JP2013246560A - Process control system, process control method and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To flexibly set an appropriate target value to a manufacturing process as much as possible without re-calculating optimization calculation even in where an operation condition is changed.SOLUTION: A plurality of evaluation amounts to be an index of evaluating the performance of an air heating furnace 400 are defined as an object function ΔP, a multipurpose optimization problem with respect to the object function ΔP is solved by multipurpose GA, and a plurality of Pareto optimal solutions are preliminarily obtained in off-line and stored. Then, according to the change of an operation condition, a Pareto optimal solution is changed into any of the plurality of stored Pareto optimal solutions, a target value pattern corresponding to the changed Pareto optimal solution is set to a controller 300.

Description

  The present invention relates to a process control system, a process control method, and a computer program, and is particularly suitable for use in changing a target value in a process control system.

  For example, in a manufacturing process such as a steelmaking process, a steelmaking process, and a rolling process in the steel industry, feedback control is often performed so that the controlled variable to be controlled becomes a target value. In such a process control system, a target value has been conventionally set by solving an optimization problem for optimizing an objective function.

  Such a manufacturing process in the steel industry or the like may include complicated reactions and processing steps, and there are a plurality of objective functions to be optimized in the optimization problem of the manufacturing process. Therefore, the multi-objective optimization problem is solved. Conventionally, in this kind of multi-objective optimization problem, the so-called weighting coefficient method is used to convert the multi-objective optimization problem into a single objective optimization problem by making the load sum of multiple objective functions a new single objective function. (See Non-Patent Document 1). In the single objective optimization problem, for example, the process control parameters are adjusted so as to minimize (or maximize) the single objective function while satisfying a plurality of constraints in the manufacturing process ( (See Patent Document 1).

JP 2002-373002 A

Obayashi Shigeru, "Multipurpose optimization and information visualization data mining", Toyota Research Report No. 58, May 2005, p. 109-116 Japan Iron and Steel Institute Joint Study Group, Rolling Theory Division, “Theory and Practice of Sheet Rolling”, Japan Iron and Steel Institute, September 1984, p. 287-291

  However, in the manufacturing process in the steel industry, the operation conditions (process state and operation policy set by the operator) change from moment to moment, and the weighting coefficient is changed each time. Therefore, in the conventional solution of the multi-objective optimization problem as described above, it is not possible to know how to adjust the target value when the operation condition is changed. For this reason, it is necessary to redo the optimization calculation.

  The present invention has been made in view of such a problem, and even when the operation condition is changed, an appropriate target value can be set in the manufacturing process flexibly and as much as possible without performing the optimization calculation again. The purpose is to do so.

  The process control system of the present invention controls a target value optimization device that sets a target value for controlling the operation of the plant, and controls the operation of the plant so that the target value is set by the target value optimization device. To perform a computer simulation of the operation of the plant and the operation of the controller so that the target value set by the controller and the target value optimization device is obtained, and as a result, evaluate the performance of the plant A process simulator that outputs a plurality of evaluation value values that are indicators of the target value optimization device to the target value optimization device, wherein the target value optimization device outputs the plurality of values output from the process simulator Deriving multiple Pareto optimal solutions by solving the optimization problem with the multiple evaluation values as objective functions using the evaluation value A rate optimum solution deriving unit, a plurality of Pareto optimum solutions derived by the Pareto optimum solution deriving unit, and a Pareto optimum solution storage unit that associates and stores a target value corresponding to the Pareto optimum solution in a storage medium The operating condition change determining means for determining whether or not the operating condition including at least one of the state quantity of the plant and the operation policy of the plant has been changed, and the operating condition is changed by the operating condition change determining means. A Pareto optimal solution selecting means for selecting any one of a plurality of Pareto optimal solutions stored by the Pareto optimal solution storage means according to the changed operating condition, and the Pareto optimal solution A target value stored in the Pareto optimal solution storage unit in association with the Pareto optimal solution selected by the selection unit is stored in the controller. And having a target value output means for force, the.

  The process control method of the present invention includes a target value optimization step for setting a target value for controlling the operation of the plant, and the operation of the plant so as to be a target value set by the target value optimization step. A computer simulation of the operation of the plant and the operation of the controller is performed so that the target value set by the control step controlled by the controller and the target value optimization step is obtained. As a result, the plant And a simulation control step of deriving a plurality of evaluation value values that are indices for evaluating the performance of the target, wherein the target value optimization step includes the plurality of evaluations derived by the simulation step. By using the value of the quantity, a plurality of parameters can be obtained by solving an optimization problem with the plurality of evaluation quantities as objective functions. A Pareto optimal solution deriving step for deriving an optimal solution, a plurality of Pareto optimal solutions derived by the Pareto optimal solution deriving step, and a target value corresponding to the Pareto optimal solution are associated with each other and stored in a storage medium The optimum solution storage step, the operation condition change determination step for determining whether or not the operation condition including at least one of the plant state quantity and the operation policy of the plant has been changed, and the operation condition change determination step If it is determined that the operation condition has been changed, a Pareto optimal solution selection step of selecting any one of a plurality of Pareto optimal solutions stored in the Pareto optimal solution storage step according to the changed operation conditions; , The Pareto associated with the Pareto optimal solution selected by the Pareto optimal solution selection step And having a target value output step of outputting to the controller a stored desired value by Tekikai storing step.

  A computer program according to the present invention causes a computer to function as each means of the process control system.

  According to the present invention, a plurality of Pareto optimal solutions are derived by solving an optimization problem having a plurality of evaluation quantities, which are indices for evaluating the performance of the plant, as objective functions. Then, the plurality of Pareto optimal solutions and target values corresponding to the Pareto optimal solutions are associated with each other and stored in the storage medium. Thereafter, when the operation condition is changed, one of the stored Pareto optimal solutions is selected again according to the changed operation condition, and is stored in association with the selected Pareto optimization. The target value is output to the controller. Therefore, even when the operating conditions are changed, an appropriate target value can be set in the manufacturing process flexibly and as much as possible without re-execution of the optimization calculation.

It is a figure which shows the 1st Embodiment of this invention and shows an example of a structure of a process control system. It is a figure which shows the 1st Embodiment of this invention and shows an example of schematic structure of a hot stove. 1 is a block diagram illustrating an example of a functional configuration of an information processing device according to a first embodiment of this invention. It is a figure which shows the 1st Embodiment of this invention and shows an example of a target value pattern. It is a figure which shows the 1st Embodiment of this invention and shows an example of a structure of an individual | organism | solid. It is a figure which shows the 1st Embodiment of this invention and shows an example of the Pareto optimal solution. It is a figure which shows the 1st Embodiment of this invention and shows an example of GUI for making an operator change a Pareto optimal solution. It is a flowchart which shows the 1st Embodiment of this invention and demonstrates an example of operation | movement of the target value optimization apparatus at the time of performing the Pareto optimal solution derivation process. It is a flowchart which shows the 1st Embodiment of this invention and demonstrates an example of operation | movement of the target value optimization apparatus at the time of performing target value change process. It is a figure which shows the 1st Embodiment of this invention and illustrates the modification of the method of selecting the Pareto optimal solution notionally. It is a figure which shows the 2nd Embodiment of this invention and shows an example of schematic structure of a hot rolling installation. It is a block diagram which shows the 2nd Embodiment of this invention and shows an example of a functional structure of information processing apparatus. It is a figure which shows the 2nd Embodiment of this invention and shows an example of a structure of an individual | organism | solid.

[First Embodiment]
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. In this embodiment, the case where the plant to be controlled is a hot blast furnace attached to a blast furnace in the steel industry will be described as an example.
FIG. 1 is a diagram illustrating an example of a configuration of a process control system according to the present embodiment.
In FIG. 1, the process control system 10 includes an information processing apparatus 100, a performance database 200, a controller 300, a hot stove 400, and a subtracter 500.

First, a schematic configuration of the hot stove 400 that is a control target of the present embodiment will be described. FIG. 2 is a diagram illustrating an example of a schematic configuration of the hot stove 400.
In FIG. 2, a hot stove 400 is a heat storage heat exchanger for supplying hot air to a blast furnace (not shown), a heat storage chamber 401 that gives heat to the blast furnace, and a combustion for heating the heat storage chamber 401. Chamber 402.

  In the combustion chamber 402, a mixed gas (fuel gas) of BFG gas and COG gas blown from the gas supply duct 403 and auxiliary combustion air blown from the auxiliary combustion air supply duct 404 are burned by the combustion burner 405, and this combustion gas is burned. Heat is stored by passing between the heat storage bricks stacked inside the heat storage chamber 401 and heating. A thermometer 413 for measuring the dome temperature is attached to the upper part of the heat storage chamber 401.

In the example shown in FIG. 2, as this heat storage brick, a high alumina brick 406 and a silica brick 407 mainly composed of silica are laminated in order from the bottom, and the high alumina brick 406 and the quartz brick 407 are laminated. A plurality of passage openings extending in the vertical direction are formed. Further, a thermometer 406 a for measuring the temperature of the high alumina brick 406 is attached to the high alumina brick 406, and a thermometer 407 a for measuring the temperature of the quartz brick 407 is attached to the silica brick 407. In detail, heat storage bricks generally have a three-layer structure of clay bricks, high alumina bricks, and quartz bricks in order from the bottom, but for the sake of simplicity, the hot stove is of the above structure. It will be explained as being.
The gas supply duct 403 is provided with a fuel gas flow rate adjustment valve 403a, and the amount of fuel gas flowing into the combustion chamber 402 can be adjusted by opening and closing the fuel gas flow rate adjustment valve 403a.

A portion of the gas supply duct 403 on the upstream side (right side in FIG. 2) from the fuel gas flow rate adjustment valve 403a branches into a BFG gas supply duct 408 and a COG gas supply duct 409.
The BFG gas supply duct 408 is connected to a blast furnace (not shown), and blows the BFG gas generated in the blast furnace to the hot stove 400.

  The BFG gas supply duct 408 is provided with a BFG gas flow rate adjustment valve 408a and a BFG gas flow meter 408b. By opening and closing the BFG gas flow rate adjustment valve 408a, the amount of BFG gas flowing into the hot stove 400 can be adjusted. Moreover, the gas inflow amount of the BFG gas flowing into the hot stove 400 can be monitored based on the measurement result of the BFG gas flow meter 408b.

The COG gas supply duct 409 is normally connected to a coke oven (not shown), and blows COG gas generated in the coke oven to the hot stove 400.
The COG gas supply duct 409 is provided with a COG gas flow rate adjustment valve 409a and a COG gas flow meter 409b. By opening and closing the COG gas flow rate adjustment valve 409a, the gas inflow amount of the COG gas flowing into the hot stove 400 can be adjusted. Moreover, the gas inflow amount of the COG gas flowing into the hot stove 400 can be monitored based on the measurement result of the COG gas flow meter 409b.

  The auxiliary combustion air supply duct 404 is connected to an oxygen generation plant (not shown) that generates oxygen, and blows air generated in the oxygen generation plant to the hot stove 400. The oxygen generation plant is also connected to a blast furnace (not shown), and the oxygen concentration of the auxiliary combustion air supplied to the hot stove 400 varies depending on the operation status of the blast furnace and the like. Further, the oxygen concentration of the auxiliary combustion air blown to the hot stove 400 can be adjusted to an arbitrary value even when the operation status of the blast furnace does not change.

  The auxiliary combustion air supply duct 404 is provided with an auxiliary combustion air flow rate adjustment valve 404a and an oxygen concentration measuring device 404b. By opening and closing the auxiliary combustion air flow rate adjustment valve 404a, the inflow amount of auxiliary combustion air flowing into the combustion chamber 402 can be adjusted. Moreover, the oxygen concentration of the auxiliary combustion air flowing into the hot stove 400 can be monitored based on the measurement result of the oxygen concentration measuring device 404b.

A gas discharge duct 410 for discharging combustion gas containing N ₂ , CO _{2 and the} like is provided at the lower end of the heat storage chamber 401.
The gas discharge duct 410 is provided with a gas discharge amount adjustment valve 410a, an exhaust gas flow meter 410b, and an exhaust gas thermometer 410c. By opening and closing the gas discharge amount adjustment valve 410a, it is possible to adjust the discharge amount of the combustion gas discharged from the gas discharge duct 410. Further, the amount of combustion gas discharged from the gas discharge duct 410, that is, the amount of exhaust gas can be monitored based on the measurement result of the exhaust gas flow meter 410b. In the present embodiment, the exhaust gas flow meter 410b can measure the amount of CO ₂ gas discharged to the gas discharge duct 410. Further, the temperature of the combustion gas, that is, the exhaust gas temperature can be monitored based on the measurement result of the exhaust gas thermometer 410c.

A room temperature air introduction duct 411 is connected to a lower end portion of the heat storage chamber 401 at a position different from the gas discharge duct 410, and room temperature air flows into the heat storage chamber 401 through the room temperature air introduction duct 411.
The room temperature air introduction duct 411 is provided with an air inflow control valve 411a. By opening and closing the air inflow control valve 411a, the amount of room temperature air flowing into the hot stove 400 can be adjusted.
The combustion chamber 402 is connected to a hot air discharge duct 412 for discharging hot air for the blast furnace. The hot air discharge duct 412 is provided with a hot air flow rate adjustment valve 412a. By opening and closing the hot air flow rate adjustment valve 412a, the flow rate of the hot air blown into the blast furnace can be adjusted.

When heat is stored in the heat storage chamber 411, the air inflow control valve 411a and the hot air flow rate control valve 412a are completely closed, and the fuel gas and the auxiliary combustion are supplied into the combustion chamber 402 through the gas supply duct 403 and the auxiliary combustion air supply duct 404. Inflow air.
The fuel gas and the auxiliary combustion air are burned by the burner 405, and the combustion gas stores the high alumina brick 406 and the quartz brick 407 through the openings formed in the high alumina brick 406 and the quartz brick 407 of the heat storage chamber 401. To do.

  When the heat storage in the heat storage chamber 401 is completed, the fuel gas flow rate adjustment valve 403a, the auxiliary combustion air flow rate adjustment valve 404a, and the gas discharge amount adjustment valve 410a are completely closed, and the room temperature is supplied to the heat storage chamber 11 via the room temperature air introduction duct 411. Allow air to flow in. The room temperature air that has flowed into the heat storage chamber 401 passes through openings formed in the high alumina brick 406 and the quartz brick 407 and is heated to 900 to 1300 ° C., and then is discharged from the hot air discharge duct 412 as hot air for the blast furnace. The

Returning to the description of FIG. 1, in the present embodiment, the hot blast furnace 400 as described above is controlled, the dome temperature U1 [° C.], the C / B ratio (ratio between the flow rate of COG and the flow rate of BFG) U2 [−. ] And the BFG flow rate U3 [kNm ³ / h] as control amounts will be described as examples. Therefore, in the present embodiment, the controller 300 causes the measured values (that is, the actual values) of the dome temperature U1, the C / B ratio U2, and the BFG flow rate U3 to be the target values set by the information processing apparatus 100. The hot stove 400 is controlled. The controller 300 can control the hot stove 400 by various known control methods such as PID control.

In addition, the controller 300 stores data on operation results in the hot stove 400 in the result database 200. In the present embodiment, the performance database 200 stores, for example, data of operation results for the latest five days for every 1 [min].
The operation result data includes, for example, output values (operation result values) output from various measuring devices installed in the hot stove 400 by operating the hot stove 400 and when the output values are obtained. The input value and target value for the hot stove 400 are included.

  As described above, the information processing apparatus 100 is for setting a target value in the process control system, and includes the target value optimizing apparatus 110 and the process simulator 150. The information processing apparatus 100 can be realized, for example, by using a measuring instrument provided in the hot stove 400 or a personal computer having an interface with the controller 300.

FIG. 3 is a block diagram illustrating an example of a functional configuration of the information processing apparatus 100.
The information processing apparatus 100 solves a multi-objective optimization problem using a multi-objective GA (genetic algorithm) and obtains a plurality of Pareto optimal solutions in advance. Thereafter, when the operation condition is changed, one Pareto optimal solution that matches the changed operation condition is selected from the plurality of Pareto optimal solutions, and a target value corresponding to the selected Pareto optimal solution is given to the controller 300. Output.

  Here, the operation conditions include a state quantity of the hot stove 400 and an operation policy of the hot stove 400. The state quantity of the hot stove 400 is an amount (for example, temperature or air flow rate) measured directly or indirectly momentarily by operating the hot stove 400. On the other hand, the operation policy of the hot stove 400 is determined by the operator based on the production plan and production management.

<Configuration of Target Value Optimization Device 110 and Process Simulator 150>
(Objective function setting unit 111)
The objective function setting unit 111 sets the objective function ΔP based on an operation from the operator.
In the present embodiment, a plurality of evaluation quantities that are indices for evaluating the performance of the hot stove 400 are used as the objective function ΔP. Specifically, in the present embodiment, the case where the objective function ΔP is represented by the following equation (1) will be described as an example.

ΔP = [ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, ΔP6] (1)
Here, ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, and ΔP6 in the equation (1) are expressed by equations (2) to (7), respectively.
ΔP1 = P11−P10 (2)
However, P11: Thermal efficiency after changing the target value, P10: Reference value of thermal efficiency ΔP2 = P21−P20 (3)
However, P21: Minimum temperature of the quartz brick after changing the target value, P20: Reference value of the minimum temperature of the quartz brick

ΔP3 = P31−P30 (4)
However, P31: CO ₂ emission after changing the target value, P30: Reference value of CO ₂ emission ΔP4 = P41−P40 (5)
However, P41: the total flow rate of COG after changing the target value, P40: the reference value of the total flow rate of COG ΔP5 = P51−P50 (6)
However, P51: The average temperature of the exhaust gas after changing the target value, P50: The reference value of the average temperature of the exhaust gas

ΔP6 = P61−P60 (7)
However, P61: input heat amount after changing the target value, P60: reference value of input heat amount In the following description, ΔP1 is thermal efficiency, ΔP2 is silica brick temperature, ΔP3 is CO ₂ emission amount, if necessary. ΔP4 is called the COG total flow rate, ΔP5 is called the exhaust gas average temperature, and ΔP6 is called the input heat amount.

As described above, in the present embodiment, the objective function setting unit 111 indicates that the thermal efficiency ΔP1, the quartz brick temperature ΔP2, the CO ₂ emission amount ΔP3, the COG total flow rate ΔP4, the exhaust gas average temperature ΔP5, and the input heat amount ΔP6 are the objective function ΔP. Is set.
The objective function setting unit 111 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 100.

(Restriction condition setting unit 112)
The constraint condition setting unit 112 sets a constraint condition when setting a target value based on an operation from the operator. In the present embodiment, the constraint conditions of the following expressions (8) to (10) are set by the constraint condition setting unit 112.
ΔP2 = 0 (8)
P21 ≧ P2min (9)
P41 ≦ P4max (10)
Here, P2min is the lower limit value of the adjustable range of the temperature of the quartz brick 407. P4max is an upper limit value of the adjustable range of the total flow rate of COG.
The constraint condition setting unit 112 can be realized by using an HD or the like provided in the information processing apparatus 100, for example.

(Pareto optimal solution derivation unit 113)
The Pareto optimal solution deriving unit 113 applies a multi-objective GA to solve a multi-objective optimization problem for the objective function ΔP set by the objective function setting unit 111 and derives a plurality of Pareto optimal solutions. The functions of the Pareto optimal solution deriving unit 113 will be described below. Note that the Pareto optimal solution deriving unit 113 is realized by using, for example, a CPU, ROM, RAM, HD, interface, and the like provided in the information processing apparatus 100.

The individual generation unit 113a sets two sets (current generation and next generation) into which N individuals enter, and randomly generates N individuals in the current generation.
The target value conversion unit 113b converts the N individuals generated by the individual generation unit 113a into a target value pattern.

  FIG. 4 is a diagram illustrating an example of the target value pattern. As described above, since the dome temperature U1, the C / B ratio U2, and the BFG flow rate U3 are controlled variables, target value patterns for these are required. Here, the target value pattern referred to in the present embodiment represents the transition of the target value from the start of combustion to the end of combustion.

  FIG. 4A shows a target value pattern 41 of the dome temperature U1. In FIG. 4A, U1max is the upper limit value of the adjustable range of the dome temperature U1, and U1min is the lower limit value thereof. It is necessary that the target value pattern 41 of the dome temperature U1 falls within the adjustable range (U1min to U1max) of the dome temperature U1.

  U10 is a reference value of the dome temperature U1, and is preset by an operator, for example. U11 is the dome temperature before the target value is changed. U12 is the dome temperature after the target value is changed. T1max is an upper limit value of the adjustable range of the switching time of the dome temperature U1, and T1min is a lower limit value thereof. It is necessary to set the target value pattern 41 of the dome temperature U1 so that the dome temperature U1 is changed within the adjustable range (T1min to T1max) of the switching time of the dome temperature U1.

  FIG. 4B shows a target value pattern 42 of the C / B ratio U2. In FIG.4 (b), U2max is an upper limit of the adjustable range of C / B ratio U2, and U2min is the lower limit. It is necessary to make the target value pattern 42 of the C / B ratio U2 fall within the adjustable range (U2min to U2max) of the C / B ratio U2.

  U20 is a reference value of the C / B ratio U2, and is set in advance by an operator, for example. U21 is the C / B ratio before the target value is changed. U22 is the C / B ratio after the target value is changed. T2max is an upper limit value of the adjustable range of the switching time of the C / B ratio U2, and T2min is the lower limit value thereof. It is necessary to set the target value pattern 42 of the C / B ratio U2 so that the C / B ratio U2 is changed within the adjustable range (T2min to T2max) of the switching time of the C / B ratio U2.

  FIG. 4C shows a target value pattern 43 of the BFG flow rate U3. In FIG. 4C, U3max is the upper limit value of the adjustable range of the BFG flow rate U3, and U3min is the lower limit value thereof. It is necessary to make the target value pattern 43 of the BFG flow rate U3 fall within the adjustable range (U3min to U3max) of the BFG flow rate U3.

  U30 is a reference value of the BFG flow rate U3, and is set in advance by the operator so that the blowing temperature matches the target value, for example. U31 is the BFG flow rate before the target value is changed. U32 is the BFG flow rate after the target value is changed. T3max is the upper limit value of the adjustable range of the switching time of the BFG flow rate U3, and T3min is the lower limit value thereof. It is necessary to set the target value pattern 43 of the BFG flow rate U3 so that the BFG flow rate U3 is changed within the adjustable range (T3min to T3max) of the switching time of the BFG flow rate U3.

FIG. 5 is a diagram illustrating an example of the configuration of an individual.
In FIG. 5, an individual 50 is, for example, 32-bit information, and includes dome temperature information 51, C / B ratio information 52, and BFG flow rate information 53.
The dome temperature information 51 further includes ΔU11, ΔU12, and ΔT11.
ΔU11 is a value corresponding to the dome temperature U11 before the target value is changed. Among the points obtained by equally dividing the range from the lower limit value U1min of the adjustable range of the dome temperature U1 to the upper limit value U1max into 16 parts, This indicates the number of the point counted from (see FIG. 4A).

ΔU12 is a value corresponding to the dome temperature U12 after the target value has been changed, and among the points obtained by equally dividing the range from the lower limit value U1min to the upper limit value U1max of the adjustable range of the dome temperature U1, into 16 This indicates the number of the point counted from (see FIG. 4A).
ΔT11 is a value corresponding to the switching time T11 of the dome temperature U1, and the range from the lower limit value T1min to the upper limit value T1max of the adjustable range adjustable range of the switching time of the dome temperature U1 is equally divided into 16 points. , Indicating the number of points counted from the bottom (see FIG. 4A).

The C / B ratio information 52 further includes ΔU21, ΔU22, and ΔT21.
ΔU21 is a value corresponding to the C / B ratio information 52 before the target value is changed, and the range from the lower limit value U2min to the upper limit value U2max of the adjustable range of the C / B ratio U2 is equally divided into 16 This indicates the number of the points counted from the bottom (see FIG. 4B).

ΔU22 is a value corresponding to the C / B ratio information 52 after the target value is changed, and the range from the lower limit value U2min to the upper limit value U2max of the adjustable range of the C / B ratio U2 is equally divided into 16 This indicates the number of the points counted from the bottom (see FIG. 4B).
ΔT21 is a value corresponding to the switching time T21 of the C / B ratio information 52, and the range from the lower limit value T2min to the upper limit value T2max of the adjustable range adjustable range of the switching time of the C / B ratio U2 is equalized. This indicates the number of points counted from the bottom among the 16 divided points (see FIG. 4B).

  The BFG flow rate information 53 further includes ΔU32 and ΔT31. In the present embodiment, it is assumed that the BFG flow rate U31 before the target value is changed is automatically adjusted by the process simulator 150 so that the blowing temperature matches the target temperature. Therefore, the individual 50 does not include a value corresponding to the BFG flow rate U31 before the target value is changed. However, a value corresponding to the BFG flow rate U31 before the target value is changed may be included in the individual 50.

ΔU32 is a value corresponding to the BFG flow rate U3 after the target value is changed. Among the points obtained by equally dividing the range from the lower limit value U3min of the adjustable range of the BFG flow rate U3 to the upper limit value U3max by 16 This indicates the number of the point counted from (see FIG. 4C).
ΔT31 is a value corresponding to the switching time T31 of the BFG flow rate U3, and the range from the lower limit value T3min to the upper limit value T3max of the adjustable range adjustable range of the switching time of the BFG flow rate U3 is equally divided into 16 points. , Indicating the number of points counted from the bottom (see FIG. 4C).

As described above, in this embodiment, each of the individuals 50 generated by the individual generation unit 113a includes, as an example of information for specifying a target value, ΔU11, ΔU12, ΔT11, ΔU21, ΔU22, ΔT21, ΔU32, ΔT31 is included, each of which is 4-bit information.
In FIG. 4, the target value patterns 41 to 43 change at the inclination of 90 [°] at the switching times T11, T21, and T31. The target value pattern may change with the inclination of the angle. In such a case, it is necessary to set the inclination information for the individual as a variable for determining the target value pattern.

Returning to the description of FIG. 3, the target value conversion unit 113b uses U11, U12, T11, U21, U22, T21, U32, and T31 calculated by the following equations (11) to (18) to generate individuals. The N individuals generated by the unit 113a are converted into target value patterns.
U11 = U1min + (U1max−U1min) × ΔU11 / 16 (11)
U12 = U1min + (U1max−U1min) × ΔU12 / 16 (12)
T11 = T1min + (T1max−T1min) × ΔT11 / 16 (13)
U21 = U2min + (U2max−U2min) × ΔU21 / 16 (14)
U22 = U2min + (U2max−U2min) × ΔU22 / 16 (15)
T21 = T2min + (T2max−T2min) × ΔT21 / 16 (16)
U32 = U3min + (U3max−U3min) × ΔU32 / 16 (17)
T31 = T3min + (T3max−T3min) × ΔT31 / 16 (18)

The simulation instruction unit 113c instructs the process simulator 150 to perform the process simulation by passing the target value pattern converted by the target value conversion unit 113b.
As shown in FIG. 3, the process simulator 150 includes a controller model 151 that models the controller 300 shown in FIG. A modeled subtractor model 153 is included, and an actual process control system is simulated (computer simulation) with software.
First, the process simulator 150 extracts “operation performance data in the hot stove 400” stored in the performance database 200. And the process simulator 150 sets the initial value for performing a simulation based on the data of the operation performance in the hot stove 400.

After the initial value is set in this way, the process simulator 150 simulates the operation of the plant control system 10 shown in FIG. 1 using the target value pattern converted by the target value conversion unit 113b. When this simulation is completed, the process simulator 150 obtains “thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5, and input heat amount ΔP6” obtained from the physical model 152 of the hot stove. And the variables used to obtain them (P10, P11, P20, P21, P30, P31, P40, P41, P50, P51, P60, P61) and the target value optimization device 110 (simulation result acquisition unit 113d) Output to.

The simulation result acquisition unit 113d acquires, from the process simulator 150, thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5, and input heat amount ΔP6 as evaluation amounts.
The fitness deriving unit 113e obtains the fitness from the “thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5, and input heat amount ΔP6” obtained by the simulation result acquisition unit 113d. Derived (calculated). Since the fitness can be derived by a known method in multipurpose GA, detailed description thereof is omitted.

The constraint condition determination unit 113f obtains the “silica brick temperature ΔP2 obtained by the simulation result acquisition unit 113d, the minimum temperature P21 of the quartz brick after changing the target value, and the total flow rate P41 of COG after changing the target value. It is determined whether or not “satisfy” the constraint conditions (equations (8) to (10)) set by the constraint condition setting unit 112.
The fitness determining unit 113g selects the fitness derived by the fitness deriving unit 113e when the constraint determining unit 113f determines that the constraint is satisfied. On the other hand, when it is determined by the constraint condition determination unit 113f that the constraint condition is satisfied, the fitness determination unit 113g cancels the fitness derived by the fitness derivation unit 113e, and sets the minimum value as the fitness To do. By doing in this way, the “silica brick temperature ΔP2 corresponding to this fitness, the minimum temperature P21 of the quartz brick after changing the target value, and the total flow rate P41 of COG after changing the target value” Output from the Pareto optimal solution finding unit 113h is prevented.

The above processing of the target value conversion unit 113b, the simulation instruction unit 113c, the simulation result acquisition unit 113d, the fitness level deriving unit 113e, the constraint condition determination unit 113f, and the fitness level determination unit 113g is the N generated by the individual generation unit 113a. It is performed individually for each individual.
The Pareto optimal solution solving unit 113h performs crossover, mutation, or copying on the individual generated by the individual generation unit 113a with a certain probability, and stores the result in the next generation. When the number of next-generation individuals reaches N by this processing, the Pareto optimal solution solving unit 113h deletes the current-generation individual and moves all the next-generation individuals to the current generation.

  The operations of the target value conversion unit 113b, the simulation instruction unit 113c, the simulation result acquisition unit 113d, the fitness level deriving unit 113e, the constraint condition determination unit 113f, the fitness level determination unit 113g, and the Pareto optimal solution solving unit 113h are as follows. Repeat until the maximum generation number G is reached. Then, the Pareto optimal solution solving unit 113h finally evaluates (ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, ΔP6) of the individual (a plurality) having the highest fitness among the individuals included in the current generation. And the target value pattern for the individual are output to the Pareto optimal solution storage unit 114.

(Pareto optimal solution storage unit 114)
The Pareto optimal solution storage unit 114 correlates the “individual evaluation amount (ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, ΔP6) and the target value pattern for the individual” output from the Pareto optimal solution solving unit 113h. Remember.
FIG. 6 is a diagram illustrating an example of the Pareto optimal solution. In the present embodiment, there are six objective functions ΔP (ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, ΔP6), but for the sake of easy understanding, here, two objective functions ΔP1 and ΔP2 are used. Pareto optimal solution is shown.
FIG. 6 shows an example in which seven Pareto optimal solutions Z1 to Z7 are obtained. Therefore, in the example shown in FIG. 6, the Pareto optimal solution storage unit 114 has values of these seven Pareto optimal solutions Z1 to Z7 (values of ΔP1 to ΔP6) and target values corresponding to the Pareto optimal solutions Z1 to Z7. The pattern is stored in association with each other.
The pareto optimal solution storage unit 114 can be realized by using, for example, an HD provided in the information processing apparatus 100.

(Operation policy acquisition unit 115)
Returning to the description of FIG. 3, the operation policy acquisition unit 115 inputs information related to the operation policy input by the operator.
The operation policy acquisition unit 115 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 100.

(Plant state / evaluation amount acquisition unit 116)
The plant state / evaluation quantity acquisition unit 116 uses the state quantity (the actual value) of the hot stove 400 and the evaluation quantities ΔP <b> 1 to ΔP <b> 6 (the actual values) of the hot stove 400 to various types provided in the hot stove 400. Obtain from a measuring instrument.
As described above, the state quantity of the hot stove 400 is an amount (for example, temperature or air flow rate) measured directly or indirectly from time to time by operating the hot stove 400. On the other hand, the evaluation amounts ΔP1 to ΔP6 in the hot stove 400 are indexes for evaluating the performance of the hot stove 400, and are, for example, longer periods (for example, in units of one day) than the period in which the state quantity of the hot stove 400 is acquired. It is calculated based on the state quantity of the hot stove 400 acquired over the period.

The evaluation values ΔP1 to ΔP6 (the actual values thereof) in the hot stove 400 are used for the target value optimization device 110 to tune the process simulator 150, for example. On the other hand, the state quantity of the hot stove 400 is used in the operation condition change determination unit 117 to determine whether or not the operation condition has been changed.
The plant state / evaluation amount acquisition unit 116 can be realized by using, for example, a CPU, ROM, RAM, and HD provided in the information processing apparatus 100.

(Operation condition change determination unit 117)
The operation condition change determination unit 117 operates based on the operation policy acquired by the operation policy acquisition unit 115 and the state quantity (actual value) of the hot stove 400 acquired by the plant state / evaluation amount acquisition unit 116. By determining whether or not at least one of the policy and the state quantity of the hot stove 400 has been changed, it is determined whether or not the operating conditions of the hot stove 400 have been changed. As described above, in the present embodiment, the operating policy and the state quantity of the hot stove 400 are included in the operating conditions of the hot stove 400. Therefore, when at least one of the operation policy and the state quantity of the hot stove 400 is changed, the operating conditions of the hot stove 400 are changed.
The operation condition change determination unit 117 can be realized by using, for example, a CPU, ROM, RAM, HD, and the like provided in the information processing apparatus 100.

(Pareto optimal solution candidate display section 118)
When the operation condition change determination unit 117 determines that the operation condition of the hot stove 400 has been changed, the Pareto optimal solution candidate display unit 118 has a plurality of Pareto optimal solutions Z1 to Z1 stored in the optimal Pareto solution storage unit 114. A GUI for allowing the operator to select Z7 (that is, the target value) is displayed on a display (for example, a liquid crystal display) connected to the information winning device 100.
FIG. 7 is a diagram illustrating an example of a GUI for causing the operator to change the Pareto optimal solutions Z1 to Z7 (that is, target values).

In FIG. 7, the GUI 70 includes a change in the operating conditions, Pareto optimal solutions (names) Z1 to Z7, evaluation amounts ΔP1 to ΔP6 that determine values of the Pareto optimal solutions Z1 to Z7, and evaluation amounts ΔP1 to ΔP1. The superiority or inferiority of ΔP6 is displayed. Further, the currently selected Pareto optimal solution (in the example shown in FIG. 7, the Pareto optimal solution Z3) is surrounded by a square.
Here, the superiority or inferiority of the evaluation amounts ΔP1 to ΔP6 indicates the relative superiority or inferiority of the evaluation amounts ΔP1 to ΔP6 in the respective Pareto optimal solutions Z1 to Z7. For example, when the Pareto optimal solution Z3 is selected The COG total flow rate ΔP4, the CO ₂ emission amount ΔP3, the quartz brick temperature ΔP2, the thermal efficiency ΔP1, the input heat amount ΔP6, and the exhaust gas average temperature ΔP5 are excellent in this order.

In the present embodiment, by operating the GUI 70, the user manually selects any one of the Pareto optimal solutions Z1 to Z7 and the information processing apparatus 100 automatically selects both. I have to.
When the user manually selects any of the Pareto optimal solutions Z1 to Z7, the operator first determines which Pareto optimal solution should be changed from the currently selected Pareto optimal solution while viewing the GUI 70, and information By operating a user interface connected to the processing device 70, an arrow 71 displayed on the GUI 70 is moved. Then, an arrow 71 is displayed beside the Pareto optimal solution desired by the operator. Thereafter, the operator operates the user interface and presses the manual button 72 to determine the Pareto optimal solution to be changed.

On the other hand, when the information processing apparatus 100 automatically selects any of the Pareto optimal solutions Z1 to Z7, the operator presses the automatic button 73. Then, one of the Pareto optimal solutions Z1 to Z7 is automatically selected as described later.
The pareto optimal solution candidate display unit 118 can be realized by using, for example, a CPU, ROM, RAM, VRAM, and HD provided in the information processing apparatus 100. In addition to the information shown in FIG. 7, information indicating the target value pattern in each Pareto optimal solution may be displayed on the GUI 70.

(Pareto optimal solution acquisition determination unit 119)
The Pareto optimal solution acquisition determination unit 119 determines which of the manual button 72 and the automatic button 73 has been pressed by an operator's operation on the GUI 70.
The Pareto optimal solution acquisition determination unit 119 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 100.

(Weighting factor deriving unit 120)
When it is determined by the Pareto optimal solution acquisition determination unit 119 that the automatic button 73 has been pressed, the weighting factor deriving unit 120 includes information on the changed operating conditions (operation policy acquisition unit 115 and plant state / evaluation amount acquisition unit 116). Weight vector W (weight) for each of thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5, and input heat amount ΔP6, based on information acquired from at least one of them Coefficient). The weight vector W is expressed by the following equation (19).

W = [W1, W2, W3, W4, W5, W6] (19)
W1 is a weight vector (weighting factor) of thermal efficiency ΔP1, W2 is a weight vector (weighting factor) of quartz brick temperature ΔP2, W3 is a weight vector (weighting factor) of CO ₂ emission amount ΔP3, and W4 is a weight vector of COG total flow rate ΔP4. (Weighting factor), W5 is a weighting vector (weighting factor) of the exhaust gas average temperature ΔP5, and W6 is a weighting vector (weighting factor) of the input heat amount ΔP6. The values of the weight vectors (weight coefficients) W1 to W6 are values of 0 or more and 1 or less, and the sum of them is 1. For example, when the thermal efficiency ΔP1 is maximized, the weight vector W is expressed by, for example, the following equation (20).
W = [1, 0, 0, 0, 0, 0] (20)
The weighting factor deriving unit 120 can be realized by using, for example, a CPU, ROM, RAM, HD, and the like provided in the information processing apparatus 100.

(Evaluation Function Deriving Unit 121)
The evaluation function deriving unit 121 uses the weight vector W derived by the weight coefficient deriving unit 120 and the values of the Pareto optimal solutions Z1 to Z7 (values of ΔP1 to ΔP6) stored in the Pareto optimal solution storage unit 114. Then, the evaluation function Q is derived (calculated) by the following equation (21), and the load sum of the weight vectors (weight coefficients) W1 to W6 is obtained. At this time, the evaluation function deriving unit 121 performs a normalization process on the evaluation amounts ΔP1 to ΔP6 so that the evaluation amounts ΔP1 to ΔP6 have substantially the same fluctuation range. Therefore, ΔP _r in the equation (21) indicates a normalized evaluation amount. In addition, the evaluation function deriving unit 121 sets the value of the Pareto optimal solution Z (ΔP) stored in the Pareto optimal solution storage unit 114 for the evaluation amount to be maximized as the thermal efficiency ΔP1 among the evaluation amounts ΔP1 to ΔP6. Is multiplied by “−1”. Therefore, among ΔP _r in equation (21), the evaluation amount to be maximized differs in sign (positive / negative) from that stored in the Pareto optimal solution storage unit 114.
Here, the normalization process and the multiplication process of “−1” are performed on the value of the Pareto optimal solution Z (the value of ΔP) stored in the Pareto optimal solution storage unit 114. What has been processed may be stored in the Pareto optimal solution storage unit 114.

In the equation (21), s is the number of evaluation amounts ΔP. In the present embodiment, since there are six evaluation amounts ΔP1 to ΔP6, s = 6.
The evaluation function deriving unit 121 can be realized by using, for example, a CPU, ROM, RAM, and HD provided in the information processing apparatus 100.

(Pareto optimal solution selection unit 122)
Based on the seven evaluation functions Q derived by the evaluation function deriving unit 121, the Pareto optimum solution selecting unit 122 is the most appropriate of the Pareto optimal solutions Z1 to Z7 stored in the Pareto optimal solution storage unit 114. Select the Pareto optimal solution. Specifically, in the present embodiment, the Pareto optimal solution selection unit 122 selects the minimum evaluation function Q among the seven evaluation functions Q derived by the evaluation function deriving unit 121 and corresponds to the evaluation function Q. Select the Pareto optimal solution. That is, in this embodiment, the Pareto optimal solution selection unit 122 selects a Pareto optimal solution that minimizes the load sum of the weight vector (weight coefficient) after the operation condition is changed.
The Pareto optimal solution selection unit 122 can be realized by using, for example, a CPU, ROM, RAM, HD, and the like provided in the information processing apparatus 100.

(Target value changing unit 123, target value storage unit 124)
When it is determined by the Pareto optimal solution acquisition determination unit 119 that the manual button 72 has been pressed, the target value changing unit 123 displays a target value pattern corresponding to the Pareto optimal solution selected by the operator using the GUI 70 shown in FIG. Read from the Pareto optimal solution storage unit 114.
On the other hand, when the Pareto optimal solution acquisition determination unit 119 determines that the automatic button 73 has been pressed, the target value pattern corresponding to the Pareto optimal solution selected by the Pareto optimal solution selection unit 122 is read from the Pareto optimal solution storage unit 114. . Then, the target value changing unit 123 rewrites the target value pattern stored in the target value storage unit 124 to the read target value pattern.
The target value changing unit 123 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 100. The target value storage unit 124 can be realized by using, for example, an HD provided in the information processing apparatus 100.

(Target value output unit 125)
When the target value pattern stored in the target value storage unit 124 is updated, the target value output unit 125 outputs the updated target value pattern to the controller 300. Thereby, the target value at the time of controlling the hot stove 400 is changed.
The target value output unit 125 can be realized by using, for example, a CPU, ROM, RAM, HD, interface, and the like provided in the information processing apparatus 100.

  In FIG. 3, when the target value pattern is not yet selected (in the initial stage of operation), for example, the target value pattern can be output to the controller 300 as follows. That is, the weighting factor deriving unit 120, the evaluation function deriving unit 121, the Pareto optimization selecting unit 122, and the target value changing unit 123 are activated to automatically select one Pareto optimal solution according to the operation condition, A target value pattern corresponding to the Pareto optimal solution is read from the Pareto optimal solution storage unit 114 and written to the target value storage unit 124. Then, the target value output unit 125 outputs the target value pattern written in the target value storage unit 124 to the controller 300. Further, when the target value pattern is not yet selected (in the initial stage of operation), a predetermined target value pattern may be output to the controller 300.

<Pareto optimal solution derivation process>
Next, an example of the operation of the target value optimization apparatus 110 when executing a Pareto optimal solution derivation process for deriving and storing a plurality of Pareto optimal solutions in advance will be described with reference to the flowchart of FIG.
First, in step S1, the objective function setting unit 111 sets an objective function ΔP based on an operation from an operator. As described above, in the present embodiment, the objective function setting unit 111 is that the thermal efficiency ΔP1, the silica brick temperature ΔP2, the CO ₂ emission amount ΔP3, the COG total flow rate ΔP4, the exhaust gas average temperature ΔP5, and the input heat amount ΔP6 are the objective function ΔP. Is set by

Next, in step S <b> 2, the constraint condition setting unit 112 sets a constraint condition when setting a target value based on an operation from the operator. As described above, in the present embodiment, the constraint conditions of the equations (8) to (10) are set by the constraint condition setting unit 112.
Next, in step S3, the individual generation unit 113a sets two sets (current generation and next generation) into which N individuals enter.
Next, in step S4, the individual generation unit 113a randomly generates N individuals in the current generation set in step S3.
Thus, in the present embodiment, for example, an example of the individual generation unit is realized by performing the process of step S4.

Next, in step S5, the target value conversion unit 113b converts the N individuals generated in step S4 into a target value pattern. As described above, the target value conversion unit 113b uses the U11, U12, T11, U21, U22, T21, U32, and T31 calculated according to the equations (11) to (18) to target N individuals. Convert to value pattern. The target value pattern is as shown in FIG.
Thus, in the present embodiment, for example, an example of the target value conversion unit is realized by performing the process of step S5.

Next, in step S6, the simulation instruction unit 113c instructs the process simulator 150 to perform computer simulation. As a result, the process simulator 150 performs a computer simulation, and as a result, “thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5 obtained from the physical model 152 of the hot stove. , And the input heat amount ΔP6 ″ and the variables (P10, P11, P20, P21, P30, P31, P40, P41, P50, P51, P60, P61) used to obtain them.
Thus, in the present embodiment, for example, an example of a simulation unit is realized by performing the process of step S6.

Next, in step S7, the simulation result acquisition unit 113d waits until the above-described information is acquired from the process simulator 150. And if the information mentioned above is acquired from the process simulator 150, it will progress to step S8.
Thus, in the present embodiment, for example, an example of a simulation result acquisition unit is realized by performing the process of step S7.
In step S8, the simulation result acquisition unit 113d acquires “thermal efficiency ΔP1, silica brick temperature ΔP2, CO ₂ emission amount ΔP3, COG total flow rate ΔP4, exhaust gas average temperature ΔP5, and input heat amount acquired from the process simulator 150 in step S7. The fitness is derived using ΔP6 ″ as an evaluation amount.
Thus, in the present embodiment, for example, an example of the fitness deriving unit is realized by performing the process of step S8.

Next, in step S9, the constraint condition determination unit 113f obtains the “silica brick temperature ΔP2 acquired by the simulation result acquisition unit 113d in step S7, the minimum temperature P21 of the quartz brick after changing the target value, and the target value. It is determined whether or not the COG total flow rate P41 ″ after the change satisfies the constraint conditions (equations (8) to (10)) set by the constraint condition setting unit 112 in step S2. As a result of the determination, if the constraint condition is satisfied, the process proceeds to step S11 described later.
On the other hand, if the constraint condition is not satisfied, the process proceeds to step S10. In step S10, the fitness determining unit 113g cancels the fitness derived by the fitness deriving unit 113e in step S8, and sets the maximum value as the fitness. Then, the process proceeds to step S11.

Then, when proceeding to step S11, the Pareto optimal solution solving unit 113h performs crossover, mutation, or copying with a certain probability on the individual generated by the individual generation unit 113a in step S4, and the result is obtained. Save to the next generation.
Next, in step S12, the Pareto optimal solution solving unit 113h determines whether or not the number of next-generation individuals has reached N. If the number of next-generation individuals is not N as a result of this determination, the process returns to step S11, and steps S11 and S12 are repeated until the number of next-generation individuals reaches N.
When the number of next-generation individuals reaches N, the process proceeds to step S13. In step S13, the Pareto optimal solution solving unit 113h deletes the current generation individual and moves all next generation individuals to the current generation.

  Next, in step S14, the Pareto optimal solution solving unit 113h determines whether or not the processing of steps S5 to S13 has been performed up to the maximum generation number G. As a result of this determination, if the processing of steps S5 to S13 has not been performed up to the maximum generation number G, the process returns to step S5, and the processing of steps S5 to S13 is performed until the processing of steps S5 to S13 is performed up to the maximum generation number G. Repeat. And if the process of step S5-S13 is performed to the maximum generation number G, it will progress to step S15.

When proceeding to step S15, the Pareto optimal solution solving unit 113h finally has an evaluation amount (ΔP1, ΔP2, ΔP3, ΔP4,...) Of the individuals having the highest fitness among the individuals included in the current generation. ΔP5, ΔP6) are adopted as Pareto optimal solutions. Then, the Pareto optimal solution solving unit 113h outputs the evaluation amount (Pareto optimal solution) of the individual and the target value pattern corresponding to the individual to the Pareto optimal solution storage unit 114. As a result, the “individual evaluation amount (ΔP1, ΔP2, ΔP3, ΔP4, ΔP5, ΔP6)” output from the Pareto optimal solution solving unit 113h and the target value pattern for the individual are associated with each other in a Pareto optimal solution. Stored in the storage unit 114.
Thus, in the present embodiment, for example, by performing the process of step S15, an example of the Pareto optimal solution solving means and the Pareto optimal solution storage means is realized. Further, for example, an example of the Pareto optimal solution deriving unit is realized by executing the flowchart of FIG.

<Target value change processing>
Next, an example of the operation of the target value optimizing device 110 when performing the target value changing process for changing the target value pattern when the operating condition of the hot stove 400 is changed will be described with reference to the flowchart of FIG. To do.
First, in step S <b> 21, the operation condition change determination unit 117 determines whether or not the operation condition of the hot stove 400 has been changed. As described above, in the present embodiment, at least one of the operation policy acquired by the operation policy acquisition unit 115 and the state quantity value of the hot stove 400 acquired by the plant state / evaluation amount acquisition unit 116 is changed. Then, it is determined that the operating condition of the hot stove 400 has been changed.
Thus, in the present embodiment, for example, an example of the operation condition change determination unit is realized by performing the process of step S21.

Next, in step S22, the Pareto optimal solution candidate display unit 118 displays a GUI 70 for causing the operator to select a plurality of Pareto optimal solutions Z1 to Z7 (that is, target values) stored in the optimal Pareto solution storage unit 114. (See FIG. 7).
Thus, in the present embodiment, for example, an example of a Pareto optimal solution candidate display unit is realized by performing the process of step S22.
Next, in step S23, the Pareto optimal solution acquisition determination unit 119 determines whether or not the changed Pareto optimal solution has been selected by the operator. As described above, in the present embodiment, when the manual button 72 is pressed by the operator, it is determined that the changed Pareto optimal solution has been selected by the operator. On the other hand, when the automatic button 73 is pressed by the operator, it is determined that the changed Pareto optimal solution has not been selected by the operator.

As a result of this determination, if the changed Pareto optimal solution is selected by the operator, the process proceeds to step S27 described later.
On the other hand, if the changed Pareto optimal solution is not selected by the operator, the Pareto optimal solution is automatically selected, and the process proceeds to step S24. In step S24, the weight coefficient deriving unit 120 derives a weight vector W (weight coefficient) based on the changed operation condition.
Thus, in the present embodiment, for example, an example of a weighting factor deriving unit is realized by performing the processing in step S24.

Next, in step S25, the evaluation function deriving unit 121 uses the weight vector W derived by the weight coefficient deriving unit 120 in step S24 and the Pareto stored in the Pareto optimal solution storage unit 114 in the process of step S15 in FIG. Using the values of the optimal solutions Z1 to Z7 (values of ΔP1 to ΔP6), the load sum of the weight vectors (weighting factors) W1 to W6 is obtained. More specifically, the evaluation function deriving unit 121 derives the evaluation function Q using equation (21).
Thus, in the present embodiment, for example, an example of the load sum deriving unit is realized by performing the process of step S25.

Next, in step S26, the Pareto optimal solution selection unit 122 selects a Pareto optimal solution that minimizes the load sum of the weight vector (weight coefficient) after the operation condition is changed. More specifically, the Pareto optimal solution selection unit 122 selects the minimum evaluation function Q among the seven evaluation functions Q derived by the evaluation function deriving unit 121 in step S25 and corresponds to the evaluation function Q. Select the Pareto optimal solution.
As described above, in the present embodiment, for example, an example of the Pareto optimal solution selection unit is realized by performing the processing of steps S23 and S26.

When the Pareto optimal solution is selected as described above, the process proceeds to step S27. When the process proceeds to step S27, the target value changing unit 123 obtains from the Pareto optimal solution storage unit 114 a target value pattern corresponding to the Pareto optimal solution determined to be selected in step S23 or the Pareto optimal solution selected in step S26. Reading and rewriting the target value pattern stored in the target value storage unit 124 to the read target value pattern.
Next, in step S28, the target value output unit 125 outputs the target value pattern rewritten (updated) by the target value changing unit 123 in step S27 to the controller 300.
Thus, in the present embodiment, for example, an example of the target value output unit is realized by performing the process of step S28.

  As described above, in the present embodiment, a plurality of evaluation quantities, which are indices for evaluating the performance of the hot stove 400, are set as the objective function ΔP, the multi-objective optimization problem for the objective function ΔP is solved by the multi-objective GA, and a plurality of Pareto optimal solutions are obtained. Is previously obtained offline and stored. Thereafter, the Pareto optimal solution is changed to one of a plurality of stored Pareto optimal solutions in accordance with the change of the operation condition, and a target value pattern corresponding to the changed Pareto optimal solution is set in the controller 300. Therefore, even when the operating conditions are changed, an appropriate target value can be set in the controller 300 as much as possible without performing the optimization calculation again.

  Further, in the present embodiment, when the operation condition is changed, a GUI 70 for causing the operator to change the Pareto optimal solution is displayed, and the operator refers to and operates this GUI 70 to select one of a plurality of Pareto optimal solutions. Select. In this GUI 70, the contents of the change in the operation conditions, the Pareto optimal solutions (names) Z1 to Z7, the evaluation amounts ΔP1 to ΔP6 for determining the values of the Pareto optimal solutions Z1 to Z7, and the superiority or inferiority of the evaluation amounts ΔP1 to ΔP6 Is displayed. Accordingly, it is possible to show the operator a guideline for actions to be taken by the operator in response to the change of the operation condition. That is, the operator can know how the evaluation amounts ΔP1 to ΔP6 are changed by changing the Pareto optimal solution, and can select an appropriate Pareto optimal solution as much as possible. .

  In the present embodiment, when the operator specifies that the Pareto optimal solution is automatically selected, a weight vector (weight coefficient) corresponding to the changed operation condition is obtained, and the obtained weight vector (weight coefficient) ) Is selected from a plurality of Pareto optimal solutions, and a target value pattern corresponding to the selected Pareto optimal solution is set in the controller 300. Therefore, it is possible to automatically set an appropriate target value in the controller 300 as much as possible.

(First modification)
Based on the constraint conditions stored in the constraint condition storage unit 112, the Pareto optimal solution selection unit 122 can automatically select the Pareto optimal solution as shown in FIG. FIG. 10 is a diagram conceptually illustrating a modification of the method for automatically selecting the Pareto optimal solution. Here, it is assumed that the constraint condition is the following expression (22).
ΔP2 ≦ T ₀ (22)
In this case, the Pareto optimal solution selection unit 122 is not the Pareto optimal solution Z1 to Z7 stored in the Pareto optimal solution storage unit 114, but the Pareto optimal solution Z1 to Z4 satisfying the equation (22) is the most Choose an appropriate Pareto optimal solution. In this case, it is possible not to obtain the weight vector W and the evaluation function Q for the Pareto optimal solutions Z5 to Z7. Therefore, the process for automatically selecting the Pareto optimal solution can be further simplified.

(Second modification)
A plurality of Pareto optimal solutions stored in the Pareto optimal solution storage unit 114 may be updated at a constant period.
Depending on the operating conditions and disturbances, reference values P10, P20, P30, P40, P50, P60 (see equations (2) to (7)) of the evaluation amounts ΔP1 to ΔP6, and reference values of the target values U1, U2, U3 U10, U20, and U30 vary. Thereby, the adjustable range (U1min to U1max, U2min to U2max, U3min to U3max) of the target values U1, U2, and U3, the range of the constraint conditions, and the evaluation for the fluctuations of the target values U1, U2, and U3 (control amounts). The sensitivity of fluctuations in the amounts ΔP1 to ΔP6 changes. Therefore, the target value pattern is changed, and the result of simulation by the process simulator 150 is changed due to the change of the target value pattern, and the fitness is also changed. Therefore, the Pareto optimal solution finally obtained is also changed. Therefore, it is possible to select a more appropriate Pareto optimal solution in response to fluctuations in operating conditions and disturbances.

(Third Modification)
The method for obtaining a plurality of Pareto optimal solutions is not limited to the multipurpose GA. For example, instead of the multi-purpose GA, a plurality of Pareto optimal solutions can be obtained using various algorithms such as NSGA-II and NCGA. Further, when sufficient time for obtaining a plurality of Pareto optimal solutions can be secured, the following may be performed. In other words, applying a weighting factor method to multiple objective functions to convert multiple objective functions to a single objective function, and using multiple single objective optimization techniques such as hill climbing while changing the weighting coefficient A Pareto optimal solution may be obtained.

(Fourth modification)
In the present embodiment, the case where the operator manually selects any of the plurality of Pareto optimal solutions and the case where the information processing apparatus 100 automatically selects one of them is shown. Only one of them may be performed.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. In the first embodiment, the case where the plant to be controlled is a hot stove has been described as an example. On the other hand, in this embodiment, the case where the plant to be controlled is a hot rolling facility will be described as an example. As described above, the present embodiment and the first embodiment differ in configuration and processing due to different control target plants. Therefore, in the description of the present embodiment, the same parts as those in the first embodiment are denoted by the same reference numerals as those in FIGS.

First, a schematic configuration of a hot rolling facility 1100 that is a control target of the present embodiment will be described. FIG. 11 is a diagram illustrating an example of a schematic configuration of the hot rolling facility 1100.
In FIG. 11, a hot rolling facility 1100 includes a heating furnace 1101, a rough rolling mill 1102, a finish rolling entry thermometer 1103, a finish rolling mill 1104, inter-stand sprays 1105a to 1105f, and a finish rolling exit temperature. A total 1106, a run-out table 1107, a strip shear 1108, and a coil winding device 1109 are provided.

The heating furnace 1101 heats the slab 1110 conveyed from a rolled steel plate production line (not shown) to a predetermined temperature.
The rough rolling mill 1102 performs rough rolling on the slab 1110 heated by the heating furnace 1101 and supplied to the hot rolling line.

A finishing rolling entry thermometer 1103 measures the temperature of the material to be rolled conveyed up to the front of the finishing mill 1104. The finishing mill 1104 continuously finish-rolls the material to be rolled. In the present embodiment, a case where the finish rolling mill 1104 is configured using seven rolling stands F1 to F7 will be described as an example.
The inter-stand sprays 1105a to 1105f are arranged between the rolling stands F1 to F7, and inject cooling water onto the material to be rolled that has been finish-rolled.

The finish rolling delivery thermometer 1106 measures the temperature of the material to be rolled that has been finish-rolled by the finish rolling mill 1104.
The run-out table 1107 cools the material to be rolled that has been finish-rolled by the finish rolling mill 1105.
The coil winding device 1109 is generally called a coiler, and has a function of winding the material to be rolled that has been cooled by the run-out table 1107. In the example shown in FIG. 11, the material to be rolled is alternately wound by the two coil winding devices 1109a and 1109b.

The strip shear 1108 has a function of cutting a material to be rolled when the material to be rolled having a predetermined length is wound on the coil winding device 1109.
The hot rolling facility is not limited to that shown in FIG. For example, a descaling device that removes the scale formed on the surface of the material to be rolled may be provided between the rough rolling mill 1102 and the finish rolling entry thermometer 1103. Moreover, the plate | board thickness meter which measures the plate | board thickness of a to-be-rolled material may be equipped in the entrance side, exit side, etc. of the rolling stands F1-F7.

The configuration of the process control system of the present embodiment is that in which the hot stove 400 of the process control system 10 shown in FIG.
In the present embodiment, the hot rolling facility 1100 as described above is a control target, and the “roll gap difference ΔGap _j standard deviation σ (ΔGap _j ) and rolling load difference ΔLoad _j between adjacent rolling stands F1 to F7. The case where the standard deviation σ (ΔLoad _j ) "is used as the control amount will be described as an example. Therefore, in the present embodiment, the controller 300 includes a standard deviation of the difference DerutaGap _j of "roll gap of the rolling stand F1~F7 adjacent σ (ΔGap _j), the standard deviation of the difference DerutaLoad _j of rolling load σ (ΔLoad _j The hot rolling facility 1100 is controlled so that the calculated value (or measured value) of “)” becomes the target value.
Here, in this embodiment, the values of the subscripts are “1”, “2”, “3”, “4”, “5”, “6”, “7”. F1 ”,“ F2 ”,“ F3 ”,“ F4 ”,“ F5 ”,“ F6 ”, and“ F7 ”.

The controller 300 stores data of operation results in the hot rolling facility 1100 in the result database 200. In the present embodiment, the result database 200 stores, for example, data of operation results for the latest five days every 1 [sec].
The operation result data includes, for example, output values (operation result values) output from various measuring devices installed in the hot rolling facility 1100 by operating the hot rolling facility 1100 and output values thereof. The input value and the target value for the hot rolling facility 1100 at the time are included.

FIG. 12 is a diagram illustrating an example of a functional configuration of the information processing apparatus 1200. The information processing apparatus 1200 according to the present embodiment uses a multi-purpose GA (genetic algorithm) to solve a multi-objective optimization problem using the thicknesses h _{1 to} h ₆ on the outlet side of the rolling stands F1 to F6 as design variables. A plurality of Pareto optimal solutions (evaluation amounts) are obtained in advance from the results, and when the operation conditions are changed, one Pareto optimal solution that matches the changed operation conditions is selected from the plurality of Pareto optimal solutions, The selected Pareto optimal solution is output to the controller 300 as a target value.

  Here, the operation condition includes a state quantity of the hot rolling facility 1100 and an operation policy of the hot rolling facility 1100. The state quantity of the hot rolling facility 1100 is an amount (for example, temperature or rolling load) measured directly or indirectly momentarily by operating the hot rolling facility 1100. On the other hand, the operation policy of the hot rolling facility 1100 is determined by the operator based on the production plan and production management.

<Configuration of Target Value Optimization Device 1210 and Process Simulator 1250>
(Objective function setting unit 1211)
The objective function setting unit 1211 sets an objective function based on an operation from the operator. In the present embodiment, a plurality of evaluation quantities that are indices for evaluating the performance of the hot rolling facility 1100 are used as an objective function. Specifically, in the present embodiment, the case where the objective function is represented by the following equations (23) and (24) will be described as an example.
σ (ΔGap _j ) → Min (23) σ (ΔLoad _j ) → Min (24)

In the equations (23) and (24), j is for specifying a rolling stand. In this embodiment, since tandem rolling is performed with seven stands (since the number of rolling stands is seven), j is expressed as the following equation (25).
j = 2, 3,..., 7 (25)
In Expressions (23) and (24), “→ Min” represents that the minimum value on the left is the optimum value.
In the equations (23) and (24), σ (X) represents the standard deviation of X (here, “ΔGap _j ”, “ΔLoad _j ”).

In the equation (23), ΔGap _j is a difference in roll gap [mm] between the adjacent rolling stands F1 to F7. ΔGap _j is specifically expressed by the following equation (26).
ΔGap _j = Gap _j −Gap _j−1 (26)
In the equation (24), ΔLoad _j is a difference in rolling load [N] between the adjacent rolling stands F1 to F7. ΔLoad _j is specifically expressed by the following equation (27).
ΔLoad _j = Load _j -Load _j-1 (27)
The objective function setting unit 1211 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 1200.

(Restriction condition setting unit 1212)
The constraint condition setting unit 1212 sets a constraint condition when setting a target value based on an operation from the operator. In the present embodiment, the constraint conditions of the following equations (28) and (29) are set by the constraint condition setting unit 1212.
ΔGap _j <0 (28)
ΔLoad _j <0 (29)
The constraint condition setting unit 1212 can be realized by using an HD or the like provided in the information processing apparatus 1200, for example.

(Pareto optimal solution deriving unit 1213)
The Pareto optimal solution deriving unit 1213 applies multi-objective GA, solves the multi-objective optimization problem for the objective function set by the objective function setting unit 1211, and derives a plurality of Pareto optimal solutions. The functions of the Pareto optimal solution deriving unit 1213 will be described below. The Pareto optimal solution deriving unit 1213 is realized by using, for example, a CPU, ROM, RAM, HD, interface, and the like provided in the information processing apparatus 1200.

The individual generation unit 1213a sets two sets (current generation and next generation) into which N individuals enter, and randomly generates N individuals in the current generation.
FIG. 13 is a diagram illustrating an example of the configuration of the individual 1300.
In FIG. 13, the individual 1300 is, for example, 32-bit information, and includes information on the sheet thicknesses h _{1 to} h ₆ on the outlet side of the rolling stands F _{1 to} F ₆ that are design variables.

The simulation instruction unit 1213b passes the information of the individual 1300 to the process simulator 1250 and instructs to perform the process simulation.
As shown in FIG. 12, the process simulator 1250 models a controller model 1251 that models the controller 300, a hot stove physical model 1352 that models the configuration of the hot rolling facility 1100, and a subtractor 500. A subtractor model 1353 is included, and an actual process control system is simulated by software (computer simulation).

The process simulator 1250 first acquires the actual value of the temperature [° C.] of the material to be rolled on the exit side of the heating furnace 1101. Instead of acquiring the actual value in this way, the temperature of the material to be rolled on the exit side of the heating furnace 1101 may be calculated by a heat transfer model.
Next, the process simulator 1250 calculates a predicted value of the temperature [° C.] of the material to be rolled on the exit side of the rough rolling mill 1102. For example, the predicted value of the temperature of the material on the exit side of the rough rolling mill 1102 can be calculated by multiplying the temperature of the material on the exit side of the heating furnace 1101 by a preset temperature drop coefficient. it can.

Next, the process simulator 1250 calculates the rolling temperature T _i [° C.] in each of the rolling stands F1 to F7. For example, the rolling temperature T _j (j = 2, 3, 4, 5, 6, 7) in each of the rolling stands F2 to F7 can be calculated by the following equation (30).

In the equation (30), T ₁ is a rolling temperature in the rolling stand F1, and is a predicted value of the temperature of the material to be rolled on the exit side of the rough rolling mill 1102 described above. _TF is a target value of the finish rolling temperature and is set in advance. n is the number of rolling stands, and is “7” in this embodiment.

Next, the process simulator 1250 calculates the rolling load [N] (predicted value) Load _i in each of the rolling stands F1 to F7 (i = 1, 2, 3, 4, 5, 6, 7). For example, the rolling load Load _i in each of the rolling stands F1 to F7 can be calculated by the following equations (31) to (33).

In the formula (31), b is a sheet width [mm] of the material to be rolled. In the equations (31) to (33), h _i is the plate thickness [mm] of the material to be rolled on the exit side of the rolling stand i. The thickness h _{i of the} material to be rolled on the exit side of the rolling stand i is information included in the information of the individual 1300 passed from the simulation instruction unit 1213b. H _i is the rolling stands F _i is the thickness of the rolled material inlet side [mm] of the same value as the thickness h _i-1 of the outlet side of the rolled material in the rolling stand F _i-1. In the equation (31), k _Pmi is the average deformation resistance [kg / mm ² ] in the rolling stand i, and is represented by the following equation (34), for example.

In the formula (34), a ₀ , a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , m, and n are constants [−] determined for each steel type. Moreover, c is carbon content% [-] and is a constant. Further, ε is a logarithmic strain [−], and ε ′ is an average strain rate [1 / sec]. T _i is the rolling temperature T _i [° C.] in each of the rolling stands F1 to F7 obtained as described above.
Further, R ′ _i in the equations (32) and (33) is a flat roll radius [mm], and is represented by the following equation (35), for example.

In the formula (35), R is a roll radius [mm]. Expression (35) includes Load _i . Therefore, in order to calculate the rolling load Load _i in each of the rolling stands F1 to F7, it is necessary to solve a two-dimensional simultaneous equation in which the rolling load Load _i and the flat roll radius R ′ _i in each of the rolling stands F1 to F7 are unknown. There is. In addition, since the method of calculating the rolling load Load _i in each of the rolling stands F1 to F7 as described above is described in Non-Patent Document 2, the detailed description thereof is omitted here.

When the rolling load Load _i at each of the rolling stands F1 to F7 is calculated as described above, the process simulator 1250 calculates the difference ΔLoad _j between the rolling loads of the adjacent rolling stands F1 to F7 by the equation (27). In the following description, the “rolling load difference ΔLoad _j between adjacent rolling stands F1 to F7” is referred to as “rolling load difference ΔLoad _j ” as necessary.
Then, the process simulator 1250 calculates a standard deviation σ (ΔLoad _j ) of the rolling load difference ΔLoad _j .

Next, the process simulator 1250 calculates a roll gap [mm] (predicted value) Gap _i in each of the rolling stands F1 to F7. For example, the roll gap Gap _i in each of the rolling stands F1 to F7 can be calculated by the following equation (36).

In Expression (36), h _C is a target value [mm] of the sheet thickness of the rolled material after finish rolling, that is, the sheet thickness h ₇ of the rolled material on the exit side of the rolling stand F7. K is a mill rigidity coefficient [N / mm], and Gap ₀ is a roll gap zero point correction term [mm] for calculating a reference roll gap as 0 (zero). Q _i is the roll bending force [N] in the rolling stand F _i . A method for calculating the roll gap Gap _i in each of the rolling stands F1 to F7 as described above is also described in Non-Patent Document 2.

When the roll gap Gap _i at each of the rolling stands F1 to F7 is calculated as described above, the process simulator 1250 calculates the difference ΔGap _j between the roll gaps of the adjacent rolling stands F1 to F7 using the equation (26). In the following description, if necessary "difference DerutaGap _j of the roll gap of the rolling stand F1~F7 adjacent" is referred to as "roll gap difference DerutaGap _j" Then, the process simulator 1250, the roll gap difference DerutaGap _j A standard deviation σ (ΔGap _j ) is calculated.

As described above, when the standard deviation σ (ΔLoad _j ) of the rolling load difference and the roll gap difference standard deviation σ (ΔGap _j ) are calculated, the process simulator 1250 can calculate the rolling load difference ΔLoad _j and the standard deviation of the rolling load difference. σ (ΔLoad _j ), roll gap difference ΔGap _j , and standard deviation σ (ΔGap _j ) of the roll gap difference are output to the target value optimization device 1210 (simulation result acquisition unit 1213c).
Simulation result acquisition unit 1213c has voted process simulator 1250, rolling load difference DerutaLoad _j, the standard deviation of the rolling load difference σ (ΔLoad _j), the roll gap difference DerutaGap _j, and standard deviations of the roll gap difference sigma a (ΔGap _j) Get as a quantity.
The fitness deriving unit 1213d derives (calculates) the fitness from the “standard deviation σ (ΔLoad _j ) of rolling load difference and standard deviation σ (ΔGap _j ) of roll gap difference” acquired by the simulation result acquiring unit 1213c. To do. Since the fitness can be derived by a known method in multipurpose GA, detailed description thereof is omitted.

The constraint condition determination unit 1213e is configured such that the “rolling load difference ΔLoad _j and roll gap difference ΔLoad _j ” acquired by the simulation result acquisition unit 1213c are set by the constraint condition setting unit 1212 (Equation (28), ( 29) It is determined whether or not the formula) is satisfied.
The fitness determining unit 1213f selects the fitness derived by the fitness deriving unit 1213d when the constraint determining unit 1213e determines that the constraint is satisfied. On the other hand, when it is determined by the constraint condition determination unit 1213e that the constraint condition is satisfied, the fitness determination unit 1213f cancels the fitness derived by the fitness derivation unit 1213d and sets the minimum value as the fitness To do.

The above processing of the simulation instruction unit 1213b, the simulation result acquisition unit 1213c, the fitness deriving unit 1213d, the constraint condition determining unit 1213e, and the fitness determining unit 1213f is performed for each of the N individuals generated by the individual generating unit 1213a. Is done individually.
The Pareto optimal solution solving unit 1213g performs crossover, mutation, or copying with a certain probability for the individual generated by the individual generation unit 1213a, and stores the result in the next generation. When the number of next-generation individuals becomes N by this processing, the Pareto optimal solution solving unit 1213g deletes the current-generation individual and moves all the next-generation individuals to the current generation.

The operations of the simulation instruction unit 1213b, the simulation result acquisition unit 1213c, the fitness deriving unit 1213c, the constraint condition determining unit 1213d, the fitness determining unit 1213e, and the Pareto optimal solution solving unit 1213f become the maximum generation number G. Repeat until. Then, the Pareto optimal solution solving unit 1213g finally has an evaluation amount (rolling load difference ΔLoad _j , standard deviation of the rolling load difference) of the individuals having the highest fitness among the individuals included in the current generation. σ (ΔLoad _j ), roll gap difference ΔGap _j , standard deviation σ (ΔGap _j ) of roll gap difference, and individuals corresponding to them (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆ ) Is output to the Pareto optimal solution storage unit 1214.
The Pareto optimal solution storage unit 1214 can be realized by using, for example, an HD provided in the information processing apparatus 1200.

(Operation policy acquisition unit 1215)
Returning to the description of FIG. 12, the operation policy acquisition unit 1215 inputs information related to the operation policy input by the operator.
The operation policy acquisition unit 1215 can be realized by using, for example, a CPU, ROM, RAM, and HD (hard disk) provided in the information processing apparatus 1200.

(Plant state quantity acquisition unit 1216)
The plant state quantity acquisition unit 1216 acquires the state quantity (actual value) of the hot rolling facility 1100 from various measuring instruments provided in the hot rolling facility 1100.
As described above, the state quantity of the hot rolling facility 1100 is an amount measured directly or indirectly from moment to moment by operating the hot rolling facility 1100. Note that the state quantity of the hot rolling facility 1100 is used by the operation condition change determination unit 1217 to determine whether or not the operation condition has been changed.
The plant state quantity acquisition unit 1216 can be realized by using, for example, a CPU, a ROM, a RAM, and an HD provided in the information processing apparatus 1200.
(Operation condition change determination unit 1217)
The operation condition change determination unit 1217 operates based on the operation policy acquired by the operation policy acquisition unit 1215 and the state quantity (actual value) of the hot rolling facility 1100 acquired by the plant state quantity acquisition unit 1216. It is determined whether or not the operating condition of the hot rolling facility 1100 has been changed by determining whether or not at least one of the policy and the state quantity of the hot rolling facility 1100 has been changed. In the present embodiment, the operation policy and the state quantity of the hot rolling facility 1100 are included in the operating conditions of the hot rolling facility 1100. Therefore, when at least one of the operation policy and the state quantity of the hot rolling facility 1100 is changed, the operating condition of the hot rolling facility 1100 is changed.
The operating condition change determination unit 1217 can be realized by using, for example, a CPU, ROM, RAM, HD, and the like provided in the information processing apparatus 1200.

(Pareto optimal solution candidate display section 1218)
When the operation condition change determination unit 1217 determines that the operation condition of the hot rolling facility 1100 has been changed, the Pareto optimal solution candidate display unit 1218 has a plurality of Pareto optimal solution stored in the optimal Pareto solution storage unit 1214. A GUI for causing the operator to select (that is, a target value) is displayed on a display (for example, a liquid crystal display) connected to the information processing apparatus 1200. In the GUI displayed here, for example, as shown in FIG. 7, the change contents of the operation condition, the Pareto optimal solution (name), and the evaluation amount σ (ΔLoad _j ) for determining the value of the Pareto optimal solution , Σ (ΔGap _j ) and superiority or inferiority of the evaluation quantities σ (ΔLoad _j ), σ (ΔGap _j ). Similarly to the GUI shown in FIG. 7, in the GUI displayed here, both the user manually selects one of the Pareto optimal solutions and the information processing apparatus 1200 automatically selects. To be able to do that.
The pareto optimal solution candidate display unit 1218 can be realized by using, for example, a CPU, ROM, RAM, VRAM, HD, and the like provided in the information processing apparatus 1200.

(Pareto optimal solution acquisition determination unit 1219)
The Pareto optimal solution acquisition determination unit 1219 determines whether the selection of the Pareto optimal solution is designated to be performed manually or automatically by the operator's operation on the GUI.
The Pareto optimum solution acquisition determination unit 1219 can be realized by using, for example, a CPU, ROM, RAM, HD (hard disk), and the like provided in the information processing apparatus 1200.

(Weighting factor deriving unit 1220)
When the Pareto optimal solution acquisition determination unit 1219 determines that the Pareto optimal solution selection is automatically performed, the weighting factor deriving unit 1220 includes information on the changed operating conditions (operation policy acquisition unit 1215 and plant state / evaluation amount). based on the information) obtained at least from one of acquisition unit 1216, a standard deviation of the rolling load difference σ (ΔLoad _j), the weight vector W for each of the standard deviations of the roll gap difference σ (ΔGap _j) "( The weight vector W is expressed by the following equation (37).
W = [W11, W12] (37)
W11 is a weight vector (weight coefficient) of the standard deviation σ (ΔLoad _j ) of the rolling load difference, and W12 is a weight vector (weight coefficient) of the standard deviation σ (ΔGap _j ) of the roll gap difference. The values of the weight vectors (weight coefficients) W11 and W12 are 0 or more and 1 or less, and their sum is 1.
The weighting factor deriving unit 1220 can be realized by using, for example, a CPU, ROM, RAM, and HD provided in the information processing apparatus 1200.

(Evaluation Function Deriving Unit 1221)
The evaluation function deriving unit 1221 includes the weight vector W derived by the weight coefficient deriving unit 1220 and a plurality of Pareto optimal solution values (σ (ΔLoad _j ), σ (ΔGap _j ) stored in the Pareto optimal solution storage unit 1214. )), The evaluation function Q is derived (calculated) by the following equation (38), and the load sum of the weight vectors (weight coefficients) W11 and W12 is obtained. In this case, the evaluation function deriving unit 1221, each evaluation quantity _{σ (ΔLoad j), σ (} ΔGap j) is to be substantially of the same range of variation, each evaluation quantity σ (ΔLoad _j), with respect to σ (ΔGap _j) Normalization process. Therefore, σ (ΔLoad _j ) and σ (ΔGap _j ) in the equation (38) indicate normalized evaluation quantities.

The evaluation function deriving unit 1221 can be realized by using, for example, a CPU, ROM, RAM, and HD provided in the information processing apparatus 1200.
(Pareto optimal solution selection unit 1222)
The Pareto optimal solution selection unit 1222 is based on a plurality of evaluation functions Q derived by the evaluation function deriving unit 1221, and a plurality of Pareto optimal solutions (σ (ΔLoad _j ), σ stored in the Pareto optimal solution storage unit 1214. (ΔGap _j )) is selected from the most appropriate Pareto optimal solution (that is, the Pareto optimal solution that minimizes the load sum of the weight vector (weighting coefficient) after the operation condition is changed). Specifically, in the present embodiment, the Pareto optimal solution selection unit 122 selects the minimum evaluation function Q from among the plurality of evaluation functions Q derived by the evaluation function deriving unit 121, and the Pareto corresponding to the evaluation function Q Select the optimal solution.
The Pareto optimal solution selection unit 1222 can be realized by using, for example, a CPU, ROM, RAM, HD, and the like provided in the information processing apparatus 1200.

(Target value changing unit 1223, target value storage unit 1224)
When the Pareto optimal solution acquisition determination unit 1219 determines that the Pareto optimal solution selection is to be manually performed, the target value changing unit 1223 is selected by the operator using the GUI displayed by the Pareto optimal solution candidate display unit 1218. The Pareto optimal solution and the individuals (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆ ) corresponding to the Pareto optimal solution are read from the Pareto optimal solution storage unit 1214.
On the other hand, when the Pareto optimal solution acquisition determination unit 1219 determines that the Pareto optimal solution is automatically selected, the target value changing unit 1223 selects the Pareto optimal solution selected by the Pareto optimal solution selection unit 1222 and the Pareto Individuals (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆ ) corresponding to the optimal solution are read from the Pareto optimal solution storage unit 1214.
Then, the target value changing unit 1223 and the Pareto optimal solution stored in the target value storage unit 1224 and the individuals (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆₎ corresponding to the Pareto optimal solution. ) Is rewritten to the read information.
The target value changing unit 1223 can be realized by using, for example, a CPU, a ROM, a RAM, and an HD (hard disk) provided in the information processing apparatus 1200. The target value storage unit 1224 can be realized by using, for example, an HD provided in the information processing apparatus 1200.

(Target value output unit 1225)
The target value output unit 1225 includes Pareto optimal solutions stored in the target value storage unit 1224 and individuals (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆ ) corresponding to the Pareto optimal solutions. When updated, the updated Pareto optimal solution and the individuals (h ₁ , h ₂ , h ₃ , h ₄ , h ₅ , h ₆ ) corresponding to the Pareto optimal solution are output to the controller 300 as target values. . Thereby, the target value at the time of controlling the hot rolling equipment 1100 is changed.
The target value output unit 1225 can be realized by using, for example, a CPU, ROM, RAM, HD, interface, and the like provided in the information processing apparatus 1200.
Even if it does as mentioned above, the effect demonstrated in 1st Embodiment can be acquired. Also in the present embodiment, various modifications described in the first embodiment can be employed.

In the present embodiment, the case where the thickness h _{i of the} material to be rolled on the exit side of the rolling stand i is used as a design variable has been described as an example. However, the distribution coefficient α _i shown in the following equation (39) is used. It may be used as a design variable.

In the equation (39), h _B is the plate thickness [mm] of the material to be rolled after rough rolling, and has the same value as the plate thickness on the entry side of the rolling stand F1. h _C is a target value [mm] of the thickness of the material to be rolled after finish rolling, and is the same value as the thickness h ₇ on the exit side of the rolling stand F7.

The embodiment of the present invention described above can be realized by a computer executing a program. Further, a means for supplying the program to the computer, for example, a computer-readable recording medium such as a CD-ROM recording such a program, or a transmission medium for transmitting such a program may be applied as an embodiment of the present invention. it can. A program product such as a computer-readable recording medium that records the program can also be applied as an embodiment of the present invention. The programs, computer-readable recording media, transmission media, and program products are included in the scope of the present invention.
In addition, the embodiments of the present invention described above are merely examples of implementation in carrying out the present invention, and the technical scope of the present invention should not be construed as being limited thereto. Is. That is, the present invention can be implemented in various forms without departing from the technical idea or the main features thereof.

DESCRIPTION OF SYMBOLS 10 Process control system 100, 1200 Information processing apparatus 110, 1210 Target value optimization apparatus 111, 1211 Objective function setting part 112, 1212 Restriction condition setting part 113, 1213 Pareto optimal solution derivation part 114, 1214 Pareto optimal solution memory | storage part 115, 1215 Operation policy acquisition unit 116, 1216 Plant state / evaluation amount acquisition unit 117, 1217 Operation condition change determination unit 118, 1218 Pareto optimal solution candidate display unit 119, 1219 Pareto optimal solution acquisition determination unit 120, 1220 Weight coefficient deriving unit 121, 1221 Evaluation function derivation unit 122, 1222 Pareto optimum solution selection unit 123, 1223 Target value change unit 124, 1224 Target value storage unit 125, 1225 Target value output unit 150, 1250 Process simulator 200 Performance database 300 Controller 00 hot air oven 500 subtractor 1100 hot rolling mill

Claims (13)

A target value optimization device for setting a target value for controlling the operation of the plant;
A controller for controlling the operation of the plant so as to be a target value set by the target value optimization device;
A plurality of indicators that are used for computer simulation of the operation of the plant and the operation of the controller so that the target value set by the target value optimization device is achieved, and as a result, evaluate the performance of the plant. A process control system having a process simulator that outputs the value of the evaluation amount to the target value optimization device,
The target value optimization device derives a plurality of Pareto optimal solutions by solving an optimization problem using the plurality of evaluation quantities as objective functions, using the values of the plurality of evaluation quantities output from the process simulator. Pareto optimal solution derivation means,
A Pareto optimal solution storage unit that stores a plurality of Pareto optimal solutions derived by the Pareto optimal solution derivation unit and a target value corresponding to the Pareto optimal solution in a storage medium in association with each other;
An operation condition change determination means for determining whether or not an operation condition including at least one of the state quantity of the plant and the operation policy of the plant is changed;
If it is determined by the operation condition change determination means that the operation condition has been changed, one of a plurality of Pareto optimal solutions stored by the Pareto optimal solution storage means is selected according to the changed operation conditions. Pareto optimal solution selection means to select,
A target value output means for outputting a target value associated with the Pareto optimal solution selected by the Pareto optimal solution selection means and stored by the Pareto optimal solution storage means to the controller. Control system. The Pareto optimal solution derivation means includes individual generation means for randomly generating a plurality of individuals each including information for specifying the target value;
Target value conversion means for converting individual information generated by the individual generation means into a target value;
Simulation instruction means for passing the target value converted by the target value conversion means to the process simulator and instructing the process simulator to perform computer simulation of the operation of the plant and the operation of the controller;
Simulation result acquisition means for acquiring values of the plurality of evaluation quantities for each of the plurality of individuals from the result of computer simulation executed by the process simulator according to an instruction from the simulation instruction means;
Fitness deriving means for deriving the fitness of the plurality of individuals according to the value of the evaluation amount acquired by the simulation result acquiring means;
Pareto optimal solution finding means that selects at least two of the plurality of individuals according to the fitness, and adopts an evaluation amount corresponding to the selected individual as a Pareto optimal solution,
The Pareto optimal solution storage means stores a plurality of Pareto optimal solutions obtained by the Pareto optimal solution search means and target values corresponding to the Pareto optimal solutions in a storage medium in association with each other. The process control system according to claim 1. The Pareto optimal solution derivation means includes individual generation means for randomly generating a plurality of individuals each including information for specifying the target value;
Simulation instruction means for passing the individual generated by the individual generation means to the process simulator as a target value and instructing the process simulator to perform computer simulation of the operation of the plant and the operation of the controller; ,
Simulation result acquisition means for acquiring values of the plurality of evaluation quantities for each of the plurality of individuals from the result of computer simulation executed by the process simulator according to an instruction from the simulation instruction means;
Fitness deriving means for deriving the fitness of the plurality of individuals according to the value of the evaluation amount acquired by the simulation result acquiring means;
Pareto optimal solution finding means that selects at least two of the plurality of individuals according to the fitness, and adopts an evaluation amount corresponding to the selected individual as a Pareto optimal solution,
The Pareto optimal solution storage means stores a plurality of Pareto optimal solutions obtained by the Pareto optimal solution search means and target values corresponding to the Pareto optimal solutions in a storage medium in association with each other. The process control system according to claim 1. When it is determined by the operation condition change determination means that the operation condition has been changed, Pareto optimal solution candidate display means for displaying information on a plurality of Pareto optimal solutions stored by the Pareto optimal solution storage means on a display device is provided. And
The said Pareto optimal solution selection means selects the Pareto optimal solution designated by the operator based on the information of the several Pareto optimal solution displayed by the said Pareto optimal solution candidate display means. The process control system according to any one of the above. A weighting factor deriving unit for deriving a weighting factor corresponding to the changed operating condition when the operating condition change determining unit determines that the operating condition has been changed;
Load sum deriving means for deriving the load sum of the weighting coefficients derived by the weighting coefficient deriving means for each of the plurality of Pareto optimum solutions stored by the Pareto optimum solution storing means with respect to the evaluation amount. ,
The said Pareto optimal solution selection means selects the Pareto optimal solution in which the load sum of the weighting coefficient derived | led-out by the said load sum deriving means becomes the minimum, The any one of Claims 1-3 characterized by the above-mentioned. Process control system.   The process control system according to claim 1, wherein the Pareto optimal solution deriving unit derives a plurality of Pareto optimal solutions offline. A target value optimization step for setting a target value for controlling the operation of the plant;
A control step of controlling the operation of the plant using a controller so as to be the target value set by the target value optimization step;
A plurality of indicators that are computer simulations of the operation of the plant and the operation of the controller so as to achieve the target value set by the target value optimization step, and as a result, evaluate the performance of the plant A process control method having a simulation step for deriving a value of an evaluation amount of
The target value optimization step derives a plurality of Pareto optimal solutions by solving an optimization problem using the plurality of evaluation amounts as objective functions, using the values of the plurality of evaluation amounts derived in the simulation step. A Pareto optimal solution derivation step,
A Pareto optimal solution storage step of storing a plurality of Pareto optimal solutions derived by the Pareto optimal solution derivation step and a target value corresponding to the Pareto optimal solution in a storage medium in association with each other;
An operation condition change determination step for determining whether or not an operation condition including at least one of the state quantity of the plant and the operation policy of the plant has been changed,
When it is determined that the operation condition has been changed by the operation condition change determination step, any one of the plurality of Pareto optimal solutions stored by the Pareto optimal solution storage step is determined according to the changed operation condition. A Pareto optimal solution selection step to be selected;
A target value output step of outputting the target value associated with the Pareto optimal solution selected by the Pareto optimal solution selection step and stored by the Pareto optimal solution storage step to the controller. Control method. The Pareto optimal solution derivation step includes an individual generation step of randomly generating a plurality of individuals each including information for specifying the target value;
A target value conversion step of converting the individual information generated by the individual generation step into a target value;
The target value converted by the target value conversion step is passed to the process simulator that executes the simulation step, and the process simulator is instructed to perform computer simulation of the operation of the plant and the operation of the controller. Simulation instruction step to perform,
From the result of the computer simulation executed by the process simulator according to the instruction by the simulation instruction step, a simulation result acquisition step for acquiring the values of the plurality of evaluation amounts for each of the plurality of individuals,
A fitness deriving step for deriving the fitness of the plurality of individuals according to the value of the evaluation amount acquired by the simulation result acquiring step;
Selecting at least two of the plurality of individuals according to the fitness, and adopting a Pareto optimal solution solving step that adopts an evaluation amount corresponding to the selected individual as a Pareto optimal solution,
In the Pareto optimal solution storage step, a plurality of Pareto optimal solutions obtained by the Pareto optimal solution solving step and a target value corresponding to the Pareto optimal solution are correlated and stored in a storage medium. The process control method according to claim 7. The Pareto optimal solution derivation step includes an individual generation step of randomly generating a plurality of individuals each including information for specifying the target value;
Passing the individual generated by the individual generation step as a target value to a process simulator for executing the simulation step, and performing computer simulation of the operation of the plant and the operation of the controller to the process simulator. A simulation instruction step for instructing;
From the result of the computer simulation executed by the process simulator according to the instruction by the simulation instruction step, a simulation result acquisition step for acquiring the values of the plurality of evaluation amounts for each of the plurality of individuals,
A fitness deriving step for deriving the fitness of the plurality of individuals according to the value of the evaluation amount acquired by the simulation result acquiring step;
Selecting at least two of the plurality of individuals according to the fitness, and adopting a Pareto optimal solution solving step that adopts an evaluation amount corresponding to the selected individual as a Pareto optimal solution,
In the Pareto optimal solution storage step, a plurality of Pareto optimal solutions obtained by the Pareto optimal solution solving step and a target value corresponding to the Pareto optimal solution are correlated and stored in a storage medium. The process control method according to claim 7. When it is determined that the operation condition has been changed by the operation condition change determination step, there is a Pareto optimal solution candidate display step for displaying a plurality of Pareto optimal solution information stored in the Pareto optimal solution storage step on a display device. And
10. The Pareto optimal solution selection step selects a Pareto optimal solution designated by an operator based on information on a plurality of Pareto optimal solutions displayed in the Pareto optimal solution candidate display step. The process control method according to any one of the above. When it is determined that the operation condition has been changed by the operation condition change determination step, a weighting factor deriving step for deriving a weighting factor according to the changed operation condition, which is a weighting factor for the evaluation amount;
A load sum deriving step for deriving a weight sum of the weighting factors derived by the weighting factor deriving means for each of the plurality of Pareto optimum solutions stored in the Pareto optimum solution storing step with respect to the evaluation amount. ,
10. The Pareto optimal solution selection step selects a Pareto optimal solution that minimizes the load sum of the weighting coefficients derived by the load sum derivation step. 10. Process control method.   12. The process control method according to claim 7, wherein the Pareto optimal solution deriving step derives a plurality of Pareto optimal solutions offline.   A computer program for causing a computer to function as each means of the process control system according to any one of claims 1 to 6.

Images