JP2023039670A - Controller for coke production process, method and program - Google Patents

Abstract

To enable a physical amount representing a carbonized state of coke to be adjusted according to a target value and reduce variations in the physical amount representing the carbonized state of coke.SOLUTION: A controller for a coke production process 100 controls a heat input so as to bring a coke temperature to close to a target temperature. A target furnace temperature calculator 102 uses a coke temperature prediction model to predict a coke temperature on the basis of an influencer including an oven battery temperature. On the basis of an evaluation function including a term representing a difference between the coke temperature predicted by the coke temperature prediction model and the target temperature, the target furnace temperature calculator calculates a target oven battery temperature for a plurality of inter-block times to come, so that the coke temperature is adjusted according to the target temperature. A heat input calculator 103 calculates a heat input according to the target oven battery temperature calculated by the target furnace temperature calculator 102. Thus, considering how the coke temperature for each block will change depending on the oven battery temperature, it is possible to dynamically calculate the target oven battery temperature for a plurality of inter-block times to come.SELECTED DRAWING: Figure 1

Description

The present invention relates to a coke production process control device, method and program.

As a technique related to control of the coke production process, there is a technique described in Patent Document 1. In Patent Document 1, a block kiln discharge method is used, and all the coking chambers provided in the coke oven are grouped into a plurality of streets (blocks). Then, each target furnace temperature is calculated by a model from the predicted coal charge amount, predicted moisture content, planned carbonization time, and actual furnace temperature of the future coal charge, and the weighted average is taken to obtain the target furnace temperature (appropriate furnace temperature). is obtained, and the input heat amount is calculated so as to achieve the target furnace bundle temperature.

JP-A-9-302350 JP 2009-75737 A

Flower Pollination Algorithm for Global Optimization, arXiv.org, Dec 19, 2013 Model predictive control-III: Generalized predictive control (GPC) and its surroundings, Shiro Masuda, Toru Yamamoto, Masahiro Oshima, System/Control/Information, 2002, Vol46(9), pp. 578-584

However, in Patent Literature 1 described above, the target furnace temperature is determined by weighting and averaging the target furnace temperatures for each pattern, so the target furnace temperature is different from the target furnace temperature for each pattern. In other words, it is a static calculation that does not take into account what values each coke temperature will take depending on the finally determined target furnace bund temperature. For this reason, it is difficult to bring the coke temperature after carbonization closer to the target temperature and suppress variations in the coke temperature after carbonization.

The present invention has been made in view of the above points, and sets the physical quantity representing the carbonization state of coke to a value according to the target value, and suppresses variations in the physical quantity representing the carbonization state of coke. The purpose is to enable the realization of

A control device for a coke production process of the present invention is a coke oven having a plurality of carbonization chambers and a plurality of combustion chambers, wherein the physical quantity representing the carbonization state of coke is a value corresponding to a target value. A control device for a coke production process that controls heat quantity, comprising: a physical quantity prediction model for predicting the physical quantity based on a first influencing factor including a furnace temperature that is the temperature of the combustion chamber; and an actual value of the first influencing factor. and at least one of a schedule value, predicting the physical quantity, calculating a value of a first evaluation function including a term representing a difference between the predicted physical quantity and a target value of the physical quantity, and calculating a target furnace temperature calculation unit for calculating a target furnace temperature based on the value of the first evaluation function calculated; and an input heat amount calculation unit for calculating an input heat amount corresponding to the target furnace temperature calculated by the target furnace temperature calculation unit And prepare.
A method for controlling a coke production process according to the present invention is a coke oven having a plurality of carbonization chambers and a plurality of combustion chambers, wherein the physical quantity representing the carbonization state of coke is a value corresponding to a target value. A control method for a coke production process that controls heat quantity, comprising: a physical quantity prediction model for predicting the physical quantity based on a first influencing factor including furnace temperature, which is the temperature of the combustion chamber; and an actual value of the first influencing factor. and at least one of a schedule value, predicting the physical quantity, calculating a value of a first evaluation function including a term representing a difference between the predicted physical quantity and a target value of the physical quantity, and calculating a target furnace temperature calculating step for calculating a target furnace temperature based on the value of the first evaluation function calculated; a process.
The program of the present invention causes a computer to function as each part of the control device for the coke production process.

According to the present invention, it is possible to set the physical quantity representing the carbonization state of coke to a value corresponding to a target value, and to suppress variations in the physical quantity representing the carbonization state of coke.

1 is a diagram showing a functional configuration of a control device for a coke production process according to a first embodiment; FIG. It is a figure which shows schematic structure of a coke oven. It is a figure which shows the outline of the appearance of the coke oven during carbonization. It is a figure which shows the outline of the appearance of the coke oven during kiln discharge (extrusion) work. FIG. 4 is a diagram showing changes in coke temperature and furnace bundle temperature from charging to extrusion in one coking chamber; 4 is a flowchart showing processing of a target furnace temperature calculator in the first embodiment; 6 is a diagram for explaining an outline of processing in the flowchart of FIG. 5; FIG. It is a flow chart which shows processing of an input calorific value calculation part in a 1st embodiment. It is a figure which shows the relationship between furnace bed temperature, coke temperature, and input heat amount. FIG. 7 is a diagram for explaining a coke temperature prediction model in the second embodiment; FIG. 4 is a characteristic diagram showing the results of Example 1. FIG. 10 is a flow chart showing processing of a target furnace temperature calculator and an input heat amount calculator in the third embodiment. FIG. 12 is a diagram for explaining an outline of processing in the flowchart of FIG. 11; FIG. 10 is a characteristic diagram showing the results of Example 2; FIG. 4 is a diagram for explaining the influence of the delay in taking out from the kiln on the coke temperature; FIG. 10 is a diagram showing the functional configuration of a control device for coke production process according to a fourth embodiment; It is a flow chart which shows processing of a target furnace temperature calculation part, an input heat quantity calculation part, and a carbonization time correction part in a 4th embodiment. FIG. 10 is a diagram for explaining an example of a delay in taking out the kiln; FIG. 10 is a diagram for explaining another example of the delay in taking out the kiln; FIG. 10 is a characteristic diagram showing the results of Example 3; FIG. 5 is a diagram for explaining the influence of the operator's judgment on the coke temperature when the kiln-out is delayed. It is a figure for demonstrating the method of correction|amendment of a dry distillation scheduled time. FIG. 12 is a diagram showing the functional configuration of a coke production process control device according to a sixth embodiment; 14 is a flow chart showing processing of a target furnace temperature calculator, an input heat quantity calculator, and a target furnace temperature corrector in the sixth embodiment. FIG. 4 is a diagram for explaining an outline of processing for correcting a target furnace bundle temperature pattern;

Preferred embodiments of the present invention will now be described with reference to the accompanying drawings.
[First embodiment]
A schematic configuration of the coke oven 1 and an overview of the coke production process will be described with reference to FIGS. 2, 3A and 3B.
In a coke oven 1 , carbonization chambers (kilns) 2 and combustion chambers 3 are alternately arranged with oven walls 4 interposed therebetween. The carbonization chamber 2 dry-distills the charged coal to obtain coke. The combustion chamber 3 keeps the carbonization chamber 2 at a high temperature by burning the fuel gas.

In the coke production process by the coke oven 1, a so-called block kiln discharging method is adopted for the kiln discharging coal charging operation. The unloading coal charging operation consists of the operation of pushing out coke from the coking chamber 2 with an extruder (kiln unloading operation) and the subsequent operation of supplying coal to the coking chamber 2 (charging operation). In addition, below, taking out from a kiln is also called extrusion. In the block kiln method, the entire carbonization chamber 2 is divided into a plurality of blocks (hereinafter referred to as blocks). In this embodiment, it is divided into 5 ways. Specifically, as shown in 1 (carbonization chambers No. 1, 6, 11, 16, . . . ), 2 (carbonization chambers No. 2, 7, 12, 17, . 8, 13, 18 . . . ), 4 streets (carbonization chamber No. 4, 9, 14, 19 . is divided by Then, the charging order of unloading from the kiln is, for example, 1st, 3rd, 5th, 2nd, and 4th order to prevent rapid temperature drop. Also, in each street, the kiln loading work is carried out in order from the young number. The time from the timing when the coal charging work for taking out the kiln is completed in a certain way to the timing when the charging work for taking out the kiln is completed in the following way (for example, in the above example, the third way after the 1st way) is called time. . Travel time is generally around 3-6 hours. In addition, this embodiment is not limited to the block kiln extraction method. For example, if each block is treated as an individual coking chamber 2 in the following description, it can also be applied to the case where the kiln charging work is carried out in units of one coking chamber 2 .

Further, in the coke oven manufacturing process, the amount of heat input to all the combustion chambers 3 is collectively adjusted, and furnace group control is performed to control the average carbonization state of each process. That is, the amount of heat input to the coke oven 1 is controlled by operating one regulating valve 5 installed for all the combustion chambers 3 . The adjustment valve 5 is a valve for adjusting the flow rate of the mixed gas of fuel gas and combustion air. In addition, the regulating valve 5 is operated via an actuator (not shown) under the control of a control device 100 for the coke production process, which will be described later. A representative value of the temperature of all the combustion chambers 3 is called a furnace bundle temperature. For example, a plurality of combustion chambers 3 out of all the combustion chambers 3 are provided with thermometers 6 for measuring the ambient temperature of the combustion chambers 3 , and the average temperature of the combustion chambers in which the thermometers 6 are provided is defined as the furnace bundle temperature. In the present embodiment, the oven temperature, which is the temperature in the combustion chamber 3 of the coke oven 1, is assumed to be the oven bundle temperature. Note that the method of the present embodiment is not limited to the case of collectively adjusting the amount of heat input to all the combustion chambers 3 . For example, when the kiln charging work is carried out in units of one coking chamber 2, a control valve and an actuator may be installed in each combustion chamber 3 to control the dry distillation state (input heat amount) for each coking chamber 2. . Further, the thermometers 6 may be installed in each of all the combustion chambers 3 or may be installed only in some of the combustion chambers 3 . For example, thermometers 6 may be installed in all the combustion chambers 3, and the temperature of each combustion chamber 3 may be used as the furnace temperature instead of the furnace bundle temperature.

As described above, coke is extruded from the coke chamber 2 by an extruder. In the example shown in FIG. 3B, the coke 10 extruded from the coke chamber 2 by the extrusion ram 7 provided in the extruder passes through the guide wheel 9 and is discharged to a fire extinguishing wheel (not shown) arranged below the guide wheel 9. and transported to the downstream process by the fire extinguishing vehicle. In addition, the guide wheel 9 moves to the position of the carbonization chamber 2 where the kiln charging work is carried out. In FIG. 3B, after the coke 10 produced in the coke chamber 2 located below FIG. The movement of guide wheel 9 to the carbonization chamber 2 positioned above in FIG. Also, in FIG. 3B, it is assumed that a thermometer 8 for measuring the temperature of the coke 10 in a non-contact manner is installed inside the guide wheel 9 . The thermometer 8 is installed so as to view the passing path of the coke 10 inside the guide wheel 9 through a window provided in the guide wheel 9 . As described above, in the present embodiment, the temperature of the coke immediately after coming out of the coke chamber 2 is measured while the coke 10 is being taken out from the kiln (at the time of extrusion). However, the temperature of the coke 10 does not necessarily have to be measured in this manner if the temperature of the coke exiting the coke chamber 2 is measured. The temperature of the coke 10 when it is discharged from the carbonization chamber 2 (at the time of extrusion) is called the coke temperature.

The coke temperature is calculated, for example, from the measured value by the thermometer 8 shown in FIG. 3B. While the coke 10 is being pushed out of the coke chamber 2 by the extrusion ram 7, the temperature of the coke 10 sequentially discharged from the coke chamber 2 is measured with a thermometer 8, and the average temperature at each time and each position measured. (A value obtained by dividing the sum of the temperatures measured at each time and each position by the number of temperature measurements) is taken as the temperature of the coke 10 produced in the coking chamber 2 . Then, the average value of the temperatures of the coke 10 produced in the coke chambers 2 belonging to one street is taken as the coke temperature (average value of the street). The street average value is the arithmetic average value (value obtained by dividing the sum of the coke temperatures produced in the coke chambers 2 belonging to one street by the number of the coking chambers 2 belonging to the street). Since the coke temperature is preferably the temperature of the coke 10 immediately after being discharged from the coke chamber 2, the coke temperature is determined as shown in FIG. 3B. The method of determining the meter and the coke temperature itself may be, for example, what is used in coke plants, and is not limited to the above.

FIG. 1 shows the functional configuration of a control device 100 (hereinafter simply referred to as a control device) for a coke production process according to the first embodiment. Note that the hardware of the control device 100 is realized by using an information processing device that includes a processor such as a central processing unit, a main storage device, an auxiliary storage device, an input device, and an output device, for example. Further, the hardware of the processing device 300 may be implemented by a PLC (Programmable Logic Controller), or may be implemented by dedicated hardware such as an ASIC (Application Specific Integrated Circuit).
The control device 100 includes an input unit 101, a target furnace temperature calculation unit 102, an input heat amount calculation unit 103, and an input heat amount setting unit 104, and the coke temperature (that is, the coke temperature during extrusion) is set in advance. The input heat amount is controlled so that it becomes a value according to the target temperature.

The input unit 101 inputs operating data of the coke production process. The operational data includes current and past operational performance values and future operational schedule values. More specifically, the actual value of the operation includes data indicating the control content and operation state of the coke production process, for example, the measured value of each thermometer 8 described above and the temperature of the furnace bed. Influence factors that affect the coke temperature and the second influencing factor, which is an influencing factor on furnace bundle temperature such as input heat amount, and coke information before being pushed into the coke oven 1. Further, the operation schedule values include, for example, various schedule values such as a coke temperature target value (target temperature), coke oven operating rate, and coal charge amount. The storage unit 200 stores operation data of the coke production process together with information such as time that shows the chronological order. For example, the sensor values of various sensors installed to monitor the coke production process, such as the values measured by the thermometer 8, are periodically stored in the storage unit 200, thereby monitoring the state of the coke production process in real time. is possible. It is also assumed that the coke temperature (average value) obtained from the measured value of the thermometer 6 as described above is also stored in the storage unit 200 as one of the operation data (actual value of the coke temperature). Although the input unit 101 inputs operation data from the storage unit 200 as an example, the input unit 101 may input operation data from an external device via a network, or the user may directly input operation data. may be The control device 100 may include the storage section 200 .

The target furnace temperature calculation unit 102 uses the coke temperature prediction model for predicting the coke temperature based on the influence factor (first influence factor) on the coke temperature including the oven bundle temperature, and is measured as described above. A target furnace for a plurality of future time (= N × time (N is an integer of 2 or more, N = 5 in FIG. 2)) so that the coke temperature in the future becomes a value according to the target temperature (target value) Calculate the mass temperature. The target furnace temperature calculator 102 has a coke temperature prediction function and a furnace bundle temperature optimization function, details of which will be described later. The coke temperature prediction function uses a coke temperature prediction model to predict coke temperature. The furnace cascade temperature optimization function calculates a target cascade temperature based on an evaluation function including a term representing the difference between the coke temperature predicted by the coke temperature prediction model and the target coke temperature.

The input heat amount calculation unit 103 calculates the input heat amount according to the target furnace bundle temperature calculated by the target furnace temperature calculation unit 102 . The input heat amount calculation unit 103 has a furnace bundle temperature prediction function and an input heat amount optimization function, which will be described later in detail. The furnace cascade temperature prediction function predicts the cascade temperature using a furnace cascade temperature prediction model that predicts the furnace cascade temperature based on factors (second influencing factors) that affect the furnace cascade temperature, including the amount of heat input. The input heat quantity optimization function calculates the target furnace temperature calculated by the target furnace temperature calculation unit 102 based on an evaluation function including a term representing the difference between the furnace temperature predicted by the furnace temperature prediction model and the target furnace temperature. The temperature is used to calculate the heat input.

The input heat amount setting unit 104 outputs the input heat amount calculated by the input heat amount calculation unit 103 to a controller for an actuator (not shown) so as to reflect the input heat amount in the coke production process. The control device of the actuator operates the adjustment valve 5 (see FIG. 2) via an actuator (not shown) to adjust the opening of the adjustment valve 5 to the opening corresponding to the input heat amount calculated by the input heat amount calculation unit 103. to

The target furnace temperature calculator 102 and input heat quantity calculator 103 will be described in detail below.
First, the furnace bundle temperature prediction model used in the input heat quantity calculation unit 103 will be described.
The furnace bundle temperature prediction model is a model that predicts the furnace bundle temperature based on the second influencing factor including the input heat amount. In this embodiment, as shown in Equation (1), a regression model is employed that includes, as an explanatory variable, past furnace bundle temperatures rather than prediction target furnace bundle temperatures. _Tro represents the temperature of the furnace bed [°C], Q represents the amount of input heat [GJ/h], S represents the amount of coal charge [ton], and W represents the moisture content of coal [%]. Further, t represents a specified time ahead of the current time (hereinafter, the specified time is assumed to be one hour). ΔT _ro (t) as the objective variable and ΔQ(t−i), ΔT _ro (t−i), ΔS(t−i) and ΔW(t−i) as the explanatory variables are the furnace bundle temperature T _ro , input heat quantity Q, furnace bundle temperature T _ro , coal charge S, and coal moisture content W are expressed by the amount of change from one hour before.

Table 1 shows the relationship between explanatory variables (○) and objective variables (★) of the furnace bundle temperature prediction model used in this embodiment. In order to consider the time delay, actual values for multiple consecutive periods in a one-hour cycle and schedule values for one hour in the future (except for the furnace bundle temperature T _ro ) are used as explanatory variables. The bundle temperature prediction model predicts the furnace bundle temperature one hour in the future, which is the objective variable. Specifically, each item indicated by a circle from 1 hour ahead (t) to 5 hours ahead (t-6) is adopted as an explanatory variable. Since the furnace bed temperature has periodic fluctuations due to operation, it is preferable to apply a moving average filter as a pretreatment to suppress periodic fluctuations.

Specific contents of objective variables and explanatory variables in the furnace bundle temperature prediction model are described below.
<Objective variable>
ΔT _ro (t): change in furnace bundle temperature T _ro from time t−1 to time t (ΔT _ro (t)=T _ro (t)−T _ro (t−1))
<Explanatory variable>
ΔQ(ti): change in input heat quantity Q from time ti-1 to time ti (ΔQ(ti)=Q(ti)-Q(ti-1)
ΔT _ro (t−i): change in furnace bundle temperature T _ro from time t−i−1 to time t−i (ΔT _ro (t−i)=T _ro (t−i)−T _ro (t -i-1)
ΔS(ti): change in the amount of charge S from time ti-1 to time ti (ΔS(ti)=S(ti)-S(ti-1)
ΔW(ti) is the amount of change in coal moisture content W from time ti-1 to time ti (ΔW(ti)=W(ti)-W(ti-1 )

Coefficients a _i , b _i , c _i , and d _i in equation (1) are explanatory variables ΔQ(t−i), ΔT _ro (t−i), ΔS(t−i), ΔW(t - is the coefficient for i). As the coefficients a _i , b _i , c _i , and d _i , the coefficients that best match the past operation results of the coke oven 1 with the form of Equation (1) are obtained separately. ₁ The coefficients a _i , b _i , c _i , and d _i can be obtained by creating a large number of teacher data as one teacher data and executing multiple regression analysis using the teacher data.

Note that the coal charge amount S is the total value of the values in all the coking chambers 2 included in the coke oven 1 to be predicted, and the coal moisture content W is the total value of all the coking chambers 2 included in the coke oven 1 to be predicted. is the mean value of the values in In addition, the coal moisture content W is represented by the mass ratio (mass%) of coal, for example. In addition, when obtaining explanatory variables ΔQ(ti), ΔT _ro (ti), ΔS(ti), and ΔW(ti), actual values are used for past values, and future values are As for the value, the schedule value (value determined in the operation schedule) or the furnace bundle temperature T _ro (t) obtained from ΔT _ro (t) already calculated by the equation (1) is used. By updating the value of t every hour, the furnace bundle temperature T _ro (t) at each future time t is calculated. In this embodiment, Equation (1) corresponds to the furnace temperature prediction model referred to in the present invention.

In applying the present invention, the furnace bundle temperature prediction model does not necessarily have to be in this form. good. Also, the model construction method does not necessarily have to be the linear formula (linear regression model) of formula (1). model), or a prediction model based on machine learning other than formula (1), or the like.

Next, the coke temperature prediction model used by the target furnace temperature calculator 102 will be described.
Here, the time from the start of charging work in one of the coking chambers 2 to the end of taking out the kiln is called a dry distillation cycle. In the present embodiment, one cycle in the carbonization cycle is the sum of the time for each of the five consecutive kiln charging operations. The coke temperature prediction model is a model that predicts the coke temperature based on the first influencing factors including the oven bund temperature. In this embodiment, a regression model based on the physical phenomenon of a single coking chamber 2 is employed as shown in Equation (2). T _co is coke temperature [°C], T _ro is furnace bed temperature [° C], t _t is quenching time [hr], t _k is carbonization time [hr], S is coal charge [ton], W is coal moisture It represents the amount [%]. Also, nj (n-9 to n shown in FIG. 4) is a variable that specifies the periods obtained by dividing the carbonization cycle by the number of streets, and each period is the same time as the time period. In this embodiment, since the kiln charging is carried out at intervals of 5 kilns, there are 5 periods from charging to extrusion in one coking chamber 2 (five periods of n to n-4 and n-5 to n -9 periods). When the values of two variables n−j are consecutive values, the periods specified by the two variables represent consecutive periods (for example, period n and period n−1 are consecutive period). For consecutive n, n-1 periods, the firing order is 1st, 3rd, 5th, 2nd, 4th, and in 1st in n period More specifically, when the kiln charging work is carried out, the period of n is the period during which the kiln charging work is carried out in one way, and the period of time is the same as the one way time. is. In addition, the period of n−1 is the period during which the kiln charging work is carried out in 4 ways, and is the same period as the time in 4 ways. Hereinafter, the street on which the kiln charging operation is performed during the nj period will be referred to as the street corresponding to the nj period.

In equation (2), ΔT _co (n), which is the objective variable, and ΔT _ro (n), Δt _t (n), Δt _k (n), ΔS (n), and ΔW (n), which are explanatory variables, are respectively Changes in coke temperature T _co , furnace bed temperature T _ro , rolling time t _t , carbonization time t _k , coal charge S, and coal moisture content W from one cycle (five times before) in the same carbonization chamber carbonization cycle Expressed in quantity. FIG. 4 shows changes in coke temperature and furnace bed temperature from charging to extrusion in one coking chamber. FIG. 4 shows two cycles (two dry distillation cycles) in the dry distillation cycle from charging to extrusion in one coking chamber.

Table 2 shows the relationship between explanatory variables (○) and objective variables (★) of the coke temperature prediction model used in this embodiment. Using the operational data during carbonization as an explanatory variable, the coke temperature prediction model predicts the coke temperature in each coke chamber, which is the objective variable. Specifically, furnace cascade temperature and run time use respective run-by-run values from charging to extrusion. For the carbonization time, coal charge amount, and coal moisture content, values for each carbonization cycle, which are given at the time of the first coal charge in the carbonization cycle, are used.

Specific contents of objective variables and explanatory variables in the coke temperature prediction model will be described below with reference to FIG. In FIG. 4, the coke temperature in the street corresponding to the period of n is the timing at which the street immediately before the street corresponding to the period of n (the last kiln-out work) is completed, or the street corresponding to the period of n The forecast is made at the timing when (the first charging work of) starts. In this case, the periods n-9 to n-1 preceding the period n are periods in which the prediction of the coke temperature has already been completed. For example, the coke temperature prediction made just prior to period n is the coke temperature prediction at the street corresponding to period n-1. In this case, in the following description, the prediction of the coke temperature on the street for the period n-1 is performed with the periods n-1 to n-9 as the periods n-2 to n-10, respectively. The start time of the kiln removal work and coal charging work may be, for example, the start time of the treatment that is supposed to be performed first in each work according to the operation manual of the coke plant. Similarly, the end of the kiln removal work and the coal charging work may be set to the time when the last treatment in each work is finished, for example, according to the operation manual of the coke plant. In FIG. 4, the period n-4 to n including the period n is called the current dry distillation cycle, the dry distillation cycle immediately before the dry distillation cycle is called the period n-9 to n-5, and the previous Let us call it the dry distillation cycle.

<Objective variable>
ΔT _co (n): The amount of change in the coke temperature in the current carbonization cycle in a certain carbonization chamber 2 from the coke temperature in the previous carbonization cycle in the carbonization chamber 2 (ΔT _co (n) = T _co (n) - T _co (n-5))
As described above, the coke temperature cannot be obtained unless the kiln is removed from the kiln, so it is calculated based on the carbonization cycle. Also, "5" in n-5 is the number of streets, and is changed according to the number of streets (this is the same for explanatory variables).

<Explanatory variable>
ΔT _ro (n−j): ΔT _ro (n−j) is calculated when predicting the coke temperature on the street corresponding to the period n and when predicting the coke temperature on the street corresponding to the period n+1 and different.
<<For predicting the coke temperature in the street corresponding to the period of n>>
When j = 0: (Target temperature in period nj) - (Actual temperature in period nj-5) (ΔT _ro (nj) = T _ro (nj) - T _ro (n−j−5))
j= ₁ , 2, . j)=T _ro (n−j)−T _ro (n−j−5)).

<<When predicting the coke temperature in the street corresponding to the period of n+1>>
When j=0, 1 (0≤j≤1): (target core temperature in period nj) - (actual temperature in period nj-5) (ΔT _ro (n+1-j) =T _ro (n+1−j)−T _ro (n+1−j−5))
j=2 _, 3, . j)=T _ro (n+1-j)-T _ro (n+1-j-5))

Since the temperature of the furnace bed is obtained every hour, the temperature of the furnace bed here is a representative value for each time period (in this embodiment, an average value for each time period). Also, "5" in nj-5 is the number of streets and is changed according to the number of streets (this is the same for other explanatory variables). Also, the period nj-5 is the period belonging to the previous carbonization cycle, and is the period one cycle before the period nj belonging to the current carbonization cycle in the carbonization cycle. If the actual value is obtained in the period nj, the actual value is used, and if the actual value is not obtained, the target value is used.

Δt _t (n−j): The method of calculating Δt _t (n−j) is also the same as ΔT _ro (n−j), when predicting the coke temperature in the street corresponding to the period of n, and the period of n+1 When predicting the coke temperature in the street corresponding to , it is different.
<<For predicting the coke temperature in the street corresponding to the period of n>>
When j = 0: (Planned time in the period of nj) - (Actual time in the period of nj - 5) (Δt _t (nj) = t _t (nj) - t _t (n−j−5))
j = 1, 2, ... (when j ≥ 1): (Actual time in the period of nj) - (Actual time in the period of nj-5) (Δt _t (nj) = t _t (n−j)−t _t (n−j−5)).

<<When predicting the coke temperature in the street corresponding to the period of n+1>>
In the case of j = 0, 1 (0 ≤ j ≤ 1): (Time as planned in the period of nj) - (Actual time in the period of nj - 5) (Δt _t (n + 1 - j) = t _t (n+1-j)-t _t (n+1-j-5))
j = 2, 3, ... (j ≥ 2): (Actual time in the period of n + 1 - j) - (Actual time in the period of n + 1 - j - 5) (Δt _t (n + 1 - j) =t _t (n+1−j)−t _t (n+1−j−5))
Similarly to ΔT _ro (n−j), in Δt _t (n−j), if the actual value is obtained in the period n−j, the actual value is used, and if the actual value is not obtained , use the planned value.

Δt _k (n): Expected carbonization time in the current carbonization cycle of a given carbonization chamber 2−Actual carbonization time in the previous carbonization cycle of the carbonization chamber 2 (Δt _k =t _k (n)−t _k (n−5) )
ΔS(n): Amount of coal charged in the current carbonization cycle of a given carbonization chamber 2−Actual amount of coal charged in the previous carbonization cycle of the carbonization chamber 2 (ΔS=S(n)−S(n−5))
ΔW: Moisture content of coal in the current carbonization cycle of a given carbonization chamber 2−Actual moisture content of coal in the previous carbonization cycle of the carbonization chamber 2 (ΔW=W(n)−W(n−5))

In the formula (2), the coke temperature T _co (n) in the period n is calculated (predicted) at the timing when the actual values of the carbonization time, coal charge, and coal moisture content in the current carbonization cycle are obtained. If so, use the actual value; otherwise, use the schedule value.
In addition, since the coal charging amount and coal moisture content in the current carbonization cycle are obtained in the last manner (corresponding to the period of n-5) in the previous carbonization cycle, 0 is attached.
Also, in applying the present invention, the coke temperature prediction model does not necessarily have to be in the form described above. good. Also, the method of constructing the model does not necessarily have to be the linear formula of formula (2), and for example, a prediction model based on machine learning other than formula (2), a physical model, or the like may be used. In this embodiment, Equation (2) corresponds to the physical quantity prediction model of the present invention.

Coefficients a _j , b _j , c, d, and e in equation (2) are explanatory variables ΔT _ro (n−j), Δt _t (n−j), Δt _k (n), ΔS(n ), the coefficient for ΔW(n). As the coefficients a _i , b _i , c _i , and d _i , the coefficients that best match the past operation results of the coke oven 1 with the form of Equation (2) are obtained separately. For example, a set of T _ro (n−j), Δt _t (n−j), Δt _k (n), ΔS(n), and ΔW(n) obtained from past operating results of coke oven 1 The coefficients a _j , b _j , c, d, and e can be obtained by creating a large number of teacher data using the data as one teacher data and executing multiple regression analysis using the teacher data.

Next, the processing of the target furnace temperature calculator 102 in the first embodiment will be described with reference to FIGS. 5 and 6. FIG.
FIG. 5 is a flow chart showing the processing of the target furnace temperature calculator 102. As shown in FIG. The flow chart of FIG. 5 is executed from time to time, such as each time a new time arrives. The timing for starting the flow chart of FIG. 5 is the timing when the final kiln removal work on each street is completed, as described with reference to FIG. 4 . However, the timing of starting the flowchart of FIG. 5 may be the timing of starting the first charging operation in each street.

In step S501, the target furnace temperature calculation unit 102, via the input unit 101, for each of the coking chambers 2, the data of influence factors that affect the coke temperature for a plurality of times from the present to the past (hereinafter referred to as carbonization information data ). This carbonization information data (the time series of the above-mentioned influence factors) may consist of actual values only, or may include schedule values if necessary, such as when actual values are insufficient. The carbonization information data is data representing the properties of coal at the time of coal charging, the state of coke at the time of extrusion, the furnace bed temperature during carbonization, etc. Specifically, the carbonization information data includes, for example, the following data. It is a thing.
- Coke temperature, carbonization time - Coal charge, coal moisture content - Furnace temperature, drying time In the present embodiment, the coke temperature and carbonization time are obtained at the end of or after the coke is discharged from the kiln in each coke chamber 2. (As noted above, the coke temperature is measured during coke extrusion, but the final coke temperature is obtained at or after the end of the blowout operation). The charging amount is obtained at or after the charging operation in each coking chamber 2 is completed. The moisture content of coal is obtained before starting carbonization of coal (before starting coal charging operation (for example, before transportation to coke oven 1)). As described above, the temperature of the furnace bundle is the average value of the temperature of all the combustion chambers in which the thermometers 6 are installed for each time period. Also, in the carbonization information data, actual values are used as the coke temperature, coal charge amount, coal moisture content, and furnace bundle temperature. However, for example, when the reliability of the schedule value is high, the schedule value may be used instead of these actual values. Also, in the carbonization information data, both the actual value and the schedule value are used as the dry distillation time and the straight time.

In step S502, the target furnace temperature calculation unit 102 selects candidates for target furnace temperature transitions over a plurality of consecutive future times (hereinafter referred to as target furnace temperature patterns) within predetermined constraints. One or more initial values of the target furnace group temperature pattern are generated. It should be noted that the first street of the future consecutive multiple street times is preferably the next street of the current street in which the charging kiln operation is being performed. This is because the coke oven 1 has a large time constant (a time delay from when the input heat amount is changed until the change affects the coke temperature). However, the first street of the future consecutive multiple street times may be two or more streets ahead of the current street. Examples of the constraint conditions include upper and lower limit values of the temperature of the furnace bed and upper and lower limit values of the amount of change in the temperature of the furnace bed. FIG. 6( a ) shows an example of a target furnace bundle temperature pattern 601 . In addition, in FIGS. 6A and 6B, control start indicates the timing when the control by the control device 100 is started. In FIGS. 6A and 6B, the time indicated as control start is the current time. The target bund temperature pattern 601 represents the target bund temperature that changes in a step-like manner at every unfolded time _tt at a plurality of unfolded times in the future. The target furnace temperature calculation unit 102 determines, for example, the target furnace temperature for each time t _t by random numbers within the restricted range, and generates a predetermined number (for example, 50 in this example) initial value of the target furnace temperature pattern. . The initial values of the predetermined number of target furnace temperature patterns are common to all the coking chambers 2 .

In step S503, the target furnace temperature calculation unit 102 calculates a target furnace temperature pattern candidate (the initial value of the target furnace temperature pattern generated in step S502 or the target furnace temperature pattern generated in step S507, which will be described later) for each coking chamber 2. temperature pattern) and the carbonization information data (carbonization time, coal charge, coal moisture content, furnace bundle temperature, passing time) captured in step S501, the calculation of the right side of the coke temperature prediction model in Equation (2) Then, the value of the left side of the coke temperature prediction model of Equation (2) is calculated, and the future coke temperature of each coke chamber 2 is predicted from the calculated value. The future coke temperature of each coke chamber 2 is calculated within the range of the period (a plurality of consecutive future times) indicated by the target furnace bundle temperature pattern 601 .

In step S504, the target furnace temperature calculation unit 102 converts the future coke temperature of each coke chamber 2 predicted in step S503 into a street average value (hereinafter referred to as coke temperature prediction value 602 called). FIG. 6(b) shows an overview of the process of obtaining the predicted value 602 of the coke temperature. Since extrusion is carried out in one of the following ways every passing time _tt , the target furnace temperature calculator 102 obtains the predicted value 602 of the coke temperature at the passing time _tt for each passing time tt. Specifically, the predicted value 602 of the coke temperature is obtained by adding together the coke temperatures of the plurality of (13 in the example of FIG. 2) coking chambers 2 that are predicted in step S503. and divided by the number of carbonization chambers 2 belonging to the street. Although the example of using the street average value as the representative value of the future coke temperature for each street to express the predicted value 602 of the coke temperature has been described, for example, the minimum value of the future coke temperature for each street may be used. can be If the minimum value is used as the representative value of the future coke temperature for each run in this way, it is possible to more reliably prevent the occurrence of so-called undercooked coke due to the coke temperature becoming too low.

In step S505, the target furnace temperature calculator 102 calculates the evaluation function _J1 of Equation (3). The evaluation function J ₁ is a term representing the difference between the predicted value 602 of the coke temperature, which is the representative value of the future coke temperature for each street, calculated in step S504, and the target temperature 603 of the coke temperature on the street (right side 1) (see FIG. 6(b)). The second term on the right side is a term provided to satisfy the lower limit constraint that the coke temperature is set in advance. Since the coke temperature affects the quality of the coke, the coke temperature is to ensure the minimum temperature required to prevent so-called undercooking of the coke. The third term on the right side is a term provided to suppress the amount of change in the target furnace group temperature, that is, the stepwise change in the target furnace group temperature pattern 601 in FIG. 6(a). This is because rapid changes in the target furnace bundle temperature are not preferable from the standpoint of furnace operation stability and the like. The fourth term on the right side is a term provided to prevent the target reactor bundle temperature from becoming high. This is because an excessively high target furnace bed temperature is not preferable from the standpoint of furnace operation stability and cost. Note that, in the present embodiment, the evaluation function _J1 corresponds to the first evaluation function according to the invention.
J ₁ = (target temperature of coke temperature - predicted value of coke temperature) + (lower limit constraint of coke temperature) + (amount of change in target furnace battery temperature) + (target furnace battery temperature) (3)
It should be noted that although step S505 is represented by an example of calculating the evaluation function using the representative value of the future coke temperature for each street, for example, the future coke temperature of each coke chamber 2 predicted in step S503, The difference between the coke temperature in each coking chamber 2 and the target temperature may be used to calculate the first term on the right side of the evaluation function J1.

In step S506, the target furnace temperature calculator 102 determines whether or not the calculation end condition is reached. For example, the calculation end condition may be that a predetermined number of repeated calculations has been reached. Alternatively, convergence of the value of the evaluation function J ₁ during repeated calculations may be set as a calculation termination condition. If it is determined that the calculation end condition is not met, the process proceeds to step S507. If it is determined that the calculation termination condition is reached, the process proceeds to step S508.

In step S507, the target furnace temperature calculation unit 102 newly generates one or a plurality of target furnace temperature patterns that are candidates for the target furnace temperature pattern, and returns to step S503. It is preferable that the number of initial values of target furnace group temperature patterns generated in step S502 and the number of target furnace group temperature patterns newly generated in step S507 be the same. For example, based on a meta-heuristics algorithm, FPA (Flower Pollination Algorithm) (see, for example, Non-Patent Document 1), a target furnace cluster temperature pattern in the direction of decreasing the evaluation function _J1 is searched, and 50 target furnace clusters are searched for. Generate a new temperature pattern. Although FPA is used as an optimization method (algorithm for solving an optimization problem) as an example, optimization methods such as GA (Genetic Algorithm) and PSO (Particle Swarm Optimization) may be used. In this way, a new target furnace group temperature pattern is generated in step S507, and the processing of steps S503 to S505 is repeated until the calculation termination condition is reached in step S506.

In step S<b>508 , the target furnace temperature calculation unit 102 outputs to the input heat quantity calculation unit 103 the target furnace temperature pattern that minimizes the evaluation function J ₁ among the candidates for the target furnace temperature pattern. For example, when the evaluation function J ₁ is obtained by multiplying each term on the right side of the equation (3) by (−1), the target furnace temperature calculation unit ₁₀₂ calculates the target furnace group Look for temperature patterns.

As described above, the coke temperature is predicted when the furnace battery temperature is changed, and the optimal furnace battery temperature pattern that satisfies the target (minimizes the evaluation function _J1 ) is determined. As a result, as shown in FIG. 8A, the target furnace bundle temperature pattern 601 can be calculated so that the coke temperature prediction value 602 becomes a value corresponding to the coke temperature target temperature 603 .

Next, with reference to FIG. 7, processing of the input heat quantity calculation unit 103 in the first embodiment will be described.
FIG. 7 is a flowchart showing the processing of the input heat amount calculation unit 103. As shown in FIG. The flow chart of FIG. 7 is executed in a control cycle of the amount of heat input to the coke oven 1 (here, for example, a one-hour cycle). Thus, the cycle for calculating the input heat quantity shown in FIG. 7 is shorter than the cycle for calculating the target furnace bundle temperature pattern shown in FIG.
In step S701, the input heat quantity calculation unit 103 receives, via the input unit 101, data of influence factors that influence the temperature of the furnace bundle for a plurality of successive one-hour periods from the present to the past for each coking chamber 2 ( (hereinafter referred to as combustion information data) (in the example shown in Table 1, each period is one hour). This combustion information data (the time series of the above-mentioned influencing factors) may consist of actual values only, or may include schedule values if necessary, such as when actual values are insufficient. Combustion information data is data representing the amount of heat input to the coke oven and the characteristics of coal during coal charging. It includes coal charge and coal moisture content.

In step S702, the input heat amount calculation unit 103 utilizes the fact that the furnace bundle temperature prediction model can be expressed in a linear form as in Equation (1), and by generalized predictive control (GPC) (see, for example, Non-Patent Document 2) , the hourly input heat amount that minimizes the evaluation function J ₂ of equation (4) is calculated. The combustion information data taken in at step S701 is used as the input data for the furnace bundle temperature prediction model. As the target value of GPC, the target furnace temperature which is represented by the target furnace temperature pattern calculated by the target furnace temperature calculator 102 as described with reference to FIG. 5 is used. The first term on the right side of the evaluation function J ₂ is the temperature predicted by the furnace temperature prediction model (hereinafter referred to as the predicted temperature of the furnace temperature) and the target furnace temperature calculation unit 102 as described with reference to FIG. It represents the difference from the target furnace bed temperature represented by the calculated target furnace bed temperature pattern. With this term, it is possible to calculate the amount of input heat that realizes the target furnace temperature pattern calculated by the target furnace temperature calculation unit 102 . The second term on the right side is a term provided to suppress the amount of change in the amount of heat input controlled by operating the regulating valve 5, here, the amount of change in the amount of heat input in adjacent time periods. . This is because a rapid change in the input heat amount means that the control valve 5 needs to be operated by a large amount, which is not preferable from the standpoint of furnace operation stability and feasibility. Note that, in this embodiment, the evaluation function _J2 corresponds to the second evaluation function of the present invention.
J ₂ = (Target temperature of furnace bed - Predicted value of temperature of furnace bed) ² + (Amount of change in input heat quantity) ² (4)
In GPC, the furnace bundle temperature prediction model of equation (1) and the evaluation function J ₂ of equation (4) are each described in vector form, and the vector form furnace bundle temperature prediction model is substituted into the vector form evaluation function. By partially differentiating the evaluation function obtained in this manner with respect to the input heat quantity, the input heat quantity that minimizes the evaluation function J ₂ of Equation (4) is obtained.

In step S<b>703 , the input heat amount calculation unit 103 outputs the input heat amount that minimizes the evaluation function J ₂ to the input heat amount setting unit 104 . In response to this, the input heat amount setting unit 104 outputs the input heat amount calculated by the input heat amount calculation unit 103 to a controller for an actuator (not shown) so as to reflect the input heat amount in the coke production process. The control device of the actuator operates the adjustment valve 5 (see FIG. 2) via an actuator (not shown) to adjust the opening of the adjustment valve 5 to the opening corresponding to the input heat amount calculated by the input heat amount calculation unit 103. to For example, if the evaluation function J ₂ is obtained by multiplying each term on the right side of equation (4) by (-1), the input heat amount calculation unit 103 calculates the input heat amount that maximizes the evaluation function J ₂ It will be.

As described above, the furnace bundle temperature is predicted when the input heat amount is changed, and the optimum input heat amount that satisfies the target (minimizes the evaluation function _J2 ) is determined. As a result, as shown in FIG. 8(b), the predicted value 801 of the furnace cascade temperature is set to a value corresponding to the target cascade temperature pattern 601 calculated by the target furnace temperature calculation unit 102. Input heat amount 802 can be calculated. That is, it is possible to calculate the input heat amount 802 that realizes the target furnace bundle temperature pattern 601 shown in FIG. 8(a).

In this embodiment, the input heat amount calculation unit 103 uses GPC to control the input heat amount, but the present invention is not limited to this. For example, similar to the flow chart of FIG. 5, input heat amount candidates may be given, and the input heat amount that minimizes the evaluation function _J2 may be searched. Alternatively, PID control may be executed to change the amount of heat input so that the actual furnace core temperature becomes a value corresponding to the target furnace core temperature pattern 601 calculated by the target furnace temperature calculator 102 .

As described above, when controlling the amount of heat input so that the coke temperature reaches a value corresponding to the target temperature, consideration is given to how each coke temperature changes due to the first influencing factor including the furnace bundle temperature. As a result, the target bund temperature (target bund temperature pattern) for multiple times in the future can be dynamically calculated. As a result, it is possible to achieve target tracking and variation suppression of coke temperature, reduce production costs (e.g. reduction of excess heat amount due to over-dry distillation), stabilize production (e.g. risk of clogging during extrusion due to un-dry distillation or over-dry distillation) effects such as avoidance of carbonization) and quality stabilization (suppression of quality variation due to variation in carbonization conditions).

In the above-described embodiment, an example was explained in which the amount of input heat is controlled so that the coke temperature becomes a value corresponding to a predetermined target temperature, but the present invention is not limited to this. The present invention can be applied as long as the input heat quantity is controlled so that the physical quantity representing the carbonization state of coke becomes a value corresponding to a predetermined target value. As a physical quantity representing the carbonization state of coke, there is, for example, the temperature of the furnace wall 4 in addition to the coke temperature. The carbonization state of coke indicates the degree to which coal has been carbonized (pyrolyzed) in manufactured coke, and is an index representing the quality of coke.

[Second embodiment]
Next, a second embodiment will be described with reference to FIG. A modification of the coke temperature prediction model will be described in the second embodiment.
The coke temperature prediction model described in the first embodiment is a regression model based on the physical phenomenon of a single coke chamber 2, but focusing on the physical phenomenon of a single coke chamber 2 is not sufficient. There is This is because the tendency of temperature change and the tendency of change in operating conditions in adjacent coking chambers 2 cannot be grasped only by focusing on the physical phenomenon of a single coking chamber 2 . In addition, it is not possible to grasp the nonlinearity due to changes in the thermophysical property values of coal and the manner in which the temperature of the furnace bundle influences each way.

Therefore, as shown in FIG. 9, a regression model 901 for predicting the coke temperature and a big data model 902 for estimating the prediction error of the regression model 901 and outputting the estimated error are used as the models used in the target furnace temperature calculation unit 102. Prepare.
The regression model 901 is represented, for example, by Equation (2), and receives operation data as input and outputs a predicted value of the coke temperature in each coke chamber.
The big data model 902 is an estimation model based on machine learning, and in this embodiment, is constructed as a non-linear model configured by a big data machine learning method. The big data model 902 receives the operation data and the predicted value of the regression model 901, estimates the prediction error of the regression model 901, and outputs the estimated error. By adding a value obtained by multiplying the estimated error by a predetermined gain K to the predicted value of the regression model 901, the predicted value is corrected. Gradient boosting decision trees (see Patent Document 2 and Non-Patent Document 8 cited in Patent Document 2) are adopted for big data machine learning techniques. Table 3 shows the relationship between explanatory variables (○) and objective variables (★) of the big data model used in this embodiment. More specifically, the operating data (including schedule values) one line ahead and the operating data five lines before the present are used as explanatory variables.

As described above, the coke temperature prediction model of the present embodiment has a hybrid model configuration in which the prediction error of the regression model 901 is estimated by the big data machine learning method and corrected. As a result, it is possible to grasp the non-linearity caused by the tendency of temperature changes in adjacent coking chambers, the tendency of changes in operating conditions, changes in coal thermophysical properties, and how the furnace bundle temperature affects each line.
It should be noted that the big data model 902 is constructed using a gradient boosting decision tree as an example, but the big data model 902 may be constructed using a machine learning method such as a neural network or deep learning.

[Example 1]
Off-line simulation of the control method according to the second embodiment (hereinafter referred to as the present control method in the description of the first embodiment) was performed using past performance data.
FIGS. 10(a) to 10(c) show the coke temperature, furnace bundle temperature, and input heat quantity, which are the results of the off-line simulation. The dotted line in the figure is the result of this control method. Further, as a comparative example (hereinafter referred to as an operator operating method), the result of operating by operating the regulating valve 5 at the operator's discretion is indicated by a solid line.
In the case of operator operation, since it is difficult to predict the future coke temperature, it is affected by operational changes and disturbances, and as shown in FIG. is occurring.
On the other hand, in this control method, in any of the verification sections A to C, it is possible to take positive actions that predict the future compared to the operator operation method (FIGS. 10(b) and 10(c)). ), the coke temperature can be set to a value corresponding to the target temperature, and the variation in the coke temperature can be suppressed. As a result, effects such as production cost reduction, production stabilization, and quality stabilization are expected.

[Third embodiment]
In the third embodiment, when the target furnace temperature calculation unit 102 calculates the target furnace temperature (target temperature pattern) for a plurality of different times in the future, the apparent target furnace temperature pattern is calculated from the viewpoint of controlling the amount of heat input. This is an example in which the feasibility of
When it is necessary to increase the amount of change in the amount of input heat to achieve the target furnace temperature pattern, the adjustment valve 5 must be operated greatly, which is not preferable from the standpoint of furnace operation stability and feasibility. It cannot be said that the target furnace bundle temperature pattern has good feasibility. If the feasibility of the target furnace temperature pattern is not good, there is a concern that it will not be possible to accurately set the coke temperature to a value corresponding to the target temperature and suppress the variation in coke temperature. be done.
Therefore, in the third embodiment, when the target furnace temperature calculation unit 102 calculates the target furnace temperature pattern, the coke temperature prediction model is given a furnace temperature that satisfies a predetermined evaluation of the amount of change in the amount of input heat. , improve the feasibility of the target furnace bed temperature pattern.

The third embodiment will be described below. Descriptions of contents common to the first embodiment will be omitted, and differences from the first embodiment will be mainly described.
In the first embodiment, the target furnace temperature calculation unit 102 independently calculates the target furnace temperature pattern and transfers it to the input heat amount calculation unit 103 . On the other hand, in the third embodiment, the target furnace temperature calculation unit 102 calculates a target furnace temperature pattern in cooperation with the input heat amount calculation unit 103, and transfers it to the input heat amount calculation unit 103. there is In the functional configuration of the control device 100 according to the present embodiment, in contrast to the functional configuration shown in FIG. 15 and 22 of fourth and sixth embodiments to be described below based on the third embodiment. Further, in the third embodiment, part of the functions of the target furnace temperature calculator 102 and the input heat quantity calculator 103 are different from those in the first embodiment. Accordingly, illustration of the functional configuration of the control device 100 according to the present embodiment is omitted, and the input unit 101, the target furnace temperature calculation unit 102, the input heat amount calculation unit 103, and the input heat amount setting unit 104 are the same as those of the first embodiment. Detailed descriptions of the same parts as those of the control device 100 are omitted.

Processing for calculating a target furnace group temperature pattern in the third embodiment will be described with reference to FIGS. 11 and 12. FIG.
FIG. 11 is a flow chart showing the processing of the target furnace temperature calculator 102 and input heat quantity calculator 103 . The flowchart of FIG. 11 is executed in regular time cycles. The timing of starting the flowchart of FIG. 11 is, for example, the timing of completion of the kiln removal work in each street, similar to the timing of starting the flowchart of FIG. It may be the timing of starting the coal charging operation in . FIG. 12 is a diagram for explaining the outline of the processing in the flowchart of FIG. 11. As shown in FIG. The upper part of FIG. 12 shows an example of the target coke temperature 603, like the upper part of FIG. 6(a).
In step S<b>1101 , the target furnace temperature calculation unit 102 takes in carbonization information data for a plurality of times from the present to the past via the input unit 101 . Step S1101 is the same processing as step S501 in FIG.

In step S<b>1102 , the input heat amount calculation unit 103 acquires combustion information data for a plurality of consecutive one-hour periods from the present through the input unit 101 . Step S1102 is the same processing as step S701 in FIG.

In step S1103, the target furnace temperature calculation unit 102 generates one or more initial values of target furnace temperature patterns that are candidates for target furnace temperature patterns. Step S1103 is the same processing as step S502 in FIG. 5, and an example of target furnace bundle temperature pattern 601 is shown in the middle of FIG. The target furnace temperature calculation unit 102 passes the generated initial value of the target furnace temperature pattern to the input heat quantity calculation unit 103 .

In step S1104, the input heat amount calculation unit 103 utilizes the fact that the furnace bundle temperature prediction model can be expressed in a linear form as in formula (1), and uses generalized predictive control (GPC) to calculate the evaluation function J Calculate the amount of heat input to minimize ₂ . The combustion information data acquired in step S1102 is used as the input data for the furnace bundle temperature prediction model. In addition, the GPC target value includes a target furnace temperature pattern candidate (an initial value of the target furnace temperature pattern generated in step S1103 or a target furnace temperature pattern generated in step S1111, which will be described later). Use group temperature. The first term on the right side of the evaluation function _J2 is the difference between the temperature of the furnace bed predicted by the temperature prediction model of the furnace bed (predicted value of the temperature of the furnace bed) and the target temperature of the target furnace bed represented by the target temperature pattern candidate. show. As described above, the cycle for calculating the input heat amount in the input heat amount calculation unit 103 is executed in a cycle of one hour. Therefore, as shown in the lower part of FIG _. The input heat amount 1201 to be minimized can be calculated. The width of each bar in the lower row of FIG. 12 represents one hour, and the length thereof represents the input heat amount. As described in the first embodiment, an example in which GPC is used in the input heat amount calculation unit 103 will be described, but the present invention is not limited to this.

In step S1105, the input heat amount calculation unit 103 uses the input heat amount calculated in step S1104 and the combustion information data acquired in step S1102 to calculate the right side of the furnace bundle temperature prediction model of formula (1). Then, the value of the left side of the furnace bundle temperature prediction model in Equation (1) is calculated, and the predicted value of the furnace bundle temperature is calculated from the calculated value. As a result, as shown in the middle part of FIG. temperature control waveform) 1202 can be calculated. The input heat amount calculation unit 103 passes the control waveform 1202 of the furnace bundle temperature to the target furnace temperature calculation unit 102 . The predicted value of the furnace bundle temperature is calculated within the range of the period indicated by the target furnace bundle temperature pattern 601 (a plurality of consecutive times in the future).

In step S1106, the target furnace temperature calculation unit 102 converts the control waveform 1202 of the furnace bundle temperature calculated in step S1105 into an average value for each time period _tt . Since the target furnace temperature calculation unit 102 executes processing in regular time cycles, it is necessary to set parameters handled here in regular time units. Therefore, the control waveform 1202 of the furnace cascade temperature represented by the furnace cascade temperature every hour is converted into a target furnace cascade temperature pattern that changes stepwise in units of time like the target cascade temperature pattern 601 . Specifically, the average value of the temperature of the furnace bed represented by the control waveform 1202 of the temperature of the furnace bed included therein is obtained for each time period t _t , and each average value is set as the target temperature of the furnace bed for each time period. By doing so, it is possible to convert to a target furnace cluster temperature pattern.

In step S1107, the target furnace temperature calculation unit 102 calculates the right side of the coke temperature prediction model of Equation (2) using the target furnace bundle temperature pattern converted in step S1106 and the carbonization information data captured in step S1101. Then, the value of the left side of the coke temperature prediction model of Equation (2) is calculated, and the future coke temperature of each coke chamber 2 is predicted from the calculated value. The future coke temperature of each coke chamber 2 is calculated within the range of the period (a plurality of consecutive future times) indicated by the target furnace bundle temperature pattern 601 .

In step S1108, the target furnace temperature calculator 102 converts the future coke temperature of each coke chamber 2 predicted in step S1107 into an average value (predicted value of coke temperature), which is an average value for each passage. Step S1108 is the same processing as step S504 in FIG. 5 (see FIG. 6B).

In step S1109, the target furnace temperature calculator 102 calculates the evaluation function J ₁ ' of equation (5). The evaluation function J ₁ ' has the same first to third terms on the right side as the evaluation function J ₁ described in the first embodiment. The fourth term on the right side is a term provided to suppress an increase in the amount of input heat. This is because an excessively large amount of input heat is not preferable from the standpoint of furnace operation stability and cost. Also, the fifth term on the right side is a term provided to set the furnace temperature to a value corresponding to the target furnace temperature pattern represented by the target furnace temperature pattern converted in step S1106. In this embodiment, the evaluation function J ₁ ' corresponds to the first evaluation function according to the invention.
J ₁ ' = (target temperature of coke temperature - predicted value of coke temperature) + (lower limit constraint of coke temperature) + (amount of change in target furnace temperature) + (magnitude of heat input) + (target temperature of furnace - Predicted value of furnace bed temperature) (5)

In step S1110, the target furnace temperature calculator 102 determines whether or not the calculation end condition is reached. Step S1110 is the same process as step S506 in FIG. If it is determined that the calculation end condition has not been met, the process advances to step S1111. If it is determined that the calculation termination condition has been reached, the process advances to step S1112.

In step S1111, the target furnace temperature calculation unit 102 newly generates one or a plurality of target furnace temperature patterns that are candidates for the target furnace temperature pattern, and returns to step S1104. Step S1111 is the same processing as step S507 in FIG. The target furnace temperature calculation unit 102 passes the newly generated target furnace temperature pattern to the input heat amount calculation unit 103, and the input heat amount calculation unit 103 executes the processes of steps S1104 and S1105. In this way, a new target furnace bundle temperature pattern is generated in step S1111, and the processing of steps S1104 to S1109 is repeated until the calculation termination condition is reached in step S1110.

In step S1112, the target furnace temperature calculation unit 102 outputs to the input heat quantity calculation unit 103 the target furnace temperature pattern that minimizes the evaluation function J ₁ ′ among the target furnace temperature patterns converted in step S1106. For example, when the evaluation function J 1 ' is obtained by multiplying each term on the right side of the equation (5) by ( _-1 ), the target furnace temperature calculation unit 102 sets the target furnace temperature to maximize the evaluation function J ₁ '. It will search for furnace bed temperature patterns.

As described above, the coke temperature is predicted when the temperature of the furnace bed is changed, and the optimum temperature pattern of the furnace bed that satisfies the target (minimizes the evaluation function J ₁ ') is determined. As a result, as shown in FIG. 8( a ), the target furnace bundle temperature pattern 601 can be calculated so that the coke temperature prediction value 602 is set to a value corresponding to the coke temperature target temperature 603 . At this time, the temperature of the furnace bundle that satisfies a predetermined evaluation of the amount of change in the amount of input heat, which in this embodiment is the evaluation of minimizing the evaluation function _J2 of formula (4) as described above, is used. . The temperature of the furnace bundle when a predetermined evaluation of the amount of change in the amount of heat input is satisfied is obtained, for example, as follows. First, the evaluation function J ₂ of the equation (4) is given to the candidates for the target furnace bundle temperature pattern, and the input heat quantity 1201 that minimizes the evaluation function J ₂ of the equation (4) is calculated. Then, the input heat amount 1201 calculated in this manner is input to the furnace bundle temperature prediction model shown in Equation (1) to calculate a furnace bundle temperature control waveform 1202 corresponding to the input heat amount 1201 . In this way, when calculating the target furnace temperature pattern 601, the furnace temperature control waveform 1202 corresponding to the input heat amount 1201 is set to satisfy the evaluation of minimizing the evaluation function _J2 of equation (4). By using it as the furnace bed temperature, the feasibility of the target furnace bed temperature pattern can be improved. In the present embodiment, the input heat amount 1201 that minimizes the evaluation function J ₂ of Equation (4) corresponds to the input heat amount that satisfies the predetermined evaluation of the amount of change in the input heat amount referred to in the present invention. Further, in the present embodiment, the control waveform 1202 of the furnace cascade temperature corresponding to the input heat amount 1201 corresponds to the furnace cascade temperature in the case of satisfying a predetermined evaluation of the amount of change in the input heat amount referred to in the present invention. Further, in the present embodiment, the control waveform 1202 of the furnace bundle temperature corresponding to the input heat amount 1201 predicts the furnace temperature when the coke production process is controlled with the input heat amount that satisfies a predetermined evaluation of the amount of change in the input heat amount. It also corresponds to a value.

Next, the input heat amount calculation unit 103 calculates the input heat amount according to the target furnace bundle temperature calculated by the processing of FIG. 11, and reflects it in the coke production process. The description is omitted here. It should be noted that since the input heat amount calculation unit 103 has already taken in the combustion information data in step S1102, step S701 need not be performed again.

As described above, as in the first embodiment, when controlling the input heat amount so as to make the coke temperature a value corresponding to the target temperature, the first influencing factor including the furnace bundle temperature determines how each coke temperature is. It is possible to dynamically calculate the target furnace temperature (target temperature pattern) for a plurality of times in the future, taking into account how the temperature changes. As a result, it is possible to set the coke temperature to a value according to the target temperature and to suppress the variation in the coke temperature, reducing production costs (for example, reducing the amount of excess carbonization heat due to excessive carbonization), (For example, avoiding the risk of clogging during extrusion due to non-distillation or overdistillation) and stabilization of quality (suppression of quality variation due to variation in dry distillation conditions) are expected. Then, when the target furnace temperature calculating unit 102 calculates the target furnace temperature pattern, the target furnace temperature pattern By improving the feasibility of , it is possible to achieve with high accuracy the setting of the coke temperature to a value corresponding to the target temperature and the suppression of variations in the coke temperature.

[Example 2]
Off-line simulation of the control method according to the third embodiment (hereinafter referred to as this control method in the description of Example 2) was performed using past performance data. As in Example 1 (see FIG. 10), the regression model 901 for predicting the coke temperature described in the second embodiment and the prediction error of the regression model 901 are used as the model used in the target furnace temperature calculation unit 102. and a big data model 902 that estimates and outputs an estimation error.
FIGS. 13(a) to 13(c) show the coke temperature, furnace bundle temperature, and input heat quantity, which are the results of the off-line simulation. The dotted line in the figure is the result of this control method. As a comparative example, a solid line indicates the result of operation by operating the regulating valve 5 at the discretion of the operator. In Example 2, offline simulation was performed under the same conditions as in Example 1, and the comparative example in Example 2 is the same as the comparative example in Example 1.
As in Example 1, this control method enables proactive actions that predict the future (FIGS. 13B and 13C), and as shown in FIG. It is possible to set the appropriate value, and it is possible to suppress the variation in the coke temperature. As a result, effects such as production cost reduction, production stabilization, and quality stabilization are expected.
In addition, in Example 1, as shown in FIG. 10C, the amount of input heat slightly changes around March 16th. On the other hand, in Example 2, as shown in FIG. 13(c), a large change in the amount of input heat is suppressed. In FIGS. 10 and 13, specific numerical values of coke temperature, furnace bundle temperature, and input heat amount are omitted, and the vertical and horizontal scales are different between FIGS. 10 and 13. By comparison, it was confirmed that the change in the amount of input heat was suppressed.

[Fourth embodiment]
The fourth embodiment is an example in which the influence of delay or early discharge from the kiln (called take-back) is taken into account.
FIG. 14 is a diagram for explaining the influence of the delay in kiln discharge on the coke temperature, in which (a) shows time-series changes in coke temperature and (b) shows time-series changes in carbonization time. In FIG. 14(a), the line parallel to the time axis with target on the side indicates the target temperature of the coke temperature, and in FIG. 14(b), the line parallel to the axis with standard on the side A straight line indicates a standard carbonization time. In addition, the kiln-launching delay influence range indicates the way coal is charged when the kiln-launching delay occurs due to an equipment trouble or the like. On streets within the range affected by delays in kiln release, the amount of delay in kiln release will result in additional carbonization. In addition, [1], ..., [5] in the figure correspond to 1 (carbonization chamber No. 1, 6, 11, 16 ...), ..., 5 (carbonization chamber No. 5). , 10, 15, 20...). As shown in FIG. 14(b), it is assumed that, for example, before the start of process 1 [1], a delay in taking out the kiln occurs due to equipment trouble or the like. In this case, as shown in the range affected by the delay in putting out the kiln in the figure, the first street [1] and the following four streets [3], [5], [2], and [4] are charged. , and the carbonization time is lengthened by the delay in kiln removal (see the upward arrow in FIG. 14(b)). As the carbonization time becomes longer, the coke temperature rises as shown in FIG. 14(a) (see the upward arrow in FIG. 14(a)). Note that FIG. 14 explained the delay in taking out the kiln, but in the case of taking it out from the kiln, the carbonization time is shortened by the take-back of taking out the kiln, and the coke temperature drops.
Therefore, in the fourth embodiment, when predicting the coke temperature in the target furnace temperature calculating unit 102, the influence of the delay or return of the kiln discharge is taken into account so that the coke temperature can be predicted more accurately.

The fourth embodiment will be described below. Descriptions of contents common to the first to third embodiments will be omitted, and differences from the first to third embodiments will be mainly described.
In the fourth embodiment, the coke temperature prediction function of the target oven temperature calculation unit 102 has a correction function of correcting the estimated carbonization time using the delay or return time when the kiln is delayed or returned. The scheduled carbonization time is an explanatory variable of the coke temperature prediction model, and by correcting the scheduled carbonization time, the influence of the delay or return of the oven is taken into account when predicting the coke temperature in the target oven temperature calculation unit 102. be able to.

Processing for calculating a target furnace group temperature pattern in the fourth embodiment will be described with reference to FIGS. 15 to 18. FIG.
First, with reference to FIGS. 17 and 18, a specific example of correcting the estimated dry distillation time using the delay or return time when the kiln is delayed or returned will be described with reference to FIGS.
In this embodiment, there are two coke ovens 1 (A oven and B oven), and an example in which an extruder is shared between the A oven and the B oven will be described. Furnace A and furnace B have the same configuration, and both are divided into 5 kilns such as 1st to 5th kilns. For example, the order of coal charging from the furnace is 1 for A furnace, 1 for B furnace, 3 for A furnace, 3 for B furnace, 5 for A furnace, 5 for B furnace, The order is 2 for furnace A, 2 for furnace B, 4 for furnace A, and 4 for furnace B.
Further, it is assumed that the control device 100 controls the input heat amount so that the coke temperature in the A furnace follows the target temperature, targeting the A furnace.

FIG. 17 shows a case where a delay in taking out the kiln occurs in the A furnace. 17 and 18, ◯ represents one carbonization chamber 2 (hereinafter referred to as kiln) in A furnace, and ● represents one carbonization chamber 2 (hereinafter referred to as kiln) in B furnace. For both A furnace and B furnace, each street includes 13 kilns, and the kiln charging work is carried out in order from the young number in each street.
Assume that the unloading pitch of each kiln (the time from the completion time of the unloading work of one kiln to the completion time of the unloading work of the next kiln) is 8 minutes. The estimated time from the start of unloading work in the first kiln to the completion of unloading work in the last kiln in one street of A furnace is 8 minutes pitch × (13 kilns - 1) = 96 minutes. be. Therefore, at the end timing of unloading the kiln on the same street as furnace A, (start time of unloading work in the first kiln + scheduled time (96 minutes)) and (actual end time of unloading work in the last kiln) ), it can be determined that a delay in taking out the kiln or taking it back has occurred on that street, and the time difference becomes the time for taking out the kiln or taking it back. In the example of FIG. 17, normal operation is in operation in 4 [4] of furnace A, but there is a 30-minute delay in unloading in 1 [1] of furnace A.

In this case, the kiln-out delay time generated in the A furnace is corrected by adding it to the subsequent four expected dry distillation times in the A furnace. In the case of this example, a 30-minute delay in unloading from the kiln is detected at the end of the unloading work of 1 [1], as described in 3 below [3], as described in 5 [5] below, and 2 street [2] and the next 4 street [4] are in a charged state. Therefore, add 30 minutes to the expected dry distillation time for each of the following 3 [3], the following 5 [5], the following 2 [2], and the following 4 [4]. . Also, for example, if a delay of 30 minutes is detected at the end of the kiln removal work in 5 [5], the following 2 [2], the following 4 [4], and the following 1 Add 30 minutes to the expected dry distillation time for each step [1] and the next 3 steps [3].
In this way, it is possible to determine the occurrence of a delay in unloading or a withdrawal in each line of A furnace at the end timing of unloading of each line of A furnace.

FIG. 18 shows a case where a delay in taking out the kiln occurs in the B furnace.
The scheduled time required from the start of unloading work in the first kiln of A furnace to the completion of unloading work in the last kiln of B furnace in one street in a set of A furnace and B furnace is the operating rate ( kiln release schedule), for example, 4 hours. Therefore, at the start timing of the unloading work on the street where furnace A is located, When there is a time difference from the start time of unloading of the first kiln on the relevant street of the furnace), the delay of unloading of furnace A or B before the start time of unloading work of the first kiln of A furnace this time. Alternatively, it can be determined that take-back has occurred, and the time difference between them becomes the take-out delay time or take-out time. In the example of FIG. 18, normal operation is performed in 4 [4] in the set of A and B furnaces, but the following 1 [1] requires 0.5 hours (30 minutes) to be removed from the kiln. A delay is occurring.

18, it is not possible to detect whether the delay or withdrawal of the kiln has occurred in the A furnace or in the B furnace in the determination of the various start timings of unloading from the kiln. Therefore, the determination at the completion timing of taking out the kiln as described in FIG. 17 is also used. In the determination at the end timing of the unloading work in each way described with reference to FIG. Therefore, from the results detected by the determination at the start timing of the work to be taken out from the kiln in each way in FIG. By subtracting the detected result (delay in discharge from the kiln or recovery time from the kiln), it is possible to detect the delay in release from the kiln or the recovery from the kiln in the B furnace. In addition, by also determining whether or not there is a delay in taking out the kiln from the kiln at the start and end of the kiln taking-out work and at the timing of the end of each kiln-out work, it is possible to detect the delay in taking out from the kiln or the take-back at an early stage.

In this case, the kiln-out delay time that occurred in the above-mentioned furnace B is as it is now (as the delay in kiln-out is detected at the start timing of the kiln-out work) and the subsequent four different dry distillation in furnace A. The scheduled dry distillation time is corrected by adding to the scheduled time. In the case of the example shown in FIG. 18, a delay of 30 minutes is detected at the start timing of the work to be taken out from the kiln in 3 [3]. ], the next 2 street [2], the next 4 street [4], and the next 1 street [1] are in a charged state. Therefore, according to the following 3 [3], following 5 [5], following 2 [2], following 4 [4], and following 1 [1], dry distillation Add 30 minutes to your scheduled time.
In this way, it is possible to determine the occurrence of a delay or reversal in the discharge from the B furnace at the start timing of each of the discharge operations from the A furnace.

FIG. 15 shows the functional configuration of the control device 100 according to the fourth embodiment. The functional configuration shown in FIG. 15 is different from the functional configuration shown in FIG. The line is changed to a double-headed line connecting the target furnace temperature calculation unit 102 and the input heat amount calculation unit 103 (see the description of the changes to the functional configuration shown in FIG. 1 of the third embodiment). Therefore, detailed descriptions of the input unit 101, the target furnace temperature calculation unit 102, the input heat amount calculation unit 103, and the input heat amount setting unit 104 that are the same as those of the control device 100 according to the first embodiment will be omitted. FIG. 16 is a flow chart showing the processing of the target furnace temperature calculator 102, input heat quantity calculator 103, and dry distillation time corrector 105. As shown in FIG. The present embodiment will be described based on the third embodiment, and processing common to the flowchart of FIG. 11 will be given the same reference numerals, and description thereof will be omitted.
In this embodiment, the optimum furnace temperature pattern is determined for furnace A, and the flowchart in FIG. ), and at the start of the A furnace (for example, the start timing of each of the kiln removal operations of the A furnace). The calculation flow does not change between "at the end of the run" and "at the start of the run", and calculations are performed by adding corrections for delays in kiln release or recovery only for the scheduled time for dry distillation to the obtained operation performance data and schedule data. do. For example, as shown in FIG. 17, in the calculation at the start, since the performance data up to 4th [4] are obtained, the calculation is performed using the performance data before 4th [4] and the future schedule. Future schedules will reflect the time of unloading or take-back detected at the beginning of one way [1].

After the target furnace temperature calculation unit 102 takes in the carbonization information data in steps S1101 and S1102, and the input heat amount calculation unit 103 takes in the combustion information data, in step S1601, the dry distillation time correction unit 105 determines whether the kiln is delayed or returned to the kiln. Detects the presence or absence of occurrence. As described with reference to FIG. 17, in the case of detecting the presence or absence of a delay in the removal from the kiln or the occurrence of a take-back at the end timing of the removal from the kiln in a certain manner in furnace A, the dry distillation time correction unit 105 corrects ( When there is a time difference between the start time of the first kiln unloading work + scheduled time) and the (actual completion time of the kiln unloading work of the last kiln), it is assumed that the kiln unloading was delayed or returned on the relevant street. can judge. If (start time of unloading work in the first kiln + scheduled time) is earlier than (actual end time of unloading work in the last kiln), there is a delay in unloading in furnace A. If it is late, it means that the A furnace is being recovered from the kiln.

In addition, as described in FIG. 18, in the case of detecting whether or not there is a delay in the removal from the kiln or the occurrence of a take-back at the start timing of the removal from the kiln as it is in the furnace A, the dry distillation time correction unit 105 (A There is a time difference between the start time of unloading work in the first kiln on the street before the specified kiln in the furnace + scheduled time) and the (start time of unloading work in the first kiln on the corresponding street in furnace A). Occasionally, it can be determined that a delay or take-back of unloading occurred in A or B furnace prior to the start time of the first unloading operation in question in A furnace. When (starting time of unloading work in the first kiln on the street before the street in A furnace + scheduled time) is earlier than (starting time of unloading work in the first kiln on the street in question in furnace A) , there is a delay in the release from the kiln in the A furnace or the B furnace, and if it is delayed, it means that the recovery from the kiln has occurred in the A furnace or the B furnace.
In either case, a threshold may be set for the time difference, and it may be detected that the kiln is delayed or returned to the kiln only when the time difference exceeds the threshold.

In step S1602, the dry distillation time correction unit 105, when detecting the occurrence of a delay or return in step S1601, sets the time difference as the delay time or return time for removal from the kiln and corrects the expected dry distillation time. Proceed to S1103. When the target furnace temperature calculation unit 102 calculates the target furnace temperature pattern in this manner, the scheduled carbonization time is corrected. As described with reference to FIGS. 17 and 18, when there is a delay in taking out the kiln, the dry distillation time correcting unit 105 adds the time required for taking out the kiln to the expected dry distillation time. Further, when recovery from kiln removal has occurred, the dry distillation time correction unit 105 subtracts the recovery time from the kiln from the expected dry distillation time. The scheduled carbonization time is an explanatory variable of the coke temperature prediction model, and by correcting the scheduled carbonization time, the influence of the delay or return of the coke temperature in step S1107 can be reflected.

Subsequent steps S1103 to S1112 are the same as in FIG.
In the examples shown in FIGS. 17 and 18, in order to detect the delay in the discharge from the kiln or the withdrawal at an early stage, the presence or absence of the delay in the discharge from the kiln or the withdrawal from the kiln is determined at the end of the furnace A and at the time of the furnace A. It was supposed to be done at the start of the street. However, the determination of the presence or absence of the kiln release delay or take-back is not limited to such timing. For example, the presence or absence of delay in kiln release or take-back may be determined at only one of the time when the running of the A furnace is finished and the time when the running of the A furnace is started. Further, for example, if furnace B is also subject to the control of the amount of heat input by the control device 100, in addition to or instead of at least one of the end of heating of furnace A and the start of heating of furnace A, the heating of furnace B is performed. At least at the end and at the start of the B furnace, the presence or absence of delay in kiln release or withdrawal may be determined. Moreover, although the present embodiment has been described based on the third embodiment, the correction function described in the present embodiment may be applied to the first and second embodiments.

As described above, when predicting the coke temperature in the target furnace temperature calculator 102, it is possible to take into account the influence of the delay or return of the kiln discharge, so that the coke temperature can be predicted more accurately. By more accurately predicting the coke temperature, it is possible to set the coke temperature to a value corresponding to the target temperature, and to reduce the production cost (reduction of the excessive carbonization heat amount due to excessive carbonization).

[Example 3]
Off-line control method (hereinafter referred to as this control method in the description of Example 3) that takes into account the delay in kiln removal as in the fourth embodiment (with correction of the expected dry distillation time) using past performance data A simulation was performed. In addition, as a comparison method, an off-line simulation of a control method that does not consider the delay in kiln discharge (no correction of the scheduled carbonization time) was performed.
FIG. 19 is a characteristic diagram showing the results of Example 3, in which (a) shows time-series changes in coke temperature, and (b) shows time-series changes in furnace bundle temperature. The solid line (∘) in the figure is the result of this control method, and the dotted line (Δ) is the result of the comparative example.
As shown in FIG. 19( a ), this control method makes it possible to adjust the coke temperature to a value corresponding to the target temperature with high accuracy. In addition, as shown in FIG. 19(b), in this control method, as a result of taking into account the effect of the longer carbonization time due to the delay in getting out of the kiln, an action is taken to reduce the temperature of the furnace cascade, thereby reducing the production cost ( It is possible to reduce the amount of excess dry distillation heat by over dry distillation.

[Fifth embodiment]
The fifth embodiment is an example in which the operator considers shortening or extending the time of the coal charging work out of the kiln with respect to the standard work time at his/her own judgment when the kiln out is delayed or returned to the kiln. . The fifth embodiment will be described below. Descriptions of contents common to the first to fourth embodiments will be omitted, and differences from the first to fourth embodiments will be mainly described.

FIG. 20 is a diagram for explaining the effect of operation based on the operator's judgment on the coke temperature when a delay in kiln output occurs. Shows time series changes. In FIG. 20(a), the line parallel to the time axis with target on the side indicates the target temperature of the coke temperature, and in FIG. 20(b), the line parallel to the axis with standard on the side A straight line indicates a standard carbonization time. Further, as explained in FIGS. 14(a) and 14(b), the range affected by the delay in taking out the kiln shows how the coal is charged when the kiln taking out is delayed due to an equipment trouble or the like. On streets within the range affected by delays in kiln release, the amount of delay in kiln release will result in additional carbonization. 20(a) and (b), similarly to FIGS. 14(a) and (b), [1], . 1, 6, 11, 16...),..., 5 (carbonization chamber No. 5, 10, 15, 20...). Also in FIG. 20(b), as in FIG. 14(b), for example, it is assumed that a delay in taking out the kiln occurs due to an equipment trouble or the like before the start of step 1 [1]. In this case, as shown in the range affected by the delay in putting out the kiln in the figure, the first street [1] and the following four streets [3], [5], [2], and [4] are charged. , and the coke temperature increases because the carbonization time increases as much as the delay in taking out the kiln (see the upward white arrow lines in FIGS. 20(a) and (b)). In such a case, in order to avoid a production reduction in the coke oven 1, the operator tends to operate (take back) on his/her own judgment such that the coal charging work out of the kiln is shorter than the standard work time. This is because the coke temperature was higher than the target temperature, and it was judged that appropriate carbonization of the coke could be achieved even if the carbonization time was shortened. For example, the inventors have found that operators tend to make take-backs in the later streets of the late-launch impact zone so as not to affect the start of the next carbonization cycle. In the drawing, the period during which the recovery is performed is indicated as the recovery period. In addition, the first [1] after the range affected by the delay in putting out the kiln is not affected by the delay in putting out the kiln, so the dry distillation time does not become longer, and the delay in putting out the kiln does not affect the dry distillation time. . According to the control described above, the input heat amount (furnace temperature) is lowered on the assumption that the coke will additionally stay in the coke chamber 2 for the delay time of taking out the coke from the kiln. For this reason, if recovery is performed based on the operator's own judgment, the coke temperature may drop more than intended because the carbonization time will be shortened due to the recovery even though the amount of heat input has been reduced. (see downward white arrow lines in FIGS. 20(a) and (b)). Note that the withdrawal period is not limited to the period shown in FIG. 20(a), and is determined according to the operator's judgment.

FIG. 21 is a diagram for explaining a method of correcting the expected dry distillation time, in which (a) is an example of time-series data of the expected dry distillation time after correction in the method of the fourth embodiment, and (b) is It is an example of time-series data of the expected dry distillation time after correction in the method of the present embodiment. In the fourth embodiment, as shown in FIG. 21( a ), when a delay in taking out the kiln occurs, the same time as the kiln taking out delay time is uniformly applied to the scheduled carbonization time for each street within the kiln taking out delay influence range. to add. Therefore, the calculation of the target furnace bundle temperature pattern and the calculation of the input heat amount are performed in a state in which the expected dry distillation time is lengthened. In such a case, if the operator decides to take back the coke, the coke temperature drops as described above, so that the coke is in a so-called undercooked state, and there is a risk that the quality of the coke will deteriorate. Here, we explained the case where there was a delay in taking out the kiln. ), which can lead to excessively high coke temperatures.

Therefore, in the fifth embodiment, as shown in FIG. 21(b), in anticipation that the operation will be performed based on the operator's judgment when the kiln is delayed or returned to the kiln, the kiln after correction in at least one way The expected dry distillation time is made different from the corrected expected dry distillation time in at least one other way. When considering the tendency due to the operator's judgment as described above, the absolute value of the correction amount for the pre-correction scheduled dry distillation time in the later time is the correction for the pre-corrected dry distillation scheduled time in the previous time. without being greater than the absolute value of the amount, the absolute value of the correction amount for the scheduled carbonization time before correction in the temporally later period for at least two streets Make it smaller than the absolute value of the correction amount with respect to time. In other words, the absolute value of the correction amount for the uncorrected expected dry distillation time in the later time period and the absolute value of the corrected amount for the uncorrected expected dry distillation time in the earlier time period, the former and the latter are different. They may be the same, but the former may not be smaller than the latter, and the latter may be smaller than the former.

The functional configuration of the coke manufacturing process control device according to the fifth embodiment is the same as the functional configuration shown in FIG. Therefore, the input unit 101, the target furnace temperature calculation unit 102, the input heat amount calculation unit 103, the input heat amount setting unit 104, and the carbonization time correction unit 105 are the same as those of the control device 100 according to the first to fourth embodiments. Detailed description is omitted. Further, in the flowchart showing the processing of the target furnace temperature calculation unit 102 and the input heat amount calculation unit 103 according to the fifth embodiment, step S1602 in FIG. 16 is changed as follows. The unit 105 performs the following processing in step S1602 of FIG.

That is, in step S1602, when the dry distillation time correction unit 105 detects the occurrence of delay or withdrawal in step S1601, the time difference is calculated as the delay time in removing the kiln or the withdrawal time in the kiln. As described in the fourth embodiment, when there is a delay in taking out the kiln, the dry distillation time correcting unit 105 adds the time required for taking out the kiln to the expected dry distillation time. Further, when recovery from kiln removal has occurred, the dry distillation time correction unit 105 subtracts the recovery time from the kiln from the expected dry distillation time.

When it is detected in step S1601 that the delay in taking out the kiln or the occurrence of the taking-out of the kiln is detected, the target furnace temperature calculating unit 102 determines that the coal is charged at the timing of the delay or the taking-out of the kiln (see FIG. 20). In order to calculate the coke temperature T _co (n) within the delay influence range) by Equation (2), in step S1602, the carbonization time correction unit 105 determines the coke temperature T co Correct the scheduled carbonization time for streets within the delay influence range (this is the same for the fourth embodiment).
As described above, in the fourth embodiment, the dry distillation time correction unit 105 adds the kiln-out delay time or subtracts the kiln-out recovery time from each scheduled dry distillation time, so the scheduled dry distillation time for each street is the same. changed in time.

On the other hand, as described above, in the present embodiment, the dry distillation time correction unit 105 corrects at least one of the streets within the kiln-out delay influence range after the timing when the kiln-out delay or recovery occurred. The scheduled carbonization time is set to a time different from at least one of the others. More specifically, the dry distillation time correction unit 105 converts the absolute value of the correction amount for the uncorrected scheduled dry distillation time in the later time period into the correction amount for the uncorrected scheduled dry distillation time in the earlier time period. The absolute value of the correction amount for the expected dry distillation time before correction in the later street for at least two streets, without being larger than the absolute value of be smaller than the absolute value of the correction amount for

For example, a positive coefficient to be multiplied by the kiln-out delay time and the kiln-out recovery time is set in advance for each street in order to correct the estimated dry distillation time in each street after the timing of the delay or withdrawal of the kiln. . In doing so, for at least two streets, the value of the coefficient for the temporally later street is not greater than the coefficient for the temporally later street than the temporally earlier street. essentially less than the value of the coefficient in the previous street. In the present embodiment as well, the kiln charging is carried out at intervals of 5 kilns, so it is sufficient to preset the coefficients up to 5 patterns ahead. For example, set the coefficients for the last two streets within the influence range of the kiln-out at the timing when the kiln-out delay or return occurs to "0.5", and set the coefficients for the remaining streets to "1.0". do.
Then, the dry distillation time correction unit 105 calculates a value obtained by multiplying the kiln-out delay time and the kiln-out recovery time by a coefficient set for each type as a correction time, and calculates the value as a correction time. By adding or subtracting the correction time, the expected dry distillation time after correction is calculated.

Here, each route in which the coal charging work is carried out next to the timing at which the kiln-out is delayed or returned to the kiln is called as follows.
In the example shown in FIG. 17 described in the fourth embodiment, there is a delay in taking out the kiln in one of the A furnaces. Therefore, the streets within the range affected by the delay in starting out in furnace A after the timing of the delay in starting out are as follows: 3, 5, 2, and 4. , the execution order of the kiln charging work is this order. Therefore, in the above-described example, the last two streets within the influence range of firing out at the timing of the delay or return of firing out are the following 2 streets and the following 4 streets, and the remaining streets are: 3 and 5 below. From the above, the coefficients for the following 3, 5, 2, and 4 are 1.0, 1.0, and 0.5, respectively. , and becomes 0.5. Therefore, assuming that the kiln-out delay time is 30 minutes, the dry distillation time correction unit 105 performs the following 3, 5, 2, and 4 as follows. The correction time is calculated, and the time obtained by adding the correction time to the expected dry distillation time (before correction) is set as the expected dry distillation time after correction.
Correction time in the following three ways: 30 minutes (= 30 minutes x 1.0)
Correction time in the following 5 ways: 30 minutes (= 30 minutes x 1.0)
Correction time in the following two ways: 15 minutes (= 30 minutes x 0.5)
Correction time in the following 4 ways: 15 minutes (= 30 minutes x 0.5)

In addition, in the example shown in FIG. 18 described in the fourth embodiment, there is a delay in taking out the kiln in one of the B furnaces. Therefore, the streets within the range affected by the delay in removing the furnace from the timing when the delay in removing the furnace occurred are as follows: 3, 5, 2, 4, and 4. It is as follows 1, and the execution order of the kiln charging work is this order. Therefore, in the example described above, the last two streets within the influence range of the firing at the timing when the delay or return of the firing occurred are the following 4 and the following 1, and the remaining streets are: It is as follows 3, as follows 5, and as follows 2. From the above, the coefficients for the next 3, next 5, next 2, next 4, and next 1 are "1.0" and "1.0", respectively. , 1.0, 0.5, 0.5. Therefore, assuming that the kiln-out delay time is 30 minutes, the dry distillation time correction unit 105 performs the following 3, 5, 2, 4, The correction time is calculated in one of the following ways, and the time obtained by adding the correction time to the expected dry distillation time (before correction) is set as the expected dry distillation time after correction.
Correction time in the following three ways: 30 minutes (= 30 minutes x 1.0)
Correction time in the following 5 ways: 30 minutes (= 30 minutes x 1.0)
Correction time in the following two ways: 30 minutes (= 30 minutes x 1.0)
Correction time in the following 4 ways: 15 minutes (= 30 minutes x 0.5)
Correction time for one of the following: 15 minutes (= 30 minutes x 0.5)

The method of correcting the estimated dry distillation time is not limited to the above calculation. For example, the coefficients for all streets may be different values. Further, depending on the characteristics of the coke oven 1, the operating conditions, the tendency of operation based on the judgment of the operator, etc., the coefficient for the subsequent operation in terms of time may be set to a larger value than the coefficient for the previous operation in terms of time. good. Also, the calculation formula is not limited to the one described above. For example, the correction time may be calculated so as to exponentially shorten the correction time for the later passage.
Also, like the fourth embodiment, the present embodiment has been described based on the third embodiment, but the correction function described in the present embodiment may be applied to the first and second embodiments.

As described above, when predicting the coke temperature in the target furnace temperature calculation unit 102, it is possible to consider the influence of the operation due to the operator's judgment when the kiln is delayed or taken back, and the coke temperature can be further reduced. can be predicted accurately. By more accurately predicting the coke temperature, it is possible to accurately set the coke temperature to a value corresponding to the target temperature, and to further reduce production costs reduction) can be achieved.

[Sixth Embodiment]
The sixth embodiment is an example of correcting the target furnace temperature (target temperature pattern) for a plurality of future times calculated by the target furnace temperature calculator 102 .
If the coke temperature prediction model described in the second embodiment is used, the coke temperature prediction accuracy is improved, but the prediction error remains. Moreover, using the big data model 902 increases the computational load and requires a huge amount of learning data for machine learning. Therefore, for example, it is not preferable to use the big data model 902 when the purpose is to reduce the calculation load or when learning data cannot be prepared. In such a case, using the regression model 901 (the coke temperature prediction model described in the first embodiment) results in a larger prediction error than using the big data model 902 . Further, in the third embodiment, since the furnace bundle temperature prediction model is used when calculating the target furnace bundle temperature pattern, the prediction error of the furnace bundle temperature prediction model also affects the target furnace bundle temperature pattern. Also, it is conceivable that the schedule values used in the prediction models including the coke temperature prediction model deviate greatly from the actual values. The prediction errors of the prediction models including these coke temperature prediction models and the deviation of the schedule values from the actual values are factors that reduce the accuracy of the target furnace bundle temperature pattern.
Therefore, in the sixth embodiment, the target furnace temperature pattern calculated by the target furnace temperature calculation unit 102 is corrected according to the difference between the target value and the actual value of the physical quantity representing the carbonization state of coke. Better feasibility of cluster temperature patterns.

The sixth embodiment will be described below. Descriptions of contents common to the first to fifth embodiments will be omitted, and differences from the first to fifth embodiments will be mainly described based on the third embodiment. As described in the first embodiment, the physical quantity representing the carbonization state of coke may be, for example, the temperature of the furnace wall 4 in addition to the coke temperature. Give an explanation.

FIG. 22 shows the functional configuration of the control device 100 according to the sixth embodiment. The functional configuration shown in FIG. 22 is different from the functional configuration shown in FIG. The single-headed arrow line is changed to a double-headed line connecting the target furnace temperature calculation unit 102 and the input heat amount calculation unit 103 (see description of changes to the functional configuration shown in FIG. 1 of the third embodiment). . Therefore, detailed descriptions of the input unit 101, the target furnace temperature calculation unit 102, the input heat amount calculation unit 103, and the input heat amount setting unit 104 that are the same as those of the control device 100 according to the first to fifth embodiments will be omitted. . FIG. 23 is a flow chart showing the processing of the target furnace temperature calculator 102, the input heat quantity calculator 103, and the target furnace temperature corrector 106. As shown in FIG. As described above, this embodiment will be described based on the third embodiment. Therefore, in the flowchart of FIG. 23, the same reference numerals are given to the processing that is common to the flowchart of FIG. 11, and the description thereof will be omitted. FIG. 24 is a diagram for explaining the outline of the process of correcting the target furnace group temperature pattern.

Processing for correcting the target furnace group temperature pattern in the sixth embodiment will be described with reference to FIGS. 22 to 24. FIG.
In FIG. 24, control start indicates the timing when the control by the control device 100 is started. The current time indicates the timing when the flowchart of FIG. 23 is started. Similar to the flowchart of FIG. 11 described in the third embodiment, the timing of starting the flowchart of FIG. , the timing of starting the charging operation on each street. Also, the flowchart of FIG. 23 is executed in regular time cycles.

The upper part of FIG. 24 shows time series changes in coke temperature. Here, the actual values of the coke temperature 2411 are indicated by black circles, and the predicted values are indicated by white circles. In addition, a solid line connects the actual values of coke temperature and a dashed-dotted line connects predicted values to show changes in coke temperature over time. The upper part of FIG. 24 shows the coke temperature 2411 as well as the coke temperature target temperature 2412 .
The middle part of FIG. 24 shows time-series changes in furnace bundle temperature. The temperature of the furnace bed 2421 is obtained every hour. Here, the actual value of the furnace bed temperature 2421 is indicated by a solid line, and the predicted value is indicated by a one-dot chain line. 24 shows a target furnace temperature pattern 2422 calculated by the target furnace temperature calculator 102 and a corrected target furnace temperature pattern 2423 corrected by the target furnace temperature corrector 106. FIG.
The lower part of FIG. 24 shows time-series changes in the amount of input heat. Here, the actual value of the input heat quantity 2431 is indicated by a solid line, and the future value is indicated by a one-dot chain line.

As described in the third embodiment, by performing the processing of steps S1101 to S1111, the target furnace group temperature pattern 2422 that minimizes the evaluation function J ₁ ' among the target furnace group temperature patterns converted in step S1106 is obtained. can get. In the middle part of FIG. 24, a target furnace bundle temperature pattern 2422 ahead of the current time is calculated at the current time.

After step S1110, step S2301 is performed. In step S2301, the target furnace temperature correction unit 106 corrects the target furnace bundle temperature pattern 2422 according to the difference between the target coke temperature value (target temperature 2412) and the actual value. For example, the target furnace temperature correction unit 106 subtracts the actual values of the coke temperature in each of the past one cycle in the carbonization cycle starting from the current time from the target value of the coke temperature (target temperature 2412). Calculate the average value with For example, the kiln loading order is 1st, 3rd, 5th, 2nd, and 4th, and the timing of starting the flow chart of FIG. It is the end timing, and the current time is the timing when the kiln removal work in 1 way is completed ([1] shown below the current time in the upper part of FIG. 24 indicates this). In this case, the actual values of the coke temperature in 1st, 4th, 2nd, 5th, and 3rd immediately before are used (shown below the current time in the upper part of FIG. 24 [1 ], [4], [2], [5], and [3] shown in order towards the left represent these things).

Assuming that k is a variable that identifies streets and that the number of streets for which the weighted average value is calculated is K (K=5 in this embodiment), the correction amount FB to be added to the target furnace bundle temperature pattern 2422 is , for example, represented by the following equations (6a) and (6b).

Here, G ₁ and G ₂ are preset positive gains. w _k is a preset positive weighting factor for street k. T _{co_p_k} is the desired coke temperature at street k. T _{co_m_k} is the actual coke temperature at street k. The fact that the correction amount FB is positive corresponds to the fact that the coke temperature is low because the target value of the coke temperature is higher than the actual value. In this case, in order to make the coke temperature reach the target temperature 2412 as soon as possible, it is conceivable to greatly increase the amount of input heat. On the other hand, the fact that the correction amount FB is negative corresponds to the coke temperature being high because the target value of the coke temperature is lower than the actual value. In this case, if the coke temperature is abruptly lowered, the coke temperature may become too low. In such a case, it is preferable to make the gain G ₁ larger than the gain G ₂ . However, the values of the gains G ₁ and G ₂ may be appropriately determined according to the characteristics of the coke oven 1, operating conditions, etc., including the magnitude relationship of the values.

The weighting factor w _k may be appropriately determined according to the characteristics of the coke oven 1, operating conditions, and the like. For example, when emphasizing the difference between the target coke temperature value and the actual coke temperature at a time close to the current time, the weighting factor w _k for the subsequent street in time is lower than the weighting factor w _k for the previous street in time. , and for at least two streets, the weighting factor w _k for the later street in time is larger than the weighting factor w _k for the earlier street in time.

In the example shown in the upper part of FIG. 24, the actual value of the coke temperature 2411 is higher than the target temperature 2412, so the correction amount FB is calculated by equation (6b). In this case, the correction amount FB has a negative value.
From the viewpoint of stable operation, etc., it is not preferable to change the target furnace bed temperature pattern too abruptly. Therefore, when the correction amount FB is not within the range of the upper and lower limits set in advance (lower limit ≤ FB ≤ upper limit), the target furnace temperature correction unit 106 sets the correction amount FB to Change to a value closer to When the correction amount FB is within the preset upper and lower limits (lower limit ≤ FB ≤ upper limit), the target furnace temperature correction unit 106 does not change the correction amount FB.

The target furnace temperature correction unit 106 adds the correction amount FB obtained as described above to the target furnace bundle temperature pattern 2422 calculated by performing the processing of steps S1101 to S1111, thereby obtaining the target after correction. A furnace cluster temperature pattern 2423 is calculated. In the example shown in the upper part of FIG. 24, since the correction amount FB is calculated as a negative value, the target furnace group temperature pattern 2423 after correction indicates a lower temperature than the target furnace group temperature pattern 2422 (see FIG. 24). 24, see the middle downward white arrow line).

As described above, the corrected target furnace group temperature pattern is calculated in step S2301. Then, in step S<b>2302 , the target furnace temperature correction unit 106 outputs the target furnace temperature pattern after correction to the input heat quantity calculation unit 103 . In this embodiment, the input heat calculation unit 103 calculates the target furnace temperature pattern calculated by the target furnace temperature correction unit 106 instead of the target furnace temperature pattern calculated by the target furnace temperature calculation unit 102 in the flowchart of FIG. Using the pattern, calculate the heat input. In step S1104, the input heat calculation unit 103 selects a target furnace temperature pattern candidate (target furnace temperature pattern generated in step S1103) instead of the target furnace temperature pattern calculated by the target furnace temperature correction unit 106. or the target furnace temperature pattern generated in step S1111) is used to calculate the amount of input heat.

As described above, by feeding back the actual value of the coke temperature, the target furnace bundle temperature pattern calculated by the target furnace temperature calculation unit 102 is adjusted so that the difference between the actual value and the target value of the coke temperature becomes small. can be corrected, and control of the input heat quantity can be executed with even higher accuracy. As a result, it is possible to more accurately set the coke temperature to a value corresponding to the target temperature, and to further reduce the production cost (reduction of the excessive carbonization heat amount due to excessive carbonization). can.
Although this embodiment has been described based on the third embodiment, the correction function described in this embodiment may be applied to the first, second, fourth, and fifth embodiments.

As described above, the present invention has been described together with the embodiments, but the above-described embodiments merely show specific examples for carrying out the present invention, and the technical scope of the present invention is not construed in a limited manner. It should not be. That is, the present invention can be embodied in various forms without departing from its technical concept or main features.
A control device for a coke manufacturing process to which the present invention is applied can be configured by a computer device having, for example, a CPU, a ROM, a RAM, etc., and its functions are realized by the CPU executing a predetermined program.
Further, the present invention provides software (program) for realizing control of the coke production process of the present invention to a system or apparatus via a network or various storage media, and the computer of the system or apparatus reads and executes the program. It can also be realized by

REFERENCE SIGNS LIST 1 coke oven 2 coking chamber 3 combustion chamber 4 furnace wall 5 regulating valve 6 thermometer for measuring furnace temperature 7 extrusion ram 8 thermometer for measuring coke temperature 9 guide wheel 10 coke 100 control device for coke production process 101 input unit 102 target oven temperature calculator 103 input heat amount calculator 104 input heat amount setting unit 105 carbonization time correction unit 106 target oven temperature correction unit 200 storage unit 601 target oven bundle temperature pattern 602 predicted value of coke temperature 603 target temperature of coke temperature 801 oven bundle Predicted temperature value 802 Input heat amount 901 Regression model 902 Big data model 1201 Input heat amount that minimizes the evaluation function 1202 Control waveform of oven bundle temperature 2411 Coke temperature (actual value, predicted value)
2412 Target temperature of coke temperature 2421 Furnace temperature (actual value, predicted value)
2422 Oven temperature pattern before correction 2423 Oven temperature pattern after correction 2421 Input heat amount S Coal charge W Moisture content of coal T _co- coke temperature T _ro Oven temperature t _k Distillation time t Time according to _t

Claims (22)

A control device for a coke production process that controls the amount of heat input to the combustion chamber in a coke oven having a plurality of carbonization chambers and a plurality of combustion chambers so that the physical quantity representing the carbonization state of coke becomes a value corresponding to a target value There is
Using a physical quantity prediction model that predicts the physical quantity based on the first influencing factor including the furnace temperature, which is the temperature of the combustion chamber, and at least one of the actual value and the scheduled value of the first influencing factor, predicting the physical quantity; calculating a value of a first evaluation function including a term representing a difference between the predicted physical quantity and a target value of the physical quantity; a target furnace temperature calculation unit that calculates the temperature;
and an input heat amount calculation unit that calculates an input heat amount corresponding to the target furnace temperature calculated by the target furnace temperature calculation unit. The input heat amount calculation unit uses a furnace temperature prediction model that predicts a furnace temperature based on a second influencing factor including the input heat amount, and at least one of the actual value and the schedule value of the second influencing factor, 2. A control device for a coke production process according to claim 1, which calculates an input heat amount according to a target furnace temperature. The input heat amount calculation unit calculates the target furnace temperature calculated by the target furnace temperature calculation unit based on a second evaluation function including a term representing the difference between the furnace temperature predicted by the furnace temperature prediction model and the target furnace temperature. 3. The coke production process control device according to claim 2, wherein the furnace temperature is used as the target furnace temperature included in the second evaluation function to calculate the input heat quantity. 4. The coke production process control device according to claim 3, wherein said second evaluation function further includes a term representing the amount of change in input heat quantity. The target furnace temperature calculation unit generates a plurality of candidates for a future target furnace temperature, uses each of the plurality of candidates as the furnace temperature included in the first influence factor in the physical quantity prediction model, and calculates the predicting a physical quantity and calculating the value of the first evaluation function, respectively, and obtaining the target furnace temperature from among the plurality of candidates based on the executed value of the first evaluation function; A coke production process control device according to any one of claims 1 to 4. The target furnace temperature calculating unit calculates the target furnace temperature using a furnace temperature that satisfies a predetermined evaluation of the amount of change in input heat quantity as the first influencing factor in the physical quantity prediction model. 6. The coke production process control device according to any one of 5. The input heat amount calculation unit calculates an input heat amount that satisfies a predetermined evaluation of a change amount of the input heat amount as an input heat amount corresponding to a furnace temperature that has not been determined as the target furnace temperature,
The target furnace temperature calculation unit calculates a predicted value of the furnace temperature when controlling the coke production process with an input heat amount that satisfies a predetermined evaluation of the amount of change in the input heat amount, and calculates a predetermined evaluation of the amount of change in the input heat amount. 7. The control device for a coke production process according to claim 6, wherein said target furnace temperature is calculated as a furnace temperature when satisfying The target furnace temperature calculation unit generates a plurality of candidates for future target furnace temperatures,
The input heat amount calculation unit uses each of the candidates for the target furnace temperature generated by the target furnace temperature calculation unit as the target furnace temperature included in the second evaluation function based on the second evaluation function. to calculate the amount of heat input,
5. The control device for a coke production process according to claim 3, wherein said target furnace temperature calculating section obtains said target furnace temperature based on a furnace temperature corresponding to the input heat amount calculated by said input heat amount calculating section. The coke production process according to any one of claims 1 to 8, wherein the cycle for calculating the input heat amount by the input heat amount calculation unit is shorter than the cycle for calculating the target furnace temperature by the target furnace temperature calculation unit. Control device. The coke production process is a coke production process in which a plurality of coking chambers are divided into a plurality of streets, and coal charging is carried out for each street,
10. The coke production process control device according to any one of claims 1 to 9, wherein the target furnace temperature is a transition of the target furnace temperature at a plurality of times in the future. 11. The control device for a coke production process according to claim 10, wherein said target furnace temperature calculation unit executes a process of calculating said target furnace temperature at regular time intervals. The physical quantity prediction model includes a regression model that predicts the physical quantity based on the first influencing factor including the furnace temperature, and an estimation model based on machine learning for estimating the prediction error of the regression model and correcting it. A control device for a coke making process according to any one of claims 1 to 11, comprising: 13. The coke production process control device according to any one of claims 1 to 12, wherein said physical quantity is coke temperature or furnace wall temperature. The coke production process is a coke production process that collectively adjusts the amount of heat input to the plurality of combustion chambers,
14. The input heat amount calculation unit calculates the input heat amount to be adjusted collectively for the plurality of combustion chambers as the input heat amount according to the target furnace temperature calculated by the target furnace temperature calculation unit. A control device for a coke production process according to any one of Claims 1 to 3. 15. The coke production process control device according to any one of claims 1 to 14, wherein the furnace temperature is a furnace bundle temperature that is a representative value of the temperatures of the plurality of combustion chambers. The physical quantity prediction model includes the expected dry distillation time as the first influencing factor,
16. The coke production process according to any one of claims 1 to 15, comprising a dry distillation time correction unit that corrects the expected dry distillation time using the delay or recovery time when the kiln is delayed or recovered. Control device. The target furnace temperature calculating unit performs a process of calculating the target furnace temperature at least either at the end of running or at the start of running,
17. The control device for a coke manufacturing process according to claim 16, wherein said carbonization time correction section executes said correction when said target furnace temperature calculation section calculates said target furnace temperature. 18. The coke manufacturing process according to claim 16 or 17, wherein the carbonization time corrector sets the corrected scheduled carbonization time in at least one street to a time different from the corrected scheduled carbonization time in at least one other street. controller. The dry distillation time correction unit makes the absolute value of the correction amount for the pre-correction scheduled dry distillation time in the later time larger than the absolute value of the correction amount for the pre-corrected dry distillation scheduled time in the previous time. For at least two different ways, the absolute value of the correction amount for the expected dry distillation time before correction in the later time is the absolute value of the correction amount for the expected dry distillation time before correction in the earlier time 19. The coke making process controller of claim 18, wherein the coke making process controller is less than . a target furnace temperature correction unit that corrects the target furnace temperature calculated by the target furnace temperature calculation unit according to a difference between a target value and an actual value of the physical quantity;
20. The control device for a coke production process according to claim 1, wherein said input heat amount calculation unit calculates an input heat amount corresponding to said target furnace temperature after being corrected by said target furnace temperature correction unit. . A coke production process control method for controlling the amount of heat input to a coke oven having a plurality of carbonization chambers and a plurality of combustion chambers so that the physical quantity representing the carbonization state of coke becomes a value corresponding to a target value There is
Using a physical quantity prediction model that predicts the physical quantity based on the first influencing factor including the furnace temperature, which is the temperature of the combustion chamber, and at least one of the actual value and the scheduled value of the first influencing factor, predicting the physical quantity; calculating a value of a first evaluation function including a term representing a difference between the predicted physical quantity and a target value of the physical quantity; a target furnace temperature calculation step of calculating the temperature;
A control method for a coke production process, comprising: an input heat quantity calculation unit step of calculating an input heat quantity corresponding to the target furnace temperature calculated in the target furnace temperature calculation step. A program for causing a computer to function as each part of the control device for the coke production process according to any one of claims 1 to 20.

Images