JP2001243301A - Device and method for preparing coal blending plan - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a blending plan preparing device by which the accuracy of a coal blending plan is increased. SOLUTION: This device is provided with a search processing means 2 capable of respectively providing solution candidates A by searching plural kinds of combination of a raw material blending ratio R for each brand capable of securing the quality of coke molded goods at a desired quality level with each of brands of available raw materials at the time point of plan preparation start. A preparation processing means 3 for selecting A so as not to generate the exhaustion of stock by investigating the generation of stock exhaustion within T when each of A is molded and produced at that R while using designated planning periods T and A, preparing a blending plan within T in the R of that A for each of selected A, preparing the blending plan upto the time of scheduled arrival of goods in the R of the selected A when the arrival of raw materials is scheduled within T, preparing the blending plan in the R based on solution candidates provided by searching the combination of R for each of new brands while including arriving raw materials within a range to reduce a change in the R of that A concerning a period following the time of scheduled arrival of goods, and providing the blending plan for respective solution candidates.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coke oven in which the quality of a product in coke production is ensured, the cost level is maintained at a desired level, and the raw material is coke oven at an optimum mixing ratio to meet the supply and demand of the raw material. BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coal blending plan creating device and a coal blending plan creating method capable of accurately obtaining coal blending conditions for charging coal.

[0002]

2. Description of the Related Art In an integrated steelmaking process, coal of various brands is stored in a coal yard (coal storage yard) separately for each brand, blended appropriately and charged into a coke oven to obtain a desired quality. The system is designed to produce coke and use the produced coke as fuel in a blast furnace. The quality of the coke used as fuel is important for the stable operation of the blast furnace in the integrated steelmaking process.

That is, coke quality is related to coke strength, and the higher the coke strength, the better. However, in order to produce high-strength coke, good-quality caking coal is necessary, and this good-quality caking coal is expensive because of its low output worldwide. On the other hand, non-fine caking coal, which is inferior in caking properties to caking coal, has a large output and is inexpensive. Therefore, from the viewpoint of cost, various kinds of coals are blended and used as a raw material for coke. Here, coal brands in coal mines in various locations are determined according to their components and quality, etc.In coke production, various brands of coal are blended to reduce cost and quality. It is important to have a composition that can produce a suitable coke.

[0004] Against this background, coal as a raw material for coke is transported far from coal mines around the world and brought into a steelworks, where it is blended to become a raw material for coke production. A large amount of coke is used in ironworks, so a large amount of coking coal is stored by brand in the coal yard of the steelworks, but the cost and quantity are commensurate with coking coal price fluctuations and production conditions. Coking coal is not always restocked with the same brand because it is replenished.

[0005] Accordingly, the brand composition of coal, which is a blending raw material when producing coke, is not constant throughout the year and is changed irregularly. Some coal brands have a unique effect on coke quality. For this reason, it is inevitable that the coke strength, which is one of the indices representing the quality of the generated coke, varies to some extent, but on the other hand, the coke strength required by the blast furnace must be guaranteed.

That is, it is an essential condition to guarantee the coke strength required by the blast furnace.

[0007] However, in order to reduce costs and secure profits, it is also important to reduce raw material costs, and it is necessary to avoid producing coke using coal of unnecessarily high quality.

Accordingly, as one method of coal blending planning to achieve such an object, a coal blending planning method applying MIP (Mixed Integer Programming) has been proposed (Operations Research, magazine, 1986). June,
p330-p335). The system employing this coal blending planning method has the following disadvantages.

[1] The start of the preparation of the mixing plan is the mixing ratio result of the previous day. If this mixing ratio does not satisfy all the constraints on coke production such as coke quality, the accuracy of the prepared mixing plan is This can not be guaranteed for several days. That is, it is an effect that an optimal solution could not be obtained due to poor accuracy of the initial value of the plan creation.

[2] Further, regarding a method of estimating coke quality which is most important for blast furnace operation, since a mathematical model approximated by a linear function is used, it cannot be said that the quality can be estimated with high accuracy. It is described in "" Formum "(Journal of the Iron and Steel Institute of Japan) Vol.1 No.8, 1996 p58-p
61 ", due to the effect of non-linear elements of the blended coal.

[0011] If the actually produced coke quality is far from the estimated coke quality (out of the restricted range), it is necessary to change the blending in a hurry and re-create a subsequent blending plan. Will happen.

[0012]

As described above, in the above-described conventional blending planning method, a blending plan is created by using a mixed integer programming method with the previous day's blending ratio result as an initial value. However, in this case, depending on the mixing ratio of the previous day, the quality of the produced coke is not an appropriate value, or
Even if it is said that the restriction condition as the blended coal is not satisfied, the blending plan becomes the initial value.

In this case, even if the plan creation method used is excellent, it cannot be said that the system can create a highly accurate and practical plan.

Further, regarding the method of estimating the coke quality,
Since a model approximated by a linear function is used, it cannot be said that quality can be estimated with high accuracy.

Further, since information on the estimated coke quality is used in the preparation of the blending plan, if the estimation accuracy of the coke quality is poor, the created blending plan cannot be said to have good accuracy.

Here, for example, even if the coke quality at the blending planning stage is within the constraints, if the produced coke greatly deviates from the estimated coke quality and does not satisfy the constraints, the blending is changed immediately. In addition, the subsequent formulation plan had to be recreated.

Therefore, development of a method for preparing a coal blending plan that does not have such a problem is desired.

Therefore, the object of the present invention is to
The accuracy of the coal blending plan can be improved,
An object of the present invention is to provide an apparatus and a method for preparing a coal blending plan, which makes it unnecessary to recreate a blending plan.

[0019]

In order to achieve the above object, the present invention is configured as follows. That is, the present invention is a coal blending plan creation device for charging a raw material product into a coke oven at an optimal blending ratio in accordance with the supply and demand of the raw material product, ensuring the quality of the product in the coke production. A solution candidate search in which a plurality of combinations of raw material compounding ratios for each brand that can ensure the quality of the product at a desired quality level for each brand of usable raw material at the start of planning are obtained to obtain a solution candidate. Processing means, using a designated planning period and the solution candidates obtained by the solution candidate search processing means,
In each of the solution candidates, it is checked whether or not stock out occurs within the specified planning period when the product is manufactured at the raw material mixing ratio, and a solution candidate without stock out is selected, and the selected solution is selected. For each candidate, create a mixing plan within the specified planning period based on the mixing ratio of raw materials in the solution candidate, and if there is a scheduled arrival of raw materials within the specified planning period, until the scheduled arrival time For the above, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate, and for the period after the scheduled arrival time, the received raw material product is changed as long as the change of the raw material mixing ratio of the solution candidate is small. By re-searching the combination of raw material compounding ratios for each new brand, including the new product, and creating a compounding plan with the raw material compounding ratios based on the solution candidates obtained, the specified planning period can be reduced. Characterized by comprising a blend planning processing means for obtaining blending plan by solution candidates to.

Second, the solution of the combination of the mixing ratio of the raw materials for each brand that can ensure the quality of the product at the desired quality level for each brand of the raw materials that can be used at the start of the planning is randomized. Solution candidate search processing means for obtaining a plurality of types of quasi-solution candidates searched for each start point by obtaining as a quasi-solution candidate the optimum solution found first by starting the search from the starting point selected in the above-mentioned, and designated plan period information And using the quasi-solution candidates obtained by the solution candidate search processing means, and when each of the quasi-solution candidates is manufactured at a raw material mixing ratio determined by the quasi-solution candidates, the stock runs out within the designated planning period. Investigate whether or not it will occur, select quasi-solution candidates with no out-of-stock occurrences, and for each of the selected solution candidates, use the raw material mixture ratio in that solution candidate within the specified planning period. If a raw material product is scheduled to be received within the designated planning period, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate until the scheduled arrival time, and the For the period after the scheduled time, new combinations of raw material mixing ratios for each brand, including incoming raw materials, were obtained by searching again as long as the change in the raw material mixing ratio of the solution candidate was small. It is characterized by comprising a blending plan creation processing means for creating a blending plan at a raw material blending ratio based on the solution candidate, thereby obtaining a blending plan for the solution candidate over the designated planning period for each solution candidate.

That is, according to the present invention, the quality of the product in the coke production is ensured, the cost level is maintained at a desired level, and the raw material is loaded into the coke oven at an optimum mixing ratio corresponding to the supply and demand of the raw material. In order to be able to accurately obtain the blending conditions of coal to be added, each brand that can ensure the quality of the product at the desired quality level with each brand of raw material that can be used at the start of planning Starting from a randomly selected start point, a solution of the combination of the raw material composition ratios is obtained and obtained as solution candidates, thereby obtaining a plurality of types of solution candidates searched for each start point. Investigate whether out-of-stock will occur within the specified planning period when products are manufactured at the ratio, and the raw material mixture ratio for each brand by solution candidates without out-of-stock will be suboptimal. And adopted, raw products blending ratio for each stock by candidate solutions with stock generated and rejected,
In addition, when a new brand is scheduled to arrive within the designated planning period, a plurality of combinations of raw material compounding ratios for each new brand including the new brand are searched for again as solution candidates, and throughout the designated planning period, This is to determine the raw material compounding ratio for each brand satisfying the constraint condition.

According to the conventional method of preparing a blending plan, the blending ratio at the start of the planning is the blending ratio of the previous day, and it cannot be said that this blending ratio satisfies all the constraints on coke production. Therefore, the drawback that the accuracy of the prepared mixture ratio could not be guaranteed over several days due to the poor accuracy of the initial value of the plan creation was determined by randomly selecting a solution search start point and finding a plurality of sub-optimal solutions. And the fact that a blending plan is created from a plurality of sub-optimal solutions obtained thereby, the problem can be solved.

That is, by randomly selecting a solution search start point and obtaining a plurality of quasi-optimal solutions, the optimal solution in the range near the start point, which is not the optimal solution for the entire solution, is obtained. As a result, a solution within a practically acceptable range can be obtained within a time within which there is no practical problem. Will be able to do it. By executing the planning using these solutions, the accuracy of the initial value of the planning is ensured, and the accuracy of the prepared mixing ratio can be guaranteed from the beginning.

Further, according to the present invention, in obtaining the quasi-solution candidate by the solution candidate search processing means, a plurality of search start points are extracted by an appropriate function, and a change amount based on a raw material mixing ratio is extracted. By performing a neighborhood search, a plurality of solution candidates for combinations of sub-optimal raw material compounding ratios for each brand are searched. As a result, the accuracy of estimating the coke strength can be maintained, and the accuracy of the prepared blending plan can be guaranteed.

That is, since the coke quality estimation method used in the conventional blending plan uses a model approximated by a linear function, the nonlinear elements of the blended coal including the estimation error and the measurement error are considered. As a result, the accuracy of estimating the coke strength was poor. Therefore, although the accuracy of the blending plan could not be guaranteed, in the present invention, the coke strength was determined by inputting the blended coal characteristics and the operating conditions as parameters to determine the coke strength by using a linear regression equation and a complementary neural network that complements the estimated value. This can be solved by applying the coke quality estimation method to be estimated. Therefore, according to the present invention, it is possible to provide an apparatus and method for preparing a coal blending plan, which can create a blending plan with higher accuracy than the conventional method.

[0026]

Embodiments of the present invention will be described below with reference to the drawings. The present invention first searches for a solution of a combination ratio combination, acquires a plurality of solution candidates,
A blending plan is created using each of the obtained solution candidates, and the result is presented. The configuration of the system of the present invention includes an input operation unit 1 for inputting conditions and the like as shown in the block diagram of FIG. A solution candidate search processing unit 2 that searches for solutions of combination ratios of various brands based on the input information from the input operation unit 1 and obtains a plurality of solution candidates. It is composed of a blending plan creation processing unit 3 for making and displaying the results, and a display unit 4 for outputting and displaying the processing results of the blending plan creation processing unit 3.

FIG. 2 is a flowchart of an initial solution search method for searching a plurality of examples of the combination ratio of each brand coal (solutions of the combination ratio of various brands) at the start of the planning.

<Search for Combination Ratio Combination> An initial solution search method will be described with reference to FIG.

(Step S0): The input operation unit 1 is operated to set a term condition. The target to be set is “target coke quality,“ initial solution candidate number ”,“ mixing ratio upper / lower limit ”,“ coal coal quality upper / lower limit ”, where“ target coke quality ”is a production target. It is a setting of how much coke quality is to be set, specifically, how much the drum strength is set as the target value, and the number of initial solution candidates is the number of initial solution candidates. That number,
“Blending ratio upper and lower limits” are the values of the maximum blending ratio and the minimum blending ratio of coal to be blended by brand.
The item contents shown in are the targets.

When the setting of the initial conditions is completed, the search for the combination of the mixing ratios can be performed.

The search process is as follows.

(Step S1): First, information on the current coal yard stock for each coal brand and information on the stock plan of each coal are input.

Here, the information of the current yard stock means, for example, in the case of coking coal to be transported by ship and supplied to the coal yard, the “name of the vessel”, the “arrival date”, and the “arrival quantity” for each brand. And "current stock quantity" and "coal grade information". Further, the coal arrival information is information composed of “vehicle name”, “arrival scheduled date and time”, “arrival quantity”, and “coal quality information” for each brand.

(Step S2) Next, based on the information on the current coal yard stock for each coal brand and the information on each coal arrival plan, the designated planning period (until the next coal arrival based on the coal arrival plan) Calculate the usable quantity of each coal brand in the period).

In this method, the stock amount is calculated in time series for each brand under the condition that "the coal of the current brand must be used throughout the designated planning period".

(Step S3): Setting information of “target coke quality”, “initial solution candidate number”, “mixing ratio upper / lower limit” and “blended coal quality upper / lower limit” set in step S0, and calculated in step S2 Based on the usable quantity of each coal brand, the target coke quality, the blending ratio of each brand, and the blending ratio that satisfies the constraints of the blended coal grade are searched. This is performed using a coke quality prediction system. Details will be described later.

(Step S4): It is checked whether or not the obtained mixing ratio satisfies the stock condition, that is, whether or not the mixing can be performed as calculated. Here, if the stock condition is not satisfied, the initial mixture ratio is searched again. That is, the process returns to the process of step S3, and the initial blending ratio is searched again by executing the process here.

As a result of the judgment in step S4, when the obtained mixture ratio satisfies the stock condition, that is, when the mixture can be performed as calculated, this is adopted as one of the initial solution candidates, and is stored in the storage means. The process proceeds to S5.

(Step S5): Here, it is checked whether or not the mixing ratio has been determined for the initially set number of initial solution candidates. That is, it is checked whether the number of initial solution candidates stored in the storage unit has reached the set number. Here, if the mixture ratio has not yet been determined for the set number of initial solution candidates, the process returns to step S3 to search for the initial mixture ratio again.
The above processing is performed. On the other hand, if the mixture ratio has been obtained for the number of initial solution candidates, the process ends.

According to the above-described method, the solutions A1, A2, A3,
... An-1 and An can be searched. Then, solutions A1, A2, A3,...
, An are provisional initial solution candidates.

Next, these tentative initial solution candidates A1,
A2, A3,..., An-1 and An create a blending plan in a designated planning period (within the period in which the coal arrival information is given).

<Method of Creating Mixing Plan> Provisional initial solution candidates A1, A2, A3,... A obtained by the processing of FIG.
A method of creating a mixing plan in a designated planning period from n−1, An will be described.

FIG. 3 is a flowchart relating to a method of creating a blending plan in a designated planning period (within the period in which coal arrival information is given), and will be described with reference to FIG. The processing in the flowchart shown in FIG. 3 corresponds to the processing content in the blending plan creation processing unit 3.

(Step S11): Each solution of a combination of a plurality (here, n) of compounding ratios searched by the processing in the procedure of FIG. 2 is set as a solution search start point in the solution space. In other words, it is verified that the mixing plan based on the mixing ratio is sufficient for each of the n tentative provisional initial solution candidates A1, A2, A3,... However, since simultaneous execution is not possible, the processing is advanced for each provisional initial solution candidate.

(Step S12) For this purpose, provisional initial solution candidates A1, A2, A3,... An-1, An are initialized so that they can be processed in order. Specifically, i = 0. Here, for convenience, i is i among n initial solution candidates A1, A2, A3,.
.., N−1, n (i = 1, 2, 3,..., N−1).

(Step S13): First, a candidate (A1, A2, A3) for the n types (total n types from the first to the nth) of the determined blending ratios of the coals of the respective brands is obtained.
A3,... An-1, An), a candidate (Ai) of the combination solution having the i-th combination ratio is selected.

(Step S14): At this time, i is n
Is checked (i> n + 1). If not, the process proceeds to step S15.
If it exceeds, it means that the processing of all candidates of A1, A2, A3,... An-1, An has been completed, and the processing is terminated.

(Step S15) Here, it is determined whether or not there is a future coal delivery schedule different from the currently used brand coal based on the arrival schedule information of each brand coal. That is, it is determined whether or not a new brand coal is scheduled to be received. If the new brand coal is not scheduled to be received, the process proceeds to step S19. If the new brand coal is scheduled to be received, the process proceeds to step S16.

(Step S16) Here, the same blending ratio is set until the arrival date of the new brand coal that arrives earliest. That is, it is assumed that the mixing ratio given by the i-th provisional initial solution candidate Ai currently selected is not changed until a new brand coal is received.

(Step S17): It is determined whether or not there is a brand that will run out of stock by the time the new brand coal arrives when the operation is performed with the blending ratio set in step S16. As a result, if there is a brand of coal that is out of stock, the i-th temporary initial solution candidate Ai currently selected is deleted from the candidates, and the process returns to step S13. At this time, the combination of the mixing ratios to be selected is i + 1
The candidate for the composition ratio of the unit is determined.

On the other hand, as a result of the out-of-stock occurrence determination, if there is no brand that causes out-of-stock, step S1
Move to the processing of 8.

(Step S18): A search is made for a solution having an optimal combination of the mixing ratio of each brand available at this time including the new brand coal to be received. In other words, the combination of the mixing ratios of the i-th tentative initial solution candidate Ai currently selected is kept to a minimum, and the optimum combination of the mixing ratios including the new brand coal to be received is obtained. Search for the best solution candidate. Here, as described above, the coal whose change ratio of the blending ratio of each brand coal is as small as possible is searched again and adopted as the optimal initial solution candidate.

For example, assuming that the tentative initial solution candidate currently selected is Ai, as a result of the above processing, the combination of the blending ratios of the tentative initial solution candidate Ai is reduced to a minimum. A new solution candidate Ai 'having the optimum combination of the mixing ratio including the new brand coal to be received is searched for and adopted. Then, the optimal initial solution candidate Ai 'is temporarily stored in the storage means.

After searching for an initial solution candidate having the optimum combination of the mixing ratio, the process returns to step S14, and it is determined whether i> n + 1. When the process returns to the step S14 again after the process of the step S18, since i has the same content as the previous i, the step S15
Move on to the judgment process.

(Step S19): If the result of determination in step S15 is that there is no new stock scheduled, then in this step S19 the designated planning period (the period during which the coal arrival information is given by the currently selected solution candidate) Create a mixing plan of (in). Then, this is stored in the storage means.

(Step S20): When the blending plan of each brand coal in the designated planning period by the currently selected solution candidate has been created, the blending planning for the designated planning period has been completed for all the solution candidates. It is determined whether or not. As a result, if the creation is completed, the process proceeds to step S21 in order to provide operator guidance on the plan creation result.
If the creation has not been completed, the process returns to step S13. Here, the solution of the combination of the (i + 1) th combination ratio is used.

In the case of an initial solution candidate that has undergone the processing of step S16 and subsequent steps, since the arrival of a new brand within the designated period has already been inserted, the determination made in step S15 that comes around for the second time is made. Then there will be no new stocks scheduled. Therefore, the process proceeds to step S19,
Create a mix plan for the specified period using the currently selected solution candidate. Candidate Ai until the nearest new stock arrival date
After the date of arrival of the new brand, a plan is created based on the mixture ratio of the solution candidate using the solution candidate Ai 'obtained by the search in step S18. Of the specified planning period. In this way, the i-th solution candidate Ai can be used to create a mixing plan when a new brand is scheduled to arrive within the designated planning period.

Thereafter, it is determined whether or not the preparation of the mixture plan for the designated plan period has been completed for all the solution candidates.

As a result, if the creation has been completed, the process proceeds to step S21. If the creation has not been completed, the process returns to step S13. Here, the combination of the (i + 1) th compounding ratio is used.

(Step S21): Here, the result of storing in step S19 the guidance of the blending plan created for each solution candidate by the combination solution candidates (up to the nth) of all the blending ratios for guidance to the operator. Is displayed on the display.

<Optimal Mixing Ratio Search Method> Next, among the above-described initial solution search methods used in the solution candidate search processing unit 2, the target coke quality and the target coke quality described in the processing in step S3 in FIG. A description will be given of a method of searching for a blending ratio that satisfies the constraints of each brand blending ratio and blended coal grade.

Assuming that the blending ratio of the i brand of coal is Ri and the j grade of the i brand is wij, the j grade Sj of the blended coal is Sj = Σ (Ri · Wij).

Next, the coke quality DI (DI is Drum Index
, Which is one of the indices indicating coke quality), as described in the above-mentioned document "Form", Can be obtained by And A sub-optimal solution that minimizes the evaluation function of is obtained by random start.

FIG. 4 shows a conceptual diagram of this. FIG. 4 shows a space (solution space) in which solutions are uniformly dispersed. In this space, a solution search start point is randomly selected, and an optimum solution (a minimum value in this case) is searched for in the vicinity thereof.

For example, the solution obtained by “Start 1” in FIG. 4 is the optimum solution in the solution group within the circled area of “Start 1”, and is the optimum solution over the entire solution. Not a good solution. Similarly, the solution found in “Start 2” is
This is the optimal solution in the solution group within the circled area of “Start 2”, and is not the optimal solution overlooking the entire solution. This will be described with reference to FIG.

FIG. 5 is a diagram simulating the search for a solution, in which the curve is a line of solution groups. In the case of this example, when searching for an optimal solution, for example, when the search is started from the point of “Start 1” to find a convergence point, the convergence point from “Start 1” is “ Similarly, when the search is started from the “start 2” point to find a convergence point, the convergence point from the “start 2” becomes “sub-optimal solution 2”. " However, this solution is a quasi-optimal solution at “Start 1” and a quasi-optimal solution at “Start 2”, and since the real convergence point is different, the searched one is a tentative solution candidate only. It cannot be said that the optimal solution was searched for as a whole. However, in the solution search based on the random start, a plurality of solution candidates are searched, so that a solution within a practically acceptable range can be calculated within a practically acceptable time.

Actually, the change amount ΔRi of the i brand compounding ratio
Searches for a sub-optimal solution by changing according to the following formula ΔRi = −ηi (∂E / ∂Ri), which is a method generally called the steepest descent method. Here, ηi is a correction coefficient.

However, in the present invention, when ηi <θ1, convergence adjustment is performed in accordance with ηi = ηi + α.
When ηi> θ2, convergence adjustment is performed according to ηi = ηi + β. However, no adjustment is made when θ1 <ηi <θ2. Note that θ1 and θ2 are threshold values, α and β are convergence coefficients, and θ1 <θ2.

Here, the estimated coke quality is obtained by a method using a neural network as shown in FIG. 6 using input elements as shown in Table 1 below.

[0070] FIG. 6 is a configuration example of a coke quality calculation processing function part of the solution candidate search processing unit 2. Here, as a configuration for obtaining coke quality, a regression equation calculation unit 51, a neural network 52, and their outputs are added. In addition to using the addition unit 53 for obtaining the predicted coke quality DI, and calculating the predicted coke quality DI from the information on the blended coal quality, the linear element given to the coke quality DI is calculated by the regression equation using the regression equation calculation unit 51. However, the nonlinear element given to the coke quality DI is complemented by the neural network 52. Therefore, the quality of coke having a nonlinear element can be estimated with high accuracy.

In the present invention, first, a search for a solution of a combination ratio is performed to obtain a plurality of solution candidates, and then, a composition plan is created using each of the obtained solution candidates, and the results are presented. In the system of the present invention having the above functions, the input operation unit 1 for inputting conditions and the like, and searching for the solution of the combination ratio of various brands based on the input information from the input operation unit 1 A solution candidate search processing unit 2 for obtaining a plurality of solution candidates, a composition plan creation processing unit 3 for creating a composition plan using each of the acquired solution candidates, and processing to present the result; The display unit 4 is configured to output and display the processing result of the creation processing unit 3, and the search of the combination ratio is performed by operating the input operation unit 1 by the operator. The initial conditions are set for each of the following items: number, blending ratio upper / lower limit, blended coal quality upper / lower limit, and information on the current coal yard inventory for each coal brand and information on the stock plan for each coal are entered, and the designated planning period (coal (The period until the next coal arrival based on the arrival plan), the instruction operation is performed to calculate the usable quantity for each coal brand.

Here, the information on the current yard stock means, for example, in the case of coking coal being supplied by ship to a coal yard, the “name of the vessel”, the “arrival date”, and the “arrival quantity” for each brand. And "current stock quantity" and "coal grade information". Further, the coal arrival information is information composed of “vehicle name”, “arrival scheduled date and time”, “arrival quantity”, and “coal quality information” for each brand. From these information, it is possible to calculate the available coal quantity during the designated planning period (within the period in which the coal arrival information is given).

By the operation of the input operation unit 1, the solution candidate search processing unit 2 searches for a combination ratio combination based on the input information, and obtains a plurality of set solution candidates.

That is, the solution candidate search processing unit 2 satisfies a target DI ± ΔDI (where DI is an abbreviation of Drum Index and is one of indexes indicating coke quality) as an initial solution,
In addition, solution candidates satisfying the stock condition of coal for each brand are obtained for a predetermined number of initial solution candidates. From these solution candidates, those that satisfy the available coal quantity, that is, the supply and demand, and further satisfy the operation constraint are finally selected as initial solution candidates.

FIG. 7 shows a display example of the finally obtained initial solution candidates. This is the name of the currently available coal brand (here, for example, A-1, A-2, A-3, A-
4, A-5, A-6, B-1, B-2, C-1, C-
, D-1 and E-1) and the initial candidates as combination candidates of the combination ratio are “solution candidate 1”, “solution candidate 2”, “solution candidate 3”,. 15 ".

For example, in the case of the example shown in FIG. 7, “solution candidate 1” is “11” for A-1, “7” for A-2, “19” for A-3, and “12” for A-4. ", A-5 to" 6 ", A-6
"5", B-1 "8", B-2 "7", C-1 "18", C-2 "1", D-1 "4", E-1 " 2 indicates that the solution candidate is to be blended at a blending ratio of “2”. “Solution candidate 2” is “20” for A-1, “17” for A-2, and “17” for A-3. 1 ", A-4
"15", A-5 "6", A-6 "9", B-1
Are mixed at the ratio of "1", B-2 is "9", C-1 is "2", C-2 is "11", D-1 is "6", and E-1 is "3". , And so on, and so on.

Based on the plural kinds of provisional initial solution candidates obtained in this way, the blending plan creation processing unit 3 executes the processing according to the flow described in FIG. Within the period in which the information was given)
This is displayed on the display unit 4 to provide guidance to the operator.

FIG. 8 shows an example in which guidance is provided to the operator.

The example shown here is the result of the blending plan processing in the designated planning period performed by the blending plan creation processing unit 3 based on “solution candidate 2” which is one of the finally obtained solution candidates. It is shown. The blending plan of coal by brand is shown for each day starting from the first day (eye). Here, on the 4th (eye), a new brand of coal arrived and became usable, so the search for the blending ratio of the coal is performed again. As a result, the blending ratios for some brands of coal have changed slightly. For example, coal of brand "B-6" is a brand newly received, and blending of coal of this brand has started at 3 [%], and accordingly, blending has been continued until now. The blending ratio of coal named "B-4" has been changed from 1% to 2%. Further, the blending ratio of “G-11” brand coal that has been continuously blended has been changed from 21 [%] to 16 [%].

Further, on the 10th (day), the stock of “B-3” brand, which has been continuously blended, has been received and the stock has been increased, and the coal of this brand is now available. From 1% to 2%, the amount of “C” which has been continued until now
The blending ratio of -2 "coal has been changed from 5% to 4%.

As shown in FIG. 8, the daily change of each coal brand composition for each solution candidate finally obtained is displayed on the screen of the display for guidance to the operator, so that the operator can display it on the display. From the displayed contents, it is possible to know the daily transition of each coal brand blend (raw material blend) in the optimal blend example for each solution candidate. In addition, if the coke is manufactured by operating with the blend (raw material blend) based on the daily transition of each coal brand blend (raw material blend) for each solution candidate finally obtained, a product with the desired coke grade can be obtained. At a price commensurate with cost.

Although the embodiments have been described above, the present invention is, in short, to ensure the quality of products in coke production, maintain the cost level at a desired level, and optimize the mixing ratio to meet the supply and demand of raw materials. In order to be able to accurately obtain the coal blending conditions for charging the raw materials into the coke oven at A search for a combination of raw material compounding ratios for each brand that can be secured at the level is started from a randomly selected start point and obtained as a solution candidate, thereby obtaining a plurality of types of solution candidates searched for each start point. For each solution candidate, if a product is manufactured at the raw material composition ratio, it is checked whether or not stock will be out of stock within the specified planning period. The material mix ratio is adopted as a suboptimal solution, the raw material mix ratio for each brand is not adopted due to solution candidates that have run out of stock, and when a new brand is scheduled to arrive within the designated planning period, the new A new search for multiple combinations of raw material mix ratios for each brand, including brands, is made to be a solution candidate, and the raw material mix ratio for each brand that satisfies the constraints is determined over the specified planning period. It was done.

In the conventional method of preparing a blending plan, the blending ratio at the start of the planning is the blending ratio of the previous day, and it cannot be said that this blending ratio satisfies all the constraints on coke production. Therefore, the drawback that the accuracy of the prepared mixture ratio could not be guaranteed over several days due to the poor accuracy of the initial value of the plan creation was determined by randomly selecting a solution search start point and finding a plurality of sub-optimal solutions. And the fact that a blending plan is created from a plurality of sub-optimal solutions obtained thereby, the problem can be solved.

That is, by selecting a solution search start point at random and obtaining a plurality of quasi-optimal solutions, the optimal solution in the range near the start point, which is not the optimal solution for the whole solution, is obtained. As a result, a solution within a practically acceptable range can be obtained within a time within which there is no practical problem. Will be able to do it. By executing planning using these solutions, there is an effect that the accuracy of the planning initial value is secured and the accuracy of the created mixing ratio can be guaranteed from the beginning.

In the conventional method of estimating coke quality used in the blending plan, since a model approximated by a linear function is used, the nonlinear elements of the blended coal including the estimation error and the measurement error are taken into account. As a result, the accuracy of estimating the coke strength was poor. Therefore, although the accuracy of the blending plan could not be guaranteed, in the present invention, the coke strength was determined by inputting the blended coal characteristics and the operating conditions as parameters to determine the coke strength by using a linear regression equation and a complementary neural network that complements the estimated value. This can be solved by applying a coke quality estimation method to be estimated. As a result, it is possible to provide an apparatus and method for preparing a coal blending plan, which can create a blending plan with higher accuracy than the conventional method.

The present invention is not limited to the above-described embodiment, but can be implemented with various modifications. The method described in the embodiments may be a computer-executable program such as a recording medium such as a magnetic disk (a floppy (registered trademark) disk, a hard disk, etc.), an optical disk (a CD-ROM, a DVD, etc.), and a semiconductor memory. , Or transmitted from a server via a communication line, and received by the receiving side for distribution.

[0087]

As described above in detail, according to the present invention, there is provided a coal blending plan preparation apparatus and method for preparing a coal blending plan with high accuracy within a practically acceptable time. it can.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the present invention, and is a schematic block diagram showing an example of a system configuration of the present invention.

FIG. 2 is a diagram for explaining the present invention, and is a diagram for explaining a method of searching for a mixture ratio at the time of creating a plan as a specific example of the present invention.

FIG. 3 is a diagram for explaining the present invention, and is a diagram for explaining a method of creating a mixing plan in a designated period as a specific example of the present invention.

FIG. 4 is a diagram for explaining the present invention, and is a diagram for explaining the concept of a sub-optimal solution deriving method as a specific example of the present invention.

FIG. 5 is a diagram for explaining the present invention, and is a diagram for explaining the concept of a sub-optimal solution deriving method as a specific example of the present invention.

FIG. 6 is a diagram for explaining the present invention, and is a diagram for explaining a coke quality estimating method as a specific example of the present invention.

FIG. 7 is a diagram for explaining the present invention, and is a diagram showing a display example of initial solution candidates as a specific example of the present invention.

FIG. 8 is a diagram for explaining the present invention, and is a diagram showing a display example of solution candidates as a specific example of the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input operation part, 2 ... Solution candidate search processing part, 3 ... Mixing plan preparation processing part, 4 ... Display part, 51 ... Regression formula calculation part, 52 ...
Neural network, 53 ... addition unit.

Claims (6)

[Claims] Claims: 1. A coal blending plan making apparatus for charging raw materials into a coke oven at an optimum blending ratio to meet the supply and demand of raw materials in order to ensure the quality of products in coke production, Solution candidate search processing means for searching a plurality of combinations of raw material compounding ratios for each brand capable of ensuring the quality of a product at a desired quality level for each brand of possible raw materials, and obtaining a solution candidate for each; Using the solution period obtained by the solution period search processing means and the solution candidate obtained by the solution candidate search processing means, whether or not stock out occurs during the specified planning period in the case where the product is manufactured at the raw material mixture ratio for each solution solution And select solution candidates that do not cause out-of-stock occurrence, and formulate a mixture plan for each of the selected solution candidates within the specified planning period based on the mixture ratio of raw materials in the solution candidates. In addition, when the raw material is scheduled to be received within the designated planning period, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate up to the scheduled arrival time, and after the scheduled arrival time, In the period of the above, as long as the change in the raw material mixing ratio of the solution candidate is small, the raw material based on the solution candidate obtained by re-searching the new raw material mixing ratio combination for each brand including the incoming raw material A blending plan creation processing means for creating a blending plan based on a product blending ratio to obtain a blending plan for each solution candidate over the designated planning period. 2. A coal blending plan for obtaining a combination of coal blending ratios for charging raw materials into a coke oven at an optimal blending ratio in response to supply and demand of raw materials while ensuring quality of products in coke production. From the starting point, the solution of the combination of raw material compounding ratios for each brand that can ensure the quality of the product at the desired quality level for each brand of raw material that can be used at the start of planning is Solution candidate search processing means for obtaining a plurality of types of quasi solution candidates searched for by starting points by obtaining the optimal solution found first after the search as a quasi solution candidate, and designated plan period information and the solution candidate search process Using the quasi-solution candidates obtained by the means, if the quasi-solution candidates are manufactured at the raw material composition ratio determined by the quasi-solution candidates, the product will run out of stock within the specified planning period. Investigate whether or not it will occur, select quasi-solution candidates with no out-of-stock occurrences, and create a mixture plan for the selected solution candidates within the specified planning period based on the raw material mixture ratio for that solution candidate In addition, when the raw material is scheduled to be received within the designated planning period, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate up to the scheduled arrival time, and after the scheduled arrival time, In the period of the above, as long as the change in the raw material mixing ratio of the solution candidate is small, the raw material based on the solution candidate obtained by re-searching the new raw material mixing ratio combination for each brand including the incoming raw material A blending plan creation processing means for creating a blending plan based on the product blending ratio to obtain a blending plan for the solution candidate over the designated planning period for each solution candidate. Picture creation device. 3. The solution candidate search processing means extracts a plurality of search start points by an appropriate function when acquiring the quasi solution candidate, and performs a neighborhood search with a change amount based on a raw material mixture ratio. 3. The coal blending plan creating apparatus according to claim 2, wherein a plurality of solution candidates of the combination of the sub-optimal raw material blending ratios for each brand are searched for. 4. A method for preparing a coal blending plan for charging a raw material into a coke oven at an optimum blending ratio in order to ensure the quality of a product in coke production and to meet the supply and demand of the raw material. A plurality of possible combinations of raw material compounding ratios for each brand that can ensure the quality of the product at the desired quality level for each brand of usable raw materials in the above are obtained, and solution candidates are obtained. From the candidate and the specified planning period,
In each of the solution candidates, it is checked whether or not stock out occurs within the specified planning period when the product is manufactured at the raw material mixing ratio, and a solution candidate without stock out is selected, and the selected solution is selected. For each candidate, create a mixing plan within the specified planning period based on the mixing ratio of raw materials in the solution candidate, and if there is a scheduled arrival of raw materials within the specified planning period, until the scheduled arrival time For the above, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate, and for the period after the scheduled arrival time, the received raw material product is changed as long as the change of the raw material mixing ratio of the solution candidate is small. By re-searching the combination of raw material compounding ratios for each new brand, including the new product, and creating a compounding plan with the raw material compounding ratios based on the solution candidates obtained, the specified planning period can be reduced. Coal blending planning method characterized by obtaining the blending plan and by solution candidate. 5. A method of preparing a coal blending plan for charging a raw material into a coke oven at an optimum blending ratio in order to meet the supply and demand of raw materials while ensuring the quality of the product in coke production. Starting from a randomly selected starting point to find a solution of the combination of raw material compounding ratios for each brand that can ensure the quality of the product at the desired quality level for each brand of usable raw materials, the optimal found first By obtaining a solution as a quasi-solution candidate, a plurality of types of quasi-solution candidates searched for each starting point are obtained, and each of the quasi-solution candidates is obtained from the obtained quasi-solution candidates and designated plan period information. Investigate whether out-of-stock will occur within the specified planning period when a product is manufactured at the determined raw material mixture ratio, and select a semi-solution candidate with no out-of-stock occurrence, and select the selected For each candidate, create a mixing plan within the specified planning period based on the mixing ratio of raw materials in the solution candidate, and if there is a scheduled arrival of raw materials within the specified planning period, until the scheduled arrival time For the above, a mixing plan is created based on the raw material mixing ratio of the selected solution candidate, and for the period after the scheduled arrival time, the received raw material product is changed as long as the change of the raw material mixing ratio of the solution candidate is small. By searching for new combinations of raw material mix ratios for each new brand, including the new product mix ratio, and creating a mix plan based on the raw material mix ratios based on the solution candidates obtained from the solution candidates, the mix plan over the specified planning period is set as the solution candidate A method for preparing a coal blending plan, which is obtained separately. 6. A method for obtaining a sub-solution candidate, extracting a plurality of search start points using an appropriate function and performing a neighborhood search with a change amount based on a raw material mixture ratio, thereby obtaining a sub-optimal 6. The coal blending plan creation method according to claim 5, wherein a plurality of solution candidates for combinations of the raw material blending ratios for each brand are searched.