JPH0777647B2 - Dimension control method in wire rod and bar rolling - Google Patents

Description

The present invention relates to a dimensional control method in continuous rolling of wire rods or bars. [Prior Art] Conventionally, the following examples have been given as the dimension control method for a wire rod or a rod. As a first example, "Plasticity and processing" vol20, No.224, P841-849
As shown in (1979-9), the deviation of the measured value of the tension from the target value was obtained by the direct tension detection of the steel bar mill.
Taking into account the mutual influence between stands, the proportional and integral control of tension is performed with the roll rotation speed as the manipulated variable. As a second example, as shown in Japanese Patent Laid-Open No. 54-128469, using the coefficient of influence of reduction on the product width dimension,
The top and bottom dimensions of the product are controlled by the roll reduction of the final stand of the common drive block mill of the wire, the product width dimension is controlled by the reduction of the first stand, and proportional control is performed by the deviation of the actual measurement value from the target value.

A third example is disclosed in Japanese Patent Laid-Open No. 57-88908.
Sea urchin, No.1 stand and No.2 of finishing block mill of wire
No. by changing the relative speed between ~ stand.
By changing the tension between 1 to No. 2 stands,
Deviation of the measured value of the exit width of any one stand from the target value
The product width is controlled according to. As a fourth example, Japanese Patent Application No. 59-218111 filed by the present inventors
As shown in No. 3, the pressure that consists of deformation, load, and temperature.
The roll speed and roll gap are
The operation amount is set to the output side of the final stand of the target rolling function.
Based on the multivariable control theory, the height and width of rolled material in
It is controlled by vector feedback. However, the control methods of the first to fourth examples have the following problems.
There is a point. That is, in the method of the first example, the gain is determined by trial and error.
There is a limit to the optimization. If the temperature and dimensional disturbance are large, the roll may be
There is a limit to high dimensional accuracy due to changes in the gap. In the method of the second example, the coefficient of influence that the reduction has on the product dimensions
Although the reciprocal of is used as the gain of proportional control, the control performance is not
It is clear. In this case, from the finishing stand to the cross-section gauge of rolled material
The problem that the control operation is delayed by the dead time due to the distance.
There is a point. In the method of the third example, a concrete method for determining the gain is shown.
The control performance is unknown. In this case as well, from the finishing stand to the rolled material cross-sectional dimension meter
The control operation is delayed by the dead time due to the distance
There is a problem. In the method of the fourth example, feedback of tension and finish size
It is a multi-variable control, and compared with the non-interference control of tension in the first example.
And high response, and the measured value by the finishing dimension meter
Since it is a clock control, it can be controlled to the target value. But
In the same way as in the first to third examples,
Controls only the dead time due to the distance to the cross-sectional dimension gauge of the rolled material
There is a problem that the operation is delayed. [Problems to be Solved by the Invention] Non-interference control of tension shown in the first example or a conventional loop
Combination of 1 input and 1 output control for each loop like control
In the case, all tensions or loops are
There is a limit to controlling the target value with high accuracy. Those of these
In the method, if the temperature and dimensional disturbance are large, the
There is a limit to the high dimensional accuracy due to the change in the clearance. In the fourth example of multivariable control of tension and finish size, all
The tension can be controlled to the target value with high response and high accuracy,
The feedback of the actual measurement value of the gauge size by the mill exit side dimension meter
The target value can be controlled by control. On the other hand, there is a loopless or loop between stands.
The maximum rolling speed of the vertical drive type mill is around 10 m / s. Therefore, in the method of the fourth example, the mill is ejected before it is controlled.
The material on the side has passed about 10 m, and the entire length of the material has passed.
It is impossible to achieve high dimensional accuracy. [Means for Solving the Problem] The present invention shows a dynamic phenomenon of continuous rolling of a wire rod or bar.
A state equation (rolling model) is created and
Applying the variable control theory, the material height and
Multi-variable control of loop height (loop height deviation between stands
Since it is proportional to the deviation of tension,
It is equivalent to multi-variable control)
It is a method to control the height and width to target values with high response and high accuracy.
The Multi-variables of material height and loop height of each stand above
In control, all loop heights are standard with high response and high accuracy.
The height of the material on the outlet side of each stand is
Calculated from the measured values of roll gap and rolling load of the stand
(Called gauge gauge calculation height)
Finishing dimension by the conventional mill exit side dimension meter for control
Higher response than feedback control using measured value of
Is. Also, the height of the material on the exit side of each stand and the height of the loop are
In variable control, loop control or tension control and
The material height on the outlet side (gauge gauge
When the machine control height is controlled, the mill exit side (finish
The height and width of the material can be controlled to be constant in the longitudinal direction with reference values.
It was confirmed by calculation and experiment that it was capable. (Moni
(No AGC compensation) Multi-variable control of material height and loop height of each stand
Controls the height and width of finishing material on the mill exit side to target values
In this method, the reference value of the material height on the delivery side deviates from the target value.
If the mill exit material height and width deviate from the target values. Therefore, at the gauge meter calculation height, the material size on the mill exit side
The deviation of the measured value of
The relationship between the height deviation and the deviation of the material size on the mill exit side
Each stun is calculated by using a formula that expresses
Output side of the material is converted to the deviation of material height and the added amount is output
By controlling it to 0 as the deviation of the material height
And the reference value of the material height on the delivery side deviates from the target value.
However, the finishing height and width can be controlled to the target values.
(With AGC compensation). In the present invention, the wire rod is formed by the mill including the multi-stage stand.
When rolling a bar or rod, the dynamic phenomenon of continuous rolling is
A non-linear unsteady model of rolling is drawn around the rolling reference state.
Deriving tension (state vector), which is derived by shaping
Mick change (in continuous time system, tension at time t
Time derivative of (state vector), time in discrete time system
Continuous tension (state vector value) at time k + 1
In the time system, at time t, in the discrete time system, the time
state vector, input vector, disturbance vector at k
Dynamic model and output vector
Statics of cuttle, state vector, and disturbance vector
Rolling process consisting of output equations that express relationships
It is expressed by a formula, and by combining it with the mill drive system model,
A state equation for the entire system is created, and the control according to the control method is performed.
Configure the state equation of the entire system used during system design and control
An output vector that represents the control amount and an input vector that represents the manipulated variable
Introduce an evaluation function consisting of and minimize the value of the evaluation function
Calculate the optimum gain of the feedback multivariable control system
It was. In actual process, online control computer
The above optimum gain is stored in (hereinafter abbreviated as "procon")
Then, the following dimension control is performed online using a process controller.
I do. Optimal operation amount (input vector) that minimizes the value of the evaluation function
Is calculated in advance offline.
Optimal gain and state vector measured online
From the output vector and the output vector
As a result, of the rolled material released from the mill's final stand
The size (output vector) is controlled to the target value. Here, roll gap, rolling load, loop height, mill exit side
Control of dimensional measurement cycle and operation of roll gap and roll speed
The cycle is about 0.05 to 0.5 s, and the shorter the cycle, the more detailed
Various controls are possible. [Operation] The present invention is intended to take a dynamic phenomenon of continuous rolling of a wire rod into
Linearizing a linear unsteady model around the rolling reference state
Dynamics of tension (state vector)
Change (in a continuous-time system, the tension (state-based
Time derivative, the time k + 1 in the discrete time system
Value of tension (state vector) at
Is at time k in the discrete time system at time t.
The state vector, input vector, and disturbance vector
And the dynamic model and output vector
Show the static relationship between the state vector and the disturbance vector
The rolling equation of state composed of the output equations
Then, by combining it with the mill drive system model, each stand
Create a state equation for the entire system and control according to the control method.
Configure and control the state equation of the entire system used during system design
From the output vector that represents the quantity and the input vector that represents the manipulated value
Introduce an evaluation function of and minimize the value of the evaluation function
Feedback multi-variable control system operation amount (input vector)
Of the rolled material released from the mill's final stand
Method (output vector) is controlled to the target value
Is a dimension control method characterized by. The symbols used for the explanation are shown in Table 1 and as follows.
Define to. In addition, each vector composed of multiple variables is defined as follows.
I mean. Here, the subscript i, j representing the stand number takes a value of i = 1,2, ..., n, j = 1,2, ... , The usage of the subscripts i and j that represent the stand No.
It is as follows. A variable whose value at the last stand No. n is always 0, for example,
Roll speed target deviation, measured deviation, stand-to-stand deviation
Use j as a subscript for the deviation of tension and the deviation of loop height
The A variable whose value at the final stand No. n is a value other than 0, eg
For example, use the subscripts for the deviation of the roll gap target value and the deviation of the actual measurement value.
And use i. Below, the time is displayed in a discrete time system. 1) Δs_i ^*(K): At each stand of the mill at time k in the discrete time system
Deviation of the rolling roll gap, that is, [{(actual measurement s_i(K)-(reference value s_oi(K))} / {No.
i Standard value of the height of the cross section of the rolled material at the stand exit side h
_0i(K)}] Here, the reference value is the path schedule value (design value)
That is.Δ_i ^*(K): Lock-on value Δsl_i ^*Deviation from = s_i ^*(K) -Δsl_i ^* Δsl_i ^*: Lock-on value = (sl_i−s_oi) / S_0iThe following description
The deviation of the variable x from the lock-on value is also shown in Table 1.
Sea urchin Δ^*= Δx^*−Δxl^* It is represented by. 2) Δt_fj ^*(K): At each stand of the mill at time k in the discrete time system
Deviation of tension, that is, [{(actual measurement t_fj(K))-(reference value t_foj(K))} /
{Standard value of average deformation resistance of rolled material in No.j stand
k_fmoj(K)}] 3) Δy_hj ^*(K): At each stand of the mill at time k in the discrete time system
Deviation of the loop height, that is, [{(actual measurement y_hj(K))-(reference value y_hoj(K))} /
{No.j Reference value y of loop height at stand exit side
_hoj(K)}] 4) Δs_{ref, i} ^*(K): Deviation of roll gap target value at No.i stand, that is,
Then, [{(target value s_{ref, i}(K))-(reference value s_{o, i}(K))}
/{No.i of the height of the cross section of the rolled material on the stand exit side
Reference value h_oi(K)}] 5) ΔN_{ref, j} ^*(K): Deviation of roll speed target value at the No.j stand,
There, [{(target value N_{ref, j}(K))-(reference value N_{o, j}(K))}
/{No.j Stand rotation speed reference value N
_oj(K)}] where the target value of the input vector is the input vector
Output from the process controller to the control unit with an operation amount given by
This is the command value to be sent. 6) Δh_n ^*(K), Δb_n ^*(K): Rolled material on the exit side of the final stand (n-th stand)
Deviation of cross-sectional height and deviation of rolled material width, ie, {(measured value h_n(K)) / (target value h_on (k))-1, (measured value b_n(K)) / (target value b_on (k))-1} Here, the target value of the output vector is the target value
Is a value controlled by. 7) ΔT_{m, i} ^*(K), ΔH_i ^*(K), Δ^*B_i(K): Average material temperature and material cross-sectional height at the No.i stand entrance side
And deviation of material cross-section width {(measured value T_{m, i}(K)) / (group
Quasi-value T_{mo, i}(K))-1, (measured value H_i(K)) / (reference value H
_{o, i}(K))-1, (measured value B_i(K)) / (reference value B
_{o, i}(K))-1} 8) Δ_n ^*(K), Δ_n ^*(K): Rolled material on the exit side of the final stand (n-th stand)
Integral value of cross-section height deviation and rolling material width deviation j: Discrete time time Δt: Control cycle 9) Δ_fj ^*(K): Integral of deviation of tension between stands at No.j stand exit side
value  j: time in discrete time system Δt: control cycle 10) Δ_hj ^*(K): Integral value of the deviation of the loop height at the No.j stand exit side j: time in discrete time system Δt: control cycle 11) (k): Shows the state vector and determines the basic state of the process.
Physical quantity. Here, between the rolls of each stand at the current time k
Deviation from lock-on value of gap, deviation of loop height, low
It consists of the deviation of the rotational speed and its derivative. Sanawa
Then, (k) = [Δ_i ^*(K), Δy_hj ^*(K), ΔN
_j ^*(K), Δ_j ^*(K)]^T Here, superscript T: transpose of matrix vector 12) (k): indicates the output vector, and the value of each element of the vector is the target
Physical quantity controlled as a value. Here, from the lock-on value of the material height at the stand exit side,
It is composed of the deviation of and the deviation of the loop height of each stand. You
That is, (k) = [Δ_i ^*(K), Δy_hj ^*(K)]^T 13) (k): A physical quantity that represents the input vector and is the manipulated variable. Here, the lock value of the target value of the roll gap of each stand is
From the reference value of the target value of roll speed
It is composed of the deviation. That is, (k) = [Δ_{ref, i} ^*(K), ΔN_{ref, j} ^*(K)]^T 14) (k): An object that shows the disturbance vector and affects the process from the outside.
Reason. Here, it is configured by the dimensions and temperature fluctuations on the entrance side of each stand.
The That is, (k) = [ΔT_mi ^*(K), ΔH_i ^*(K), ΔB
_i ^*(K)]^T In general, disturbances are often unknown, and conventional feedback
(4) In the multivariable control theory, the design is done without considering it. Hereinafter, the details of the present invention will be described by taking a 4-stand mill (n = 4) as an example.
explain. Figure 1 shows the structure of the loop between the stands. Numbers 1 and 2 indicate No.i and No.i + 1 stand rolls, respectively. 3 and 4 are guide rolls, and 5 is a kick roll. 6 is the loop height detector, 7 is the pass line, and 8 is the material.
I will. Fig. 2 (a) shows a four-stand mill as an object and the control of the present invention.
The control system is shown in Fig. 2 (b).
-It shows the control system of the AGC + monitor AGC compensation. 9 and 11 are horizontal rolls, 10 and 12 are vertical rolls, and 13 is a dimensional gauge on the mill exit side. 14 is a control process computer. 15 indicates the material. FIG. 3 shows an off-line calculation process and an operation of the control method of the present invention.
The flowchart of online control is shown below. The off-line processing is the continuous rolling phenomenon of the wire and the controller
A model that represents the internal mix with a mathematical model (state equation)
Conversion (this way in this specification) and the inventor
Control system design using multivariable control theory applied to continuous rolling?
It is composed of This process determines the optimum gain that determines the optimum manipulated variable.
Is determined. Online processing is the collection of measured data, offline
Storage of calculated optimum gain, optimum using measured data
From operation amount calculation and output of the optimum operation amount to the operation end
It is constructed and calculated by the process computer. Next, the contents of offline calculation processing and online calculation processing
I will describe. First, each item of offline calculation processing is explained. Here, the numbers outside the margin of the flowchart shown in FIG.
It is No. To achieve optimal multivariable control, all possible pressures are
In order to determine the optimum condition from the rolled state, rolling phenomenon
Mathematical model that represents all feasible states of the controller and controller
(State equation) is required. Here, since the multivariable control theory is linear, the mathematical model
Also needs to be expressed linearly. For the subject of the present invention, the modeling is as follows. Modeling of control devices (roll reduction system, roll rotation system)
(Step 1): (mill drive system model) When the roll reduction control system is approximated by a first-order lag system, Δs_i ^*(S) = Δs_{ref, i} ^*(S) / (1 + T ・ s) i: Stand No., T_s: Time constant where s: corresponds to time t of continuous time system by Laplace transform.
If the variable s is set as d / dt and expressed in a continuous time system, dΔs_i ^*(T) / dt = -Δs_i ^*(T) / T_s + Δs_{ref, i} ^*(T) / T_s When expressed in a discrete time system discretized with a control period Δt,
At time k, Δs_i ^*(K + 1) = a_siΔs_i ^*(K) + b_siΔs
_{ref, i} ^*(K) (1) where k in () is a positive integer in the discrete time system.
Time, k + 1 is the time after the control cycle Δt from k. Well
A_si, b_siIs the coefficient determined by the time constant T and discretization
Represents. Next, when the roll rotation control system is approximated by a quadratic system, ΔN_i ^*(S) / ΔN_{ref, i} ^*(S) = ω_ni ²/ (1 + 2ξω_nis + ω_ni ²) Where ω_n: Each velocity, ξ: Damping coefficient When expressed in discrete time system, Δ at time k_i ^*(K + 1) = a_NiΔN_i ^*(K) + b_NiΔ_i ^*(K)
+ C_NiΔN_{ref, i} ^*(K) (2) where a_Ni, b_Ni, c_NiIs each velocity ω_n, Damping coefficient ξ and separation
Represents the coefficient determined by the dispersion. Modeling the dynamics of rolling phenomena and the overall state of the system.
Equation (Step 2): The wire rolling phenomenon depends on the material on the inlet side, temperature change (disturbance), and reduction operation.
Due to the dynamic changes in the roll rotation operation
The inter-tand tension and the height and width of the material on the outlet side dynamically change.
Are transmitted to the downstream stand after the material transfer time
It is a complex dynamic phenomenon. The dynamic phenomena of rolling at each stand are analyzed by
The non-linear continuous rolling model consisting of degrees
Around the table) and make it dimensionless (dimensional only in time)
Die of tension (state vector) derived by
Namic change (in continuous time system, at time t
Time derivative of force (state vector), in discrete time system,
The tension (state vector value) at time k + 1 is
In continuous time system, at time t, in discrete time system, time
State vector, input vector, disturbance vector at time k
Dynamic model and output expressed using toll values
Vector, state vector, static of disturbance vector
State of rolling consisting of output equations expressing various relationships
It is expressed by the equation (linear rolling model). For details of the linear rolling model, please refer to
Spring Lecture p533-536 "Equation of state of wire rod rolling")
It is listed. It is discretized with the control period Δt, and the above equations (1) and (2) are used.
When combined with the mill drive system of, the equation of state of the entire system is
Is obtained as. x (k + 1) = A'x (k) + B'u (k) + E'w
(K) (3) y (k) = C′x (k) + F′w (k) (4) where state vector x, output vector y, input vector u,
The disturbance vector w is as follows. x (k) = [Δs_i ^*(K), Δt_fj ^*(K), ΔN
_j ^*(K), Δ_j ^*(K)]^T (5) y (k) = [Δh_n ^*(K), Δb_n ^*(K), Δt
_fj ^*(K)]^T (6) u (k) = [Δs_{ref, i} ^*(K), ΔN_{ref, j} ^*(K)]
^T (7) w (k) = [ΔT_mi ^*(K), ΔH_i ^*(K), ΔB
_i ^*(K)]^T (8) Here, the subscripts i = 1, ..., N, j = 1, ..., N−1, n: the number of stands. Coefficient matrices A ′ to F ′ are partial derivatives of rolling characteristics and drive system characteristics.
This is a matrix whose elements are coefficients. Furthermore, the material transfer from No, i stand to No.i + 1 stand
Sending time t_diThe approximate number of steps of l_i(L_iIs t_di≒ l_i・ Δt and
0 or a positive integer such that
The material transfer delay formula is ΔT_{mi + i} ^*(K) = α_iΔT_mi ^*(K-l_i) (9) ΔH_{i + i} ^*(K) = Δb_i ^*(K-l_i) (10) ΔB_{i + i} ^*(K) = Δh_i ^*(K-l_i) (11) It is given by i = 1,2, ..., n-1. Where the coefficient α_iRepresents the temperature change of the material
Computer simulation using a nonlinear rolling model with a coefficient
It is calculated by the rate. If there is a loop between the stands as shown in Fig. 1,
1) Bending moment and bending acting on loop
From the relationship of the ratio and 2) Hooke's law, the tension t_fiAnd loop
Length l_siThe relationship is t_fi−t_foi= C_1i(L_si-L_soi) / L_soi (12) where subscript 0: reference state, C_1i: Determined by rolling condition
Is a constant. Loop height y_hiAnd loop length l_siRelationship is geometrically determined
(Y_hi−y_hoi) / Y_hoi= C_2i(L_si-L_soi) / L_soi (13) where C_2i: Constants determined by rolling conditions (12), (1
From formula 3) loop height Δy_hj ^*And tension Δt_fj ^*The function of is Δy_hj ^*(K) = C_2i/ C_ij× k_fmoj・ Δt_fj ^*(K) (14) Modeling during control system design (state equation during control system design)
(Step 3): Original state method expressing dynamic phenomena of rolling and mill drive system
Control appropriate variables according to mill type and control method from equation
Used when designing a control system that is configured by selecting the amount of operation and the amount of operation
The state equation is called the state equation when designing the control system. In this control system, the material height and tension on the outlet side of each stand are adjusted.
The corresponding loop height is the control amount (output vector). Therefore, the tension Δt in the systems (3) to (8) is_fj ^*To (1
Loop height Δy_hj ^*Replaced with
In addition, the output vector is calculated from the lock-on value of the output material height.
Of the loop height and the deviation of the loop height from the reference value.
The gap between the radius and its target value by the deviation from the lock-on value
And the equation of state when designing the control system is given by equations (15) to (20).
Growls. (K + 1) = A (k) + B (k) + E (k)
 (15) (k) = C (k) + F (k) (16) where, state vector, output vector, input vector
The disturbance vector is as follows. (K) = [Δ_i ^*(K), Δy_hj ^*(K), ΔN
_j ^*(K), Δ_j ^*(K)]^T (17) (k) = [Δ_n ^*(K), Δy_hj ^*]^T (18) (k) = [Δ_{ref, i} ^*(K), ΔN_{ref, j} ^*(K)]
^T (19) (k) = [ΔT_mi ^*(K), ΔH_i ^*(K), ΔB
_i ^*(K)]^T(20) where the subscripts i = 1, ..., N, j = 1, ..., N−1, n: Number of stands. For other control methods, the processing method is the same, but each vector
The variables that make up are different. For example, the
When only the loop height is the controlled variable, the manipulated variable is the roll speed.
Control is possible only with the target value of the number of turns. Therefore, (1
From the state vector of Eq. (7), the deviation of the measured value of the roll gap is
Difference, deviation of the material height on the output side from the output vector of Eq. (18),
The deviation of the roll gap target value from the input vector of equation (19)
Exclude the equation of state when designing the control system. Control system design (step 4): The control system design procedure is described below. Evaluation using equations of state (15) to (20) when designing the control system
Optimal input vector u that minimizes the valence function_optOut (k)
Realized by feedback of force vector and partial state vector
Derive the method. Below is a description of a discrete-time system that is easy to adjust in the actual process.
The multivariable control theory is explained. Evaluation function for state equations (15) and (16)Consider optimal control that minimizes. Where is
The deviation is the difference between the inputs and is given by the following equation. (K) = (k) − (k) = − (k) (22) (∵ (k) = 0, target value) (k) = (k) − (k−1) (23) Also, Q, Assume that R is a weight matrix and (k) is a step disturbance
Do. Control input is control deviation, partial state vector_sThe fi
Feedback (k) =-F₁(K-1) -F₂[(K)-(k-
1)]-F₃[_s(K)-_s(K-1)] + (k-
1) It is realized by (24). Partial state_sIsAt this time, the optimum gain F₁, F₂, F₃Is F = [F₁｜ F₂｜ F₃] = (R + Γ^TSΓ)^-1Γ^TSΦ (26) S = H^TQH + Φ^TSΦ-Φ^TS Γ (R + Γ^TSΓ)^-1・ Γ^T
It is given by SΦ (27). Here, Φ, Γ, H are error equations of the state equations (15) and (16).
Stem_e (K + 1) = Φ_e(K) + Γ (k) (28) (k) = H_e(K) (29)Here is the coefficient matrix of. The deviation of the material height on the output side of Eq. (17) does not include monitor AGC compensation,
It is calculated as follows for each case. (1) Without monitor AGC compensation Gauge meter type Δ_i ^*= Δ_i ^*+ Α_i・ ΔP_i ^*/ M_i ^*(I = 1 to 4 or 3,4)
(31a) (2) With monitor AGC compensation Gauge meter type Δ_i ^*= Δ_i ^*+ Α_i・ ΔP_i ^*/ M_i ^*+ Δh_{M, i} ^*(I = 1 to 1
4or3,4) (31b) where M_i ^*Is the dimensionless Mill constant (M_i ^*= H_oi・ M_i・ P_oi),
α_iIs the scale factor. The first and second terms on the right side of the equations (31a) and (31b) are known
It is a pedometer type. Δh of the third term on the right side of equation (31b)_{M, i} ^*Is the dimension on the delivery side of the rolling mill
The deviation from the target value of the measured value of No.3, No.4 stand
Effect of material height on deviation of material size on the delivery side of rolling mill
Use the coefficient to determine the deviation of the material height of the No. 3 and No. 4 stand outlets.
Monitor AGC obtained by converting, multiplying by gain, and integrating
This is a compensation term. Monitor AGC compensation Δh_{M, i} ^*Is obtained by the following procedure. Height of exit side of finishing stand of dimension meter 13 Δh_Four ^*， Outside width Δh_Four ^*
From the sample value of, the reader stand (second from the last)
And finishing stand (No.3, No.4 stamp in this embodiment)
De) Outer material height monitor correction amount Δh_{x, 3} ^*, Δh_{x, 4} ^*Δh_{x, 3} ^*= (K₂・ Δh_Four ^*−Δb_Four ^*) / K₁ (32) Δh_{x, 4} ^*= Δh_Four ^* Calculate with (33). Where k₁, k₂Is the influence coefficient k₁= (∂Δb_Four ^*/ ∂Δs₃ ^*) / (∂Δh₃ ^*/ ∂Δs₃ ^*(34) k₂= (∂Δh_Four ^*/ ∂Δs_Four ^*) / (∂Δh_Four ^*/ ∂Δs_Four ^*It is given by (35). Next, monitor AGC compensation Δh_{M, i} ^*(K) is a leader stun
De (i = 3), finishing stand (i = 4) Outlet side material height
Monitor correction amount Δh_{x, 3} ^*(K), Δh_{x, 4} ^*Integrate (k)Is calculated by. However, T_iGoes out from the No.i stand
Material transfer time to size meter, K_{M, i}Is the monitor integral gain
is there. Determination of optimum control gain (step 5): Weighting matrices Q and R of equations (26) and (27) and moni of equation (36)
Compensation integral gain K_{M, i}Repeats the simulation
It was decided by and. Here, the weight matrix Q by which the deviation of the control amount is multiplied is a multivariable
Of the control variables that require high response and small deviation.
Increase the size of the element of Q by which it is multiplied. Well
In addition, the weight matrix R by which the deviation of the manipulated variable is multiplied is a multivariable operation.
In the amount of work, the amount of control that is controlled by the operation amount is required.
The higher the response speed is, the more important R
Reduce the size of the prime. However, the weights Q and R are multiplied by the same physical quantity.
Among the elements, the magnitude relation of the elements is controlled by the corresponding control.
The magnitude response speed or the manipulated variable change speed.
However, it does not correspond between different physical quantities. At this time, the optimum input u_opt(K) is_opt (K) =-F₁(K-1) -F₂[(K)-
(K-1)]-F₃[_s(K)-_s(K-1)] +
_opt(K-1) It is realized by (24). In addition, F1, F2, and F3 in equation (24) are the optimum feedback gains.
Process computer during online processing
It is embedded in and memorized. Next, the online calculation process is explained. This process is executed by the process computer. Obtained by offline processing and stored in the process computer
Optimal feedback gain and online measurement
The optimal control using the state vector and output vector
Calculate the input and output the value to the controller as the manipulated variable
Online processing of known feedback multivariable control
Perform online control in the same way as. This publicly-known online process is performed, for example, in the well-known document: Showa.
60th Plastic Working Spring Lecture p101-104
Variable control (2) (Study on rolling characteristics of wire rod 9
Report) ”(see especially Chapter 3 and Table 3)
It is a thing. [Example] Fig. 1 shows the structure of a loop between stands. Fig. 2 shows the target 4-stand mill and the control system of the present invention.
System. The No. 1 star mill with the 4-stand mill under the rolling conditions shown in Table 2
Input side material average temperature, input side material height / width is ΔT_m1 ^*=-
0.028, ΔH₁ ^*= ΔB₁ ^*= 0.007 changed in steps
We examined the case. (1) Example without monitor AGC compensation The lock-on value of the gauge meter matches the standard state.
Case: Figures 4 to 6 show the stand outlet materials without control.
Fee height Δh_i ^*, Outer material width Δb_i ^*， Loop height Δy_hj ^*Su
It is an illustration of a step response. Here, material temperature and material dimensional fluctuations are No. 1 stand
The arrival time is t = 3 seconds. 7 to 11 are based on loop control + 4-stand AGC
Height of each stand output side with control Δh_i ^*, Output width Δb
_i ^*， Loop height Δy_hj ^*, Roll gap Δs_i ^*And roll
Rotation speed ΔN_i ^*FIG. 6 is an exemplary diagram of a step response of. At this time, monitor compensation is not performed and Δh_{M, i} ^*= 0
I said. The weight matrix Q, R is set to Q = diag (1,1,1,1,0.001,0,002,0.005) (37) R = diag (10,10,10,10,1,1.2,1.5) (38) ． Where diag represents the diagonal elements of the matrix. At the time of settling with short loop height and exit stand height / width
It can be seen that the offset is controlled to 0 between. (2) Example without monitor AGC compensation The lock-on value of the gauge meter deviates from the reference state.
Case: Material temperature and dimensional disturbance is ΔT from the material tip_mi ^*= -0.028,
ΔH_i ^*= ΔB_i ^*Consider the case where only = 0.007 is added
The Figures 12 to 14 show loop control without monitor compensation +
Outer height of each stand by all 4 stands AGC Δh_i ^*， Outside
Width Δb_i ^*， Loop height Δy_hj ^*FIG. The lock-on value of the gauge meter deviates from the standard state.
Therefore, there is an offset in the height and width of the material on the exit side of the finishing stand.
Has occurred. Here, the material tip reaches the No. 1 stand.
The time at which t was set is t = 0.2 seconds. (3) Example with monitor AGC compensation The lock-on value of the gauge meter deviates from the reference state.
Case: Material temperature and dimensional disturbance is ΔT from the material tip_m1 ^*= -0.028,
ΔH₁ ^*= ΔB₁ ^*Consider the case where only = 0.007 is added
The Figures 15 to 17 show loop control with monitor compensation +
Outer height of each stand by all 4 stands AGC Δh_i ^*， Outside
Width Δb_i ^*， Loop height Δy_hj ^*Here is an example of the response of. this
, The monitor compensation integral gain is K_{M, 3}= 0.025, K_{M, 4}= 0.01
I set it to 0. Finishing stand Outside material height and width offset
It can be seen that the dot is controlled to 0. In this embodiment, a side loop (material
Showing continuous rolling with the weight of the material can be ignored)
However, there is an up-loop or down-loop between the stands.
Even in some cases, the loop height considering the weight of the material
And the tension acting on the rolled material
If configured, dimension control is possible in the same way as in the above embodiment
Is. In addition, loopless rolling with no loops between stands
Then, instead of the loop height, the tension is controlled as a controlled variable.
The size can be controlled by the method. In this embodiment, the wire rod and rod 2 shown in FIG.
The dimensional control method in the mill is shown, but one stand
3 roll mill with 4 or 3 rolls each, 4 rolls
Dimension control for roll mills is also a coefficient matrix of equation of state
Substituting the values of AF for the values of the mill, the method of the invention
The target value can be controlled with high response and high accuracy. In addition, the dimensions of loopless rolling of shaped steel mills and pipe mills
Also in legal control, tension is used as a controlled variable instead of loop height,
Replace the values of the coefficient matrices A to F of the equation of state with the values of the mill
Then, the method of the present invention can control the target value with high response and high accuracy.
It is possible. The same applies to the dimension control of hot strip mills and cold strip mills.
The values of the coefficient matrices A to F of the equation of state are placed in the mill values.
In other words, according to the method of the present invention, the target value can be set with high response and high accuracy.
It is controllable. In this example, the evaluation function used when designing the controller is used.
Although the quadratic form was used as the form of the number,
Defined and feedback controlled multivariable system in the same way.
If the control system is configured and the optimum gain is calculated, it is the same as this embodiment.
Like this, online dimension control is possible. [Effect of the Invention] As described above, in the case of this embodiment according to the present invention,
In this case, the control based on the multivariable control theory is controlled as follows.
It was revealed that the performance was excellent. (1) Loop height control and gauge gauge for each stand
The settling time is short and the offset height is limited to 0.
It is possible. (2) Offset material height and width of finishing stand
It can be controlled to 0.

[Brief description of drawings]

Figure 1 shows the structure of the loop between stands. 2 (a) and 2 (b) are configuration diagrams showing an embodiment of the present invention. FIG. 3 is a diagram showing a flowchart composed of off-line calculation processing and online control showing from the modeling of the present invention to the determination of the optimum feedback control gain based on the multivariable control theory. FIGS. 4 to 6 show the outlet material height Δ without control.
It is an illustration of the step response of h _i ^* , exit material width Δb _i ^* , loop height Δy _hj ^* . Figures 7 to 11 show loop height control + 4-stand AGC.
Example of step response of output material height Δh _i ^* , output material width Δb _i ^* , loop height Δy _hj ^* , roll gap Δs _i ^* and roll rotation speed ΔN _i ^* when (without monitor compensation) Is. 12 to 14 show loop height control when the lock-on value of gauge meter calculation deviates from the standard state + 4 stands in total
It is an illustration of step response of outgoing material height Δh _i ^* , outgoing material width Δb _i ^* , loop height Δy _hj ^* by AGC (without monitor compensation). Figures 15 to 17 show loop height control when the lock-on value of gauge meter calculation deviates from the standard state + all 4 stands
It is an illustration of step response of _output material height Δh _i ^* , output material width Δb _i ^* , loop height Δy _hj ^* by AGC (with monitor compensation). The numbers in Figure 1 are as follows. 1: No.i stand roll, 2: No.i + 1 stand roll, 3, 4:
Guide roll, 5: kicking roll, 6: loop height detector,
7: Pass line, 8: Material The numbers in Fig. 2 are as follows. 9,11: Horizontal roll, 10, 12: Vertical roll, 13: Output side dimension gauge, 14:
Control Calculator, 15: Material

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Harutoshi Ogai Harutoshi Ogai 1-1-1, Edamitsu, Hachimanto-ku, Kitakyushu-shi, Fukuoka Inside the 3rd Technical Research Laboratories, Nippon Steel Corporation (72) Naoki Tanaka 12 Nakamachi, Muroran-shi, Hokkaido Address Nippon Steel Co., Ltd. Muroran Works (72) Inventor Hiroshi Hagiwara 23-15, Suzuko-cho, Kamaishi City, Iwate Prefecture Kanaishi Works, Nippon Steel Co., Ltd. (56) Reference JP 62-244508 JP, A) JP-A-62-244510 (JP, A)

Claims (2)

[Claims] 1. When rolling a wire rod or bar with a mill consisting of a multi-stage stand, a mill drive system linear model of the roll gap and roll speed is created, and the measured values and reference of material dimensions and temperature at the entrance side of each stand. The deviation (disturbance vector) is calculated from the value, and the deviation of the roll gap of each stand from the reference value, the deviation of the measured loop height from the reference value, and the deviation from the reference value of the roll rotation speed Vector), and the control amount (output vector) from the deviation of the material height on the delivery side obtained from the deviation of the roll gap of each stand and the deviation of the actual measured value of the rolling load from the reference value and the deviation of the loop height between the stands. ), The operation amount (input vector) is constructed from the deviation of the roll gap of each stand and the target value of the roll speed and the reference value, and the dynamic phenomenon of continuous rolling is defined by the disturbance vector and the state vector. It is represented by a state equation using the control vector, output vector, and input vector, and it is combined with the mill drive system model to create a state equation for the entire system, and the control variables and manipulated variables are selected according to the mill type and control method. Feedback multivariable control that minimizes the value of the evaluation function by constructing the state equation of the entire system used in designing the control system and introducing an evaluation function consisting of an output vector that represents the controlled variable and an input vector that represents the manipulated variable. A dimension control method characterized in that the manipulated variable (input vector) of the system is obtained and the dimension of the rolled material released from the final stand of the mill is controlled to a target value. Here, the mill drive system model is a model that represents the relationship between the time delay of the measured value and the target value of the control device (roll reduction system, roll rotation system). 2. When rolling a wire rod or a bar rod by a mill consisting of multi-stage stands, a mill drive system wire rod model of roll gap and roll speed is created, and the measured value and standard of material size and temperature at the entrance side of each stand. The deviation (disturbance vector) is calculated from the value, and the deviation of the roll gap of each stand from the reference value, the deviation of the measured loop height from the reference value, and the deviation from the reference value of the roll rotation speed Vector) and the deviation of the material height on the delivery side of each stand from the target value of the measured value of the material size on the delivery side of the rolling mill, It is converted to the deviation of the material height on the outlet side of each stand by the calculation formula or the measured value that expresses the relationship of the deviation of the material dimension on the side, and the added amount is taken as the deviation of the material height on the outlet side and the loop of each of the stands. Controlled by height deviation Quantity (output vector), and the operation amount (input vector) is constructed from the deviation of the roll gap of each stand and the target value of roll speed and the reference value, and the dynamic phenomenon of continuous rolling is defined by the disturbance vector and state. It is expressed by a state equation using a vector, an output vector, and an input vector, and by combining it with a mill drive system model, a state equation of the entire system is created, and a control amount and an operation amount according to the mill type and control method are selected. Feedback multivariable control that minimizes the value of the evaluation function by constructing the state equation of the entire system used in designing the control system and introducing an evaluation function consisting of an output vector that represents the controlled variable and an input vector that represents the manipulated variable. A dimension control method characterized in that the manipulated variable (input vector) of the system is obtained and the dimension of the rolled material released from the final stand of the mill is controlled to a target value. Here, the mill drive system model is a model that represents the relationship between the time delay of the measured value and the target value of the control device (roll reduction system, roll rotation system).