) is obtained by Taylor expansion of the plate speed formula and obtaining only the first derivative.

Δv _0i = v _Ri Δf _i + (1 + f _i Δv _Ri (8) Here, the i-th stand is considered as an independent control system, and a vector (state vector) having a state variable representing its internal state as an element is X _i. , And a vector (manipulation vector) U _i having the input as an element, X _i and U _i are respectively expressed as follows.

X _i = [Δh _i , Δv _Ri , ΔT _Vi ] ^T U _i = [ΔS _Pi , Δv _Pi ] ^{T Using} these notations, the above equations (1) to (8) are rearranged to form the differential equation Consider to represent with.

First, the following expression (9) is obtained from the expressions (1) to (3).

Next, the equation (6) is modified to obtain the following equation (10).

Next, the following equation (11) is obtained from the equations (4), (5), (7) and (8).

From the above equations (9) to (11), the roll opening command deviation, which is a control amount, for the output side plate thickness deviation Δh _i , the roll speed deviation Δv _Ri , and the exit side tension deviation ΔT _{i which} are the manipulated variables in this embodiment. The relationship between ΔS _Pi and roll speed command deviation Δv _Pi is
It became clear in the form of a differential equation.

Taking the 3 tandem mill as an example, these relational expressions are expressed in the form of a matrix as shown in FIG. In the figure,
The terms on both sides of (a) and (b) correspond to each other. In addition, a _ij , b _ij , and d in FIG.
_ij indicates the coefficient of each variable in the expressions (9) to (11).
For example, in equation (10), when i = 2, in the first term on the right side Therefore, it corresponds to a ₅₅ in FIG. Similarly, Is. The third term on the right side is recognized as an observable disturbance.

According to the concept of the present invention, the illustrated state equation is divided into a plurality of blocks so as to include at least one manipulated variable, and these blocks are made non-interfering with each other. In the case of the present embodiment, a specific dividing method is preferably set for each stand. Between stands, they are related only by the rolled material that connects them, that is, only by the rolling tension, whereas in the stands, not only tension but also plate thickness, roll speed and tension are related to each other. This is because the manipulated variables are closer to each other than between the stands.
According to FIG. 3, it can be seen that the amount of state related between the stands is only tension. For example, terms that affect the second stand to the first _{stand, a 34, a 35, a} 36
Only, both of which relate to tension deviations ΔT ₁ and ΔT ₂ . On the other hand, looking inside the 1st stand,
There are terms ₁₁ , a ₁₃ , a ₂₂ , a ₃₁ , a ₃₂ , and a ₃₃ , which are interrelated over Δh ₁ , Δv _R1 , and ΔT ₁ .

Next, a specific method of decoupling between stands of the first term (system matrix A) on the right side in FIG. 3 will be described by taking the second stand as an example.

When the equations related to the second stand are extracted from the state equation shown in FIG. 3, the following equations (12) to (14) are obtained.

The term where the subscript of each variable is 1 or 3, that is, (1
The underlined terms in Eqs. 1) to (13) are interference terms from other stands, and by canceling them out, decoupling can be achieved.

FIG. 4 shows a control system for such decoupling. Hereinafter, each interference term will be described.

The tension deviation ΔT ₁ 201 of the first stand is calculated by the adder 203 via the block a ₄₃ 202 as shown in the first term on the right side of the equation (12).
Entered in. For example, "block a ₄₃ " means "a
Amplifier with a gain of ₄₃ ". The output of the adder is Δ _2, which is integrated by the integrator 204 and becomes the plate thickness deviation Δh ₂ 205. Therefore, as fed back and gain combined in full Eid Fuo word for non-interfering, via a non-interfering gain f ₂₁ 206 a [Delta] T ₁ 201, and inputs to the control matrix 207 as input. The output of the control matrix 207 is applied to the adder 203. If the output of the adder 203 does not include the ΔT ₁ component, non-interference is established. That is, from the condition of b ₃₃ × f ₂₁ × ΔT ₁ + a ₄₃ × ΔT ₁ = 0, It is understood that it suffices to select a gain that

Next, a method of canceling the underlined interference term in equation (14) will be described.

To the tension deviation ΔT ₁ of the first stand, block a ₆₃ 208, adder 2
It is added to the adder 210 via 09. The plate thickness deviation Δh ₃ of the third stand is added to the adder 212 via the block a ₇₆ 211, the speed deviation Δv _R3 is added to the adder 212 via the block a ₈₆ 213, and the tension deviation ΔT ₃ is added to the block a ₉₆ 214. To the adder 212, and the outputs of the adders 210 and 212 are added to the adder 213.

On the other hand, in this embodiment, the blocks f ₂₅ 214, f ₂₂
215, f ₂₃ 216, f ₂₄ 217 and block 218 are used. ΔT ₁
Δh ₃ to the adder 219 via the block f ₂₅ 214 with gain a ₆₃
Is added to the adder 219 via the block f ₂₂ 215 with gain a ₇₆ and Δv
_R3 is added to the adder 219 via the block f ₂₃ 216 with gain a ₈₆ and Δ
T ₃ is applied to the adder 219 via the block f ₂₄ 217 having the gain a ₉₆ , and the output of the adder 219 is input to the block 220 as the speed command deviation via the block 218. Block 2
The output of 20 goes through an integrator 221, a block 222, and an adder 210,
It is added to the adder 213.

At this time, the gain of the block f ₂₆ is calculated by the integrator 220, the block 22
1,222, reciprocal of transfer function up to adders 210, 213, that is, If S: Laplace operator is used, the gain from the adder 219 to the adder 213 becomes −1 equivalently, the effects of ΔT ₁ , Δh ₃ , Δv _R3 , and ΔT ₃ are canceled out, and the non-interference with respect to the second stand. To establish.

In the same manner as described above, the decoupling can be performed for each of the other stands, and the decoupling between the blocks is established by these operations.

If the blocks having the same number of inputs and the same number of outputs are regarded as −input−output blocks and decoupling is performed, decoupling between the blocks and decoupling between the input and output can be realized.

A state equation of a control target (including a non-interference control system) adopting such a configuration will be described with reference to FIG. Using the relationship of the block 150, which represents the state of each stand of the rolling mill, it is expressed as a control object represented by the block 160, and each block A ₁ , A ₂ , A ₃ , B ₁ , B ₂ , B ₂ , When B ₃ is decomposed and rearranged, it becomes three equations of state like block 170. In this case, since the state equations are each of the third order, there is an effect that the optimum control (described later) can be realized only by a simple matrix calculation of 3 × 3.

Next, assuming that non-interference between blocks has been achieved,
A case where so-called optimum control is applied to the rolling mill control of this embodiment will be described. Optimal control is a control method that appropriately adjusts states and manipulated variables to minimize their deviation or minimizes manipulated variables.The classical control system was focused only on the relationship between input and output in the past. Unlike the above, since the internal structure of the controlled object can be controlled in an intrusive manner, its effect on improving the characteristics is expected.

In the present embodiment, as an example, the optimum control system 112 for controlling the second stand 102 in FIG. 1 will be described. The optimum control system 112 inputs the above-mentioned state vector X and visible disturbance W and performs a predetermined calculation. (Described later). Then, the result is added to the calculation result for the non-interference in the non-interference control system 3 to control the actuator 117 of the second stand.

FIG. 6 shows the internal structure of the optimum control system. In the figure, 10
Reference numeral 2 denotes a second stand which is a controlled object, and as described above, uses ΔS _P2 and Δv _P2 as manipulated variables and Δh ₂ , Δv _R2 and ΔT ₂ as controlled variables (state variables). Δh ₂ is added to the adder 251 via F ₁₁ 250, Δv _R2 is added to the adder 251 via F ₁₂ 252,
ΔT ₂ is applied to the adder 251 via F ₁₃ 253. Note that F _ij
Is a feedback parameter obtained from the theory of optimal control. The output of the adder 251 is input to the adder 254.

Similarly, Δh ₂ is fed to adder 256 via F ₂₁ 255 and Δv _R2 is fed to F ₂₂ 2
57 is input to the adder 256, ΔT ₃ is input to the adder 256 via F ₂₃ 258, and the output of the adder 256 is input to the adder 259.
Δh ₂ is input to the adder 261 and ΔT ₂ is input to the adder 261, and the output of the adder 261 is added to the adders 263 and 254.
ΔT ₂ is input to the adder 266, Δh ₂ is input to the adder 266, and the output of the adder 266 is input to the adder 269 via the adder 268 and the integrator 269.

The visible disturbance w ₂ is, for example, the entrance side plate thickness deviation ΔH _i in the above-described embodiment, and is optimally controlled with respect to m _i input to the adder 254 via m ₁ 270 and to the adder 259 via m ₂ 271. It is a feedforward parameter obtained from the theory of.

As described above, when the present invention is applied to the control of the tandem mill,
In addition to the original effects of the present invention, there are the following effects. That is, since it is possible to control each stand independently, when the number of stands is changed, only a partial change of the stand is required to change the control device, and it is possible to significantly change the entire system. Since no major changes are required, the control system is easier to handle.

For example, when adding a stand, conventionally, it was necessary to construct a new control system for the entire tandem mill including the added stand, and a major change was forced. Since it is made non-interfering, the control of the entire tandem mill can be executed by newly adding the control device only for the additional stand. Similarly, in the case where a certain stand of the tandem mill is used as a dummy stand (for example, during roll change during running), only the function of the control device portion that controls the stand is removed.

Next, the present invention is also applicable to a single stand mill. In this case, the blocks and the non-interference between the blocks can be realized regardless of the manipulated variables and the controlled variables, so that the degree of freedom in controlling the rolling mill can be increased, and precise control can be achieved.

Although the present invention has been described above by taking the control of the rolling mill as an example, the present invention is not limited to this and can be widely applied to a multivariable control system. Although a detailed description is omitted to avoid complication, as an application example, decoupling of a drive system and a steering system in an automobile, decoupling between axes of a multi-axis robot,
This includes decoupling between power supply stations in the power system.

Therefore, the present invention will be described below in a general form. FIG. 7 shows an outline of a control system to which the present invention is applied.
In the figure, the state 2 of the controlled object 1 is input to the non-interference control system 3 and the optimum control systems 4, 5, and 6. The optimum control systems 4,5,6 use the state 2 to determine the manipulated variables 7,8,9 for the controlled object 1.
The non-interference control system 3 receives the above-mentioned state 2, the manipulated variables 7, 8 and 9 of the optimum control systems 4, 5 and 6, and the optimum manipulated variable 1 and the non-interference control are combined.
0, 11, 12 are determined and output to the controlled object 1. The present invention is
As described above, the controlled object 1 is divided into a plurality of blocks so that each block has at least one manipulated variable, and at least one block has two or more manipulated variables or controlled variables. And for each block,
It controls the controlled object. The state equation of controlled object 1 is given by the following equation: = AX + BU (15). Here, if the system matrix A is divided into finite blocks, and the diagonal block A _D and the non-diagonal block are A _N , the cistam matrix can be expressed by the following equation (16).

A = A _D + A _N (16) Similarly, control matrix B is divided into diagonal block B _D and non-diagonal block B _N.
Split into.

B = B _D + B _N (17) Next, the state vector X is divided into N blocks. However, this N division depends on the control target and cannot be uniformly determined. The viewpoint when dividing is that the degree of the relationship between the manipulated variables or the controlled variables among the divided blocks is
It should be smaller than the degree of the relationship between the manipulated variables or controlled variables in each block.

By dividing the state vector into N, the corresponding operation vector U is also divided into N. Similarly, the system matrix and the control matrix are also divided into N.

That is, the system matrix is decomposed into the system submatrix whose elements are a _ij , the control matrix into the operation submatrix b _ij , the state vector into the state small vector X _i , and the operation vector into the operation small vector U _i. It FIG. 9 shows the equation (15) in the form of a matrix using these. Here, the matrix obtained by dividing the system matrix A of FIG. 9 into A _D and A _N and the control matrix B are
Figure 10 shows the matrix divided into B _D and B _N.

FIG. 8 shows the relationship among the control target 1, the non-interference control system 3, and the optimum control system 4 for controlling a certain block among the control targets.

The controlled object in FIG. 8 is a block diagram of the equations (15), (16), and (17). That is, the adder 13 includes a state vector X14 and a diagonal block system matrix A _D 15
And the product of X14 and the off-diagonal block system matrix A _N 16 are input. That is, the product of the system matrix A obtained by multiplying the expression (16) by X and the state vector X14 is generated in the adder 13 (the following expression (18)).

A _D X + A _N X = AX (18) The adder 17 has a product of the operation vector U18 and the diagonal control matrix B _D 19 and the adder 13 through the operation vector U18 and the non-diagonal control matrix B _N 20. The product of and is input. That is, U18
The product of the control matrix B and the operation vector U represented by the following formula (19), B _D U + B _N U = BU (19) multiplied by is input. Then, the adder 17 obtains the following equation (20), AX + BU (20), which is the sum of the product of the system matrix A and the state vector X obtained by the equation (18), and their values are given by the equation (15). Therefore, it is equal to the differential value 21 of the state vector X. Since X14 is the integration of 21, 21 is integrated through the integration matrix 22, and the state vector X14 is obtained. In S, S is a Laplace operator and I is an identity matrix.

Here, assuming that the state vector X is an nth-order vector and the operation vector U is m inputs, the system matrix is an n × n suppression matrix and the control matrix is an n × m matrix. Considering the internal state of the controlled object 1, that non-interference control is performed means that the system matrix A and the control matrix B are only diagonal blocks. Considering the adder 17, the product of the state vector X14 as an interference term and the off-diagonal system matrix A _N 16 is added in a feedback manner, and the product of the operation vector U18 of one interference term and the off-diagonal control matrix B _N is added. Is added as a feedforward. In order to efficiently perform the non-interference control, it is preferable to perform the feedforward compensation for the feedforward element and the feedback compensation for the feedback element. This is because the control system enables parallel handling.

Since the control matrix B adds the operation vector U to the derivative of the state vector X in a feedforward manner, this non-interference component is obtained by the feedforward compensation.

The command vector r23 generated in the non-interference control system 3 becomes the operation vector U via the feedforward compensation G24,
It is added to the control matrix B.

In this case, assuming that the general form of the control matrix B when the non-interference control is performed is B ′ _D and the non-interference control (feedforward term) is performed, the following equation (21) is established.

(B _D + B _N ) G = B _D ′ (21) This is multiplied by the pseudo-inverse matrix of (B _D + B _N ) on both sides of equation (21) from the front according to the method generally used in this field. , (22) is obtained.

G = (B _D + B) ⁻ B _D (22) However, FIG. 11 shows a specific example of the method for realizing the pseudo inverse matrix (22) of (B _D + B) ⁻ : B _D + B. This shows the relationship between the feedforward compensation element G24, the operation vector U18, the diagonal and off-diagonal control matrices B _D and B _N, and the differential value 21 of the state vector in FIG. Of these, u _i related to the i-th block and its command vector 23 are illustrated. The interference terms to other blocks will be considered with reference to FIG.

Considering the influence from the i-th block to the j-th block, the command vector r _i 23 becomes the operation small vector u _i 18 via the non-interference feedforward compensation element g _ii 50, and the control small matrix B _ij 51 Then, it becomes the differential vector 21 of the j-block small state vector X _j via the adder 54. In addition to the figure, the adder 54 also receives an amount based on r other than the j-th one. When these relationships are expressed by an equation, the following equation (23) is obtained.

X _ji = b _ij · g _ii r _i (23) where _ji is the interference from the i-th block. To cancel this, add the newly added non-interference block element g
Add r _i to the manipulated small vector u _j via _ij 52. The manipulated variable applied from u _j is added to the derivative of the small state vector X _j via the control submatrix b _jj . The value is given by the following equation (24). _ji ′ = b _ij g _ij r _i (24) (22) When the sum of Eq. (23) becomes zero, the interference from i-lock to j-block disappears (Equation (25)).

b _ij · g _ij r _i + b _jj g _ij r _i = 0 (25) By _{modifying the} formula (25), the following formula (26) is obtained.

^{_{g ij = b jj - b jj}} g ii ... (26) these operations sequentially row Tsuteiki, (until 1 input 1 output) each element until the scalar non-interference can be achieved as a whole is performed.

Next, the decoupling of the interference component of the system matrix will be considered. Since the state is fed back by the feed back component, it cannot be canceled by the feed forward operation of the operation amount of the interference component. Therefore enter the state vector X the fed back compensation element F _B 25 of Figure 8, F _B 25
Is added to the adder 26, and the output of the adder 26 is added to the feedforward compensation element G24. Then, the output is added to the differential value 21 of the state vector via the control matrix B.
On the other hand, regarding the interference term, the state vector X14 is added to the differential value 21 of the state vector via the non-diagonal system matrix A _N 16 (the second side on the left side of the equation (27)). Section). When this term is canceled, non-interference is established. Therefore, decoupling is achieved by determining the gain of the feedback compensation element F _B 25 so that the following equation holds. _ji = BGF _B X + A _N X = 0 (27) By transforming the equation (26), the following equation (28) is obtained.

_{BGF B = -A N F B =} -G - B - A N ... (28) are intended only for these non-interference control system 3, except always preparative interference term of the controlled object, a desired response to a control object It is not a control system that allows it. Therefore, a control system for determining the manipulated variable of each block is necessary for that purpose. Since deblocking between blocks has already been achieved,
As for the control system, each block is considered as an independent control target, and any type of control system may be used as long as it controls them individually. As described above, it is a great effect of the present invention that the degree of freedom in selection of the control system can be increased. Here, the case where the optimum control system is adopted will be described.

In Fig. 8, state vector X14 and observable disturbance W
Is added to the optimum control system 4, and the optimum control system calculates based on these inputs and outputs it to the adder 26. In the adder 26, the output of the optimum control system 4 and the output of the front non-interference feedback compensation element F _B 25 are added, and the result is input to the feedforward compensation element 24.

The optimum control system 4 can be used for each block. As an example, FIG. 12 shows the internal structure of the optimal control system for the i-th block. In the figure, the observable disturbance W _i is fed to an adder 28 via an optimum controlled object 26 (hereinafter simply referred to as a controlled object) including the controlled object 1 and the non-interference control system 3 and an optimum feedforward element M27. Added.

A command vector r _i for obtaining the desired response of the control system is input to the adder 30, and the output of the controlled object 26
y _i is subtracted and input, and the deviation is input to the optimum feed back element F _i 29 together with the state vector X _i of the controlled object 26 via the integration element 31. The output of the optimal feedback element 29 is applied to adder 28. The output of the adder 28 is applied as a command 7 to the controlled object.

The method of determining the values of the optimum control elements M _i 27 and F _i 29 described above is described in, for example, Proceedings of the Japan Society of Measurement and Control, Volume 17, No. 2 (April 1981).
Mon) Discussed on pages 16 to 21.

That is, the state equation of the controlled object _i = A _i X _i + B _i v _i + E _i W _i (29) y = C _i X _i + D _i v _i and the evaluation function J = ∫ (X _i ^T QX _i + v _i ^T Rv _i ) dt (30) is calculated to find the compensation elements F _i and M _i .

F _i is the Riccati equation P _i {A _i −B _i R _i ⁻¹ D _i ^T Q _i C _i } + {A _i ^T −C _i ^T Q _i D _i R _i ⁻¹ B of the following equation (31). _i ^T } P _i
−P _i B _i R ⁻¹ B _i P _i + C _i ^T {Q _i −Q _i D _i R _i ⁻¹ D _i ^T Q _i } C _i = 0 ... Using the solution P of (31), the following equation (32 ) Is required.

F _i = − (P _i + D _i ^T Q _i D _i ) ⁻¹ (D _i ^T Q _i C _i + B _i P _i ) ... (32) M _i is calculated using the above F _i by the following equation (33). , (34), Γ _i =-(A _i + B _i F _i ) ^-1 B _i (33) Θ _i =-(A _i + B _i F _i ) ^-1 E _i (34) The following equation (35) ) Can be obtained as.

M _i = − (Γ _i ^T P _i Γ _i ) ⁻¹ P _i ^T P _i Θ _i (35) In the above example, the optimal control system 4 is treated as an optimal reguulator, but the optimal reguulator is optimal. Since the extension to the servo is easy, the present invention can be applied to both.

As the control means for each block used in the present invention, various control systems other than the above-mentioned optimum control system can be applied. For example, pole placement method by state feedback, PID
Control, etc. In particular, according to the above-described embodiment, since the stands of the tandem mill are made non-interfering with each other, the normal PID control can be applied to each stand, and a complicated system is controlled by such a simple control system. It is also an important effect of the present invention to be able to do so.

〔The invention's effect〕

As described above, according to the present invention, even for a large-scale controlled object having a large number of controlled variables, this is divided into blocks and the interference is made between the blocks. An appropriate non-interference control between the control amounts can be achieved without inviting.

[Brief description of drawings]

1 and 2 are diagrams showing a rolling mill control system according to one embodiment of the present invention, and FIGS. 3 and 5 are diagrams showing calculation formulas used in one embodiment of the present invention in the form of a matrix, 4 and 6 are views showing the configuration of a rolling mill control system according to an embodiment of the present invention, and FIGS. 7, 8, 11 and 12 are for explaining the present invention in general. Are conceptual diagrams, and FIGS. 9 and 10 are diagrams showing the calculation formulas used in the present invention in the form of a matrix. 101,102,103… Rolling roll, 104… Rolling material, 108,109,110
... converter, 111, 112, 113 ... optimal control system, 114, 115 ... dead time element, 3 ... non-interference control system.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Harumi Maruyama Inventor Harumi 4026 Kuji-cho, Hitachi City, Ibaraki Prefecture Hitachi Research Laboratory, Hitachi Ltd. (72) Goo Shimizu 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture (56) References JP 61-125603 (JP, A) JP 59-226903 (JP, A)

Claims (6)

[Claims] 1. A non-interference control method for controlling a controlled object controlled by a plurality of manipulated variables and at least one controlled variable by dividing it into partial control systems, wherein the controlled object is at least one manipulated variable and at least two or more. And a first compensating element for canceling the control amount of the other partial control system that affects the control amount of the one partial control system. A second compensating element for determining the manipulated variable for the controlled object is added to the controlled variable of
To the control amount including the second compensation element and the control amount including the second compensation element, a third compensation element for canceling the influence exerted from the other partial control system on the one partial control system is added, Three
The non-interference control method, wherein the controlled object is controlled by using the controlled variable compensated by the compensating element as described above as the operation amount of the controlled object. 2. The partial control system according to claim 1, when the one of the control amounts fluctuates, the partial control system is more likely to change than other control amounts included in the partial control system to which the control amount belongs. The non-interference control method is characterized in that the variation of the control amount included in the other partial control system is divided so as to be smaller. 3. The device according to claim 1, wherein the controlled object is a tandem rolling mill having a plurality of rolling stands, and the partial control system is each rolling stand of the tandem rolling mill. Non-interference control method. 4. A non-interference control device for controlling a controlled object controlled by a plurality of manipulated variables and at least one controlled variable by dividing it into a plurality of partial control systems, wherein each controlled object has at least one manipulated variable. And a control means for calculating and determining an operation amount for the controlled object for each of the partial control systems, and an interference between the controlled variables between the partial control systems. The first non-interference control means for inputting the control amount of the partial control system to calculate a compensation element for the control amount and the interference between the manipulated variables between the partial control systems are canceled so as to cancel. As described above, a calculation result of the control means and the first non-interference control means is input, a compensation element for the operation amount of the control means is calculated, and the compensated operation amount is output to the control target. Decoupling control apparatus characterized by comprising a non-interference control means. 5. The control means according to claim 4, when one of the control amounts fluctuates, the control means changes more than other control systems included in a partial control system to which the control amount belongs. It is provided for each divided partial control system so that the control amount included in the other partial control system is smaller than the fluctuation of the other control system included in the other partial control system. Non-interference control device. 6. The apparatus according to claim 4, wherein the controlled object is a tandem rolling mill having a plurality of rolling stands, and the control means operates for each stand of the tandem rolling mill. Characteristic non-interference control device.