JP6299155B2 - Cast planning device, method and program - Google Patents

Description

  The present invention relates to a cast planning apparatus, method, and program for planning a casting plan in a steelmaking process using equipment including a converter and continuous casting equipment.

  In the steel manufacturing industry, product standards and sizes are extremely diverse, and there is an increasing demand for compliance with delivery dates and shortening delivery times according to the customer's product use schedule. On the other hand, in the manufacturing industry, from the viewpoint of productivity improvement by mass production, it is required to produce multiple orders of the same chemical composition of products in steelmaking facilities in batch units. In the steel manufacturing industry, The steelmaking facility is basically a facility aimed at mass production of steel of the same composition. However, the manufacturing process consists of multiple manufacturing facilities such as steelmaking, rolling, refining, and shipping. Pursuing the productivity of lots in the steelmaking process decreases the productivity of other manufacturing facilities, and the lots in steelmaking facilities. Since the summarization leads to concentration of manufacturing load in the downstream process and causes an increase in work in progress and an increase in the manufacturing period, it is required to create a steel-out lot considering the trade-off between the manufacturing processes. In addition, pre-fabrication for making a lot causes extra product inventory and a corresponding increase in work period. In other words, it is necessary to create a steel frame arrangement plan that can cast steel of the same component as much as possible in a steel making facility while achieving equalization of load in each manufacturing process and delivery date management.

  Furthermore, in recent years, in order to further reduce production costs and improve product quality, the costs associated with changing the steelmaking lot have been minimized, and the casting position and the number of continuous castings have been regulated to improve the slab quality. A cast plan that satisfies the casting constraints must be formulated. For example, even when it is necessary to continuously cast molten steel with different steel output components, in order to minimize the mixing of different steel types and reduce scrapping costs, continuous casting of molten steel with similar chemical components as much as possible or heavy specific gravity Restrictions on continuous casting in order and restrictions on the charge of the specified steel output component to prohibit casting at the beginning of continuous casting in order to improve the quality of the slab, limiting the number of continuous casting charges in one cast A cast plan that takes into account the restrictions to be made must be made.

Patent Document 1 includes at least the input means for fetching information on the production type order information, the process processing occurrence probability for each production type, and information on the planning policy, the weight, the production type, and the steel output request date for a plurality of orders. and order database storing means for storing the order information in the order database comprising, a manufacturing varieties model storage means for storing a manufacturing varieties model is a step processes occurrence probability for each production varieties, based on the information of the order database, orders an order matrix creating means for creating a matrix, and the process capability upper limit, the process is calculated from the weights for the sum of the lag tapping amount in the previous row tapping quantity, as the step processing occurrence probability and tapped volume and weights for step load leveling degree represented by the absolute value of the difference between the load and Engineering enough capacity to at least set the weights for the evaluation value related tapping lot larger, the It is expressed by a weighted linear sum of the planning policy setting means, at least the added value of the delayed steel amount and the preceding steel amount, the process load level, and the evaluation value related to the steel production lot expansion. an evaluation function, the optimization calculation means in the minimum or maximum in the range satisfying the constraint that the amount tapping becomes less tapping capacity upper limit value, calculates the tapping frame allocation planning and manufacturing products classification appropriated frame, A steelmaking plan planning result display means for displaying a steelmaking plan planning result comprising the steelmaking frame arrangement plan and the production type allocation frame, a steelmaking plan planning result registration means for registering the steelmaking plan planning result, A method for planning a steel frame arrangement provided with the above is disclosed.

  Further, in Patent Document 2, a production plan is created from the last process to the upper process in order from the final process to the upper process, and the upper process is manufactured based on the arrangement constraints based on the entire order group obtained thereby. Create a production plan for the lot, create a production plan for the production lot of the upper process based on detailed assortment constraints based on the closest start of the order group obtained by that, and created by the above process In addition, the processing process end time information of the production lot of the upper process is connected on one time axis while adjusting the time, and the production plan is sequentially made from the passing process to the final process based on the connected upper process processing end time information. A method for creating a production plan for steel products is proposed.

Japanese Patent No. 5000547 JP 2003-256020 A

  The technique disclosed in Patent Document 1 is a technique for determining the amount of steel output for each production type, and does not consider the order of charge casting. In addition, since a mathematical optimization technique called multi-objective mixed integer programming is used, all constraints must be expressed in a linear form. However, among the complicated casting constraints, there are constraints that are difficult to represent in a linear format, and there are constraints that take too much time for optimization calculation even if expressed in a linear format. It is difficult to create a casting plan that satisfies complex casting constraints in a practical time.

  In addition, the method disclosed in Patent Document 2 is a rough plan, in which only daily capacity constraints and delivery date constraints are considered, and a daily steel charge is determined. This is a method of determining the casting order and casting of the steel output charge within the day in consideration of operational constraints. As described above, since the fine plan is prepared by correcting only a part of the rough plan, the result of the rough plan greatly affects the accuracy of the fine plan. However, the rough plan does not take any operational restrictions into consideration, and it is unlikely that the obtained rough plan is good for the planning of the fine plan. By sequentially correcting in the direction, the difference between the rough plan and the fine plan is accumulated in the future fine plan. Therefore, it is difficult to make a fine plan with an excellent evaluation value in the whole planning period.

  The present invention has been made in view of the above points, and is excellent in lot aggregation, process load leveling and delivery date optimization over the entire planning period, and can be solved by a mathematical optimization method. The aim is to be able to develop a good casting plan that satisfies difficult and complicated casting constraints in a practical time.

The gist of the present invention is as follows.
(1) A cast planning device for planning a cast plan in a steelmaking process using equipment including a converter and continuous casting equipment,
Order database storage means for storing order information including at least the weight, the production type, and the steel extraction request date for a plurality of orders in the order database;
Based on the information in the order database, the steel product types with similar production specifications are aggregated as one production type, and the orders in which the production type and the date of steel output request are the same are aggregated by weight as the same order group Order matrix creating means for creating a matrix;
Production type model storage means for storing a production type model that is the probability of process processing occurrence for each type of production;
At least the process capacity upper limit, the weight related to leveling of the process load, the weight related to the difference between the steel output request date and the steel output date, the weight related to the expansion of the steel output lot, and the weight related to the product quality are set as planning policy parameters. Planning policy setting means,
Using the order matrix, the production model, and the planning policy parameters, at least an evaluation value related to leveling of the process load, an evaluation value related to a difference between a steel output request date and a steel output date, and an evaluation related to expansion of a steel output lot The evaluation function expressed by a weighted linear sum with the value is minimized or maximized within a range satisfying the steel output amount constraint that the amount of steel output is equal to or less than the upper limit of the steel output capacity. A first optimization calculating means for calculating an allocation frame by production type, which is the amount of steel output, by a mathematical optimization method;
An initial cast plan creation means for creating an initial value of a cast plan based on the production type allocation frame calculated by the first optimization calculation means;
A weighted linear sum of at least the evaluation value for leveling the process load, the evaluation value for the difference between the date of requesting steel production and the date of steel production, the evaluation value for expanding the steel production lot, and the number of violations of casting constraints related to product quality The initial value of the cast plan is searched so that the evaluation function represented by can be minimized or maximized within a range that satisfies the constraint conditions including the steel output amount constraint that the steel output amount is equal to or less than the steel output capacity upper limit value. A second optimization calculating means for calculating a cast plan by a technique;
Output means for outputting the cast plan calculated by the second optimization calculation means,
The initial cast plan creation means determines the number of casts for each day in the planning period and the number of charges for each cast, and assigns a production type to each charge.
(2) The cast planning apparatus according to (1), wherein the first optimization calculation means uses multi-objective mixed integer programming as a mathematical optimization method.
(3) In the second optimization calculation means, the initial value of the cast plan created by the initial cast plan creation means is set as the current solution and the optimal solution,
A neighborhood solution is created by correcting a part of the current solution, a predetermined evaluation value is calculated, it is determined whether the neighborhood solution is made the current solution according to the predetermined evaluation value, and the neighborhood solution is In the case of a solution, if the predetermined evaluation value of the neighborhood solution is better than that of the optimum solution, the neighborhood solution is repeated as an optimum solution under a predetermined convergence determination rule. The cast planning apparatus according to (1) or (2).
(4) The cast planning apparatus according to (3), wherein the second optimization calculation means creates a neighborhood solution by exchanging charges from the current solution.
(5) In the second optimization means, instead of adding the number of violations as a weighted linear sum to the evaluation function for the casting constraints that must be observed among the casting constraints related to the product quality, the constraints The cast planning apparatus according to any one of (1) to (4), which is added to a condition.
(6) A cast planning method for planning a cast plan in a steelmaking process using equipment including a converter and continuous casting equipment,
The order matrix creation means manufactures one type of steel material with similar manufacturing specifications based on order database information that stores order information including at least the weight, production type, and date of request for steel production for multiple orders. A step of creating an order matrix in which orders that are aggregated as varieties and that have the same production varieties and steel output request dates are aggregated by weight as the same order group;
The planning policy setting means includes at least a process capacity upper limit value, a weight related to leveling of the process load, a weight related to the difference between the steel output request date and the steel output date, a weight related to the expansion of the steel output lot, and a weight related to the product quality. Setting as a planning policy parameter;
The first optimization calculation means uses the order matrix, the production type model that is the probability of occurrence of process processing for each production type, and the planning policy parameter, and at least an evaluation value related to leveling of the process load, The amount of steel output in which the amount of steel output is below the upper limit of the steel output capacity, using an evaluation function expressed by a weighted linear sum of the evaluation value for the difference between the requested date and the date of steel output and the value for the expansion of the steel output lot. Calculating the allocation limit by production type, which is the amount of steel output according to the production type by day, by a mathematical optimization method with the minimum or maximum within a range that satisfies the constraints;
An initial cast plan creation means creating an initial value of a cast plan based on the calculated appropriation frame by production type;
The second optimization calculation means includes at least an evaluation value relating to leveling of the process load, an evaluation value relating to a difference between the steel output request date and the steel output date, an evaluation value relating to the expansion of the steel output lot, and a casting constraint relating to product quality. The evaluation function represented by the weighted linear sum with the number of violations of the above, so as to be the minimum or maximum within the range that satisfies the constraint condition including the steel output amount constraint that the steel output amount is below the upper limit of the steel output capacity Calculating a cast plan by a search method using an initial value of the cast plan;
An output means for outputting the calculated cast plan;
In the step of creating the initial value of the cast plan, the number of casts of each day in the planning period and the number of charges of each cast are determined, and a production type is assigned to each charge.
(7) A program for drafting a cast plan in a steelmaking process using equipment including a converter and continuous casting equipment,
Order database storage means for storing order information including at least the weight, the production type, and the steel extraction request date for a plurality of orders in the order database;
Based on the information in the order database, the steel product types with similar production specifications are aggregated as one production type, and the orders in which the production type and the date of steel output request are the same are aggregated by weight as the same order group Order matrix creating means for creating a matrix;
Production type model storage means for storing a production type model that is the probability of process processing occurrence for each type of production;
At least the process capacity upper limit, the weight related to leveling of the process load, the weight related to the difference between the steel output request date and the steel output date, the weight related to the expansion of the steel output lot, and the weight related to the product quality are set as planning policy parameters. Planning policy setting means,
Using the order matrix, the production model, and the planning policy parameters, at least an evaluation value related to leveling of the process load, an evaluation value related to a difference between a steel output request date and a steel output date, and an evaluation related to expansion of a steel output lot The evaluation function expressed by a weighted linear sum with the value is minimized or maximized within a range satisfying the steel output amount constraint that the amount of steel output is equal to or less than the upper limit of the steel output capacity. A first optimization calculating means for calculating an allocation frame by production type, which is the amount of steel output, by a mathematical optimization method;
An initial cast plan creation means for creating an initial value of a cast plan based on the production type allocation frame calculated by the first optimization calculation means;
A weighted linear sum of at least the evaluation value for leveling the process load, the evaluation value for the difference between the date of requesting steel production and the date of steel production, the evaluation value for expanding the steel production lot, and the number of violations of casting constraints related to product quality The initial value of the cast plan is searched so that the evaluation function represented by can be minimized or maximized within a range that satisfies the constraint conditions including the steel output amount constraint that the steel output amount is equal to or less than the steel output capacity upper limit value. A second optimization calculating means for calculating a cast plan by a technique;
A program for causing a computer to function as output means for outputting a cast plan calculated by the second optimization calculation means,
The initial cast plan creation means determines the number of casts for each day of the planning period and the number of charges for each cast, and assigns a production type to each charge.

  According to the present invention, a high-quality cast that is excellent in lot consolidation, process load leveling, and delivery date optimization over the entire planning period, and that satisfies complex casting constraints that are difficult to solve by mathematical optimization techniques. Plans can be made in a practical time.

It is a figure which shows schematic structure of the cast plan planning apparatus which concerns on embodiment. It is a flowchart which shows the outline of the cast plan planning method by the cast plan planning apparatus which concerns on embodiment. It is a characteristic view which shows an example of the weight function provided to an evaluation function. It is a figure for demonstrating allocation with each charge and manufacturing kind in an initial cast plan preparation means. It is a flowchart which shows the process in the 2nd optimization calculation means. It is a figure for demonstrating an example of the preparation method of the neighborhood solution in a 2nd optimization calculation means. It is a figure which shows the cast plan of this invention. It is a figure which shows the conventional cast plan. It is the schematic of an example of the thick board manufacturing process in the steel industry.

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings.
First, an example of a schematic configuration of a manufacturing process (steel making process) of a thick steel plate (thick plate) which is a representative product in the steel industry will be described with reference to FIG. In FIG. 9, the arrows indicate the flow of work in progress.

  In the converter 301, a steel output component, which is a chemical component of a steel intermediate product (molten steel) in a high-temperature molten state, is adjusted in units of, for example, about 300 tons, and the steel is discharged into a molten steel pan. The steel output unit in the converter 301 is called charge.

  The continuous casting facility 302 continuously casts molten steel produced in the converter 301 for a plurality of charges, and then cuts it to a specified length to produce a plate-like intermediate product called a slab of about 20 tons, for example. To do. A series of production units in the continuous casting equipment 302 is called casting. Although it depends on the manufacturing specifications, the 8 to 12 charges are generally manufactured as one cast.

  In the rolling equipment 303, the slab is heated and then formed to a predetermined thickness and width.

  The refining (cutting) equipment 304 cuts to the size of the custom specification, the refining (correction) equipment 305 performs correction to ensure the quality of the shape, etc., and the refining (care) equipment 306 is used to ensure the quality. Products that have undergone care and have been all processed are placed in a warehouse 307. In addition, the product cut | disconnected to the size of order specification is called a plate.

  Table 1 shows an example of the size (unit) of a representative minimum production lot in each production facility of the thick plate production process. In this example, the converter 301 is about 100 times as large as 3 ton of the final product (minimum production lot in the warehouse), and the continuous production facility 302 is about 800 times larger than the final product. Manufacturing as a unit is necessary from the viewpoint of productivity and yield. However, if the productivity and yield of converters and continuous casting equipment are given priority and the delivery date is made up to the previous order and the production lot is increased, there is a problem that the product inventory increases. Further, if the production lot of the converter and continuous casting equipment is increased without considering the production load of the finishing equipment 304 to 306, it will lead to an increase in work due to concentration of the production load and an increase in the manufacturing period. That is, it is also important to level the manufacturing load in each manufacturing apparatus. As described above, it is important to formulate a casting plan in which the production start timing is determined so as to satisfy the conflicting problems of expansion of the production lot, leveling of the production load, and compliance with the delivery date.

  The purpose of drafting the cast plan is to simultaneously satisfy the mutually contradicting requirements of the unfinished steel orders that have been put in place, such as expansion of production lots, compliance with delivery dates, and leveling of production processes. It is also an issue to satisfy the conflicting requirements at the same time. In order to solve this problem, the steel product types with similar manufacturing specifications are aggregated as one manufacturing type, and the order data is obtained by making the orders with the same manufacturing type and the desired date of steel production into the same order group. Create a cast plan that meets the requirements of simplification, lower-dimensional handling, expansion of production lots, compliance with delivery dates, and leveling of production processes.

  The continuous casting equipment 302 is the most important process for a steel mill that changes liquid steel (molten steel) to solid steel (slab), and greatly affects manufacturing cost and product quality. For example, in the continuous casting equipment 302, the molten steel whose components are adjusted in the converter 301 is continuously cooled to form a slab. However, the molten steel is mixed in the portion where the molten steel components are switched, and thus cannot be used as a product. (Scrap). In addition, the size of the dissimilar steel type mixing part (debrised part) where the molten steel component changes depends on the component of the front and rear charge and the specific gravity, and the components are basically similar, in order of specific gravity (higher specific gravity first) When cast, the different steel type mixing part becomes smaller. On the other hand, from the viewpoint of quality, the top charge for starting casting of the continuous casting equipment 302 is not stable, and the quality of the slab is inevitably lowered, so that a high quality slab cannot be cast at the top of the cast. In addition, when a charge of a special component is cast, it is necessary to cast at the end of the casting because of a problem of deterioration in quality of a charge to be cast next. There are many such casting constraints, and in order to reduce manufacturing costs and improve product quality, it is important to create a casting plan that complies with these casting constraints.

FIG. 1 shows a schematic configuration of a cast planning apparatus according to the present embodiment. This cast plan planning apparatus includes an order database storage means 101, an order matrix creation means 102, a production type model storage means 104, a planning policy setting means 105, a first optimization calculation means 106, an initial cast plan creation means 107, a second Optimization calculation means 108, cast plan result display means 109 and cast plan result registration means 110, which are output means. The order database storage unit 101 stores order information including at least the weight, the production type, and the date of steel extraction request for a plurality of orders in the order database. The production type model storage means 104 stores a production type model which is the probability of occurrence of process processing for each production type.
Hereinafter, the function of each means will be described in detail.

FIG. 2 shows an outline of a cast plan making method by the cast plan making apparatus according to the present embodiment.
First, the order matrix creation means 102 reads the order information of the production type from the order database storage means 101 (step S201), aggregates the types of steel materials with similar production specifications as one production type, and produces the production type and the steel output request. An order matrix 103 is created in which orders having the same date are aggregated by weight as the same order group (step S202).

  Next, the planning policy setting means 105 sets planning policy parameters that are various conditions for formulating a cast plan from information related to the planning policy (step S204). More specifically, as the planning policy parameters, for example, the process capability upper limit value (process load upper limit value), the first and second optimization calculation times, the first and second optimization calculation convergence conditions, and the priority of each evaluation index (Weight regarding leveling of process load, weight regarding difference between output date and output date, weight related to expansion of output steel lot, weight related to product quality, etc.).

  Next, in the first optimization calculation means 106, the production type model is read from the production type model storage means 104, and the order matrix 103 is read from the order matrix creation means 102 (step S203). Then, using the planning policy parameters read from the production model, the order matrix 103, and the planning policy setting means 105, at least an evaluation value related to the leveling of the process load and an evaluation value related to the difference between the steel output request date and the steel output date And an evaluation function expressed by a weighted linear sum of the evaluation value related to the steel output lot expansion and a constraint expression expressing the steel output amount constraint that the steel output amount is equal to or less than the upper limit of the steel output capacity by a mathematical expression ( Step S205). Then, optimization calculation is performed by minimizing or maximizing the evaluation function by the multi-purpose mixed integer programming method (step S206), and an allocation frame for each production type that is a steel output amount for each production type is calculated.

Next, the initial cast plan creating means 107 creates an initial cast plan that is an initial value of the cast plan based on the appropriation frame for each product type calculated in step S206 (step S207).
Next, in the second optimization calculation means 108, at least an evaluation value related to leveling of the process load, an evaluation value related to the difference between the steel output request date and the steel output date, an evaluation value related to the steel output lot expansion, and a line Constraint conditions including the output amount constraint that the steel output is below the upper limit of the steel output capacity, using an evaluation function expressed as a weighted linear sum of the evaluation values related to the number of violations of casting constraints related to product quality that is difficult to express A cast plan is calculated by a search method using the initial value of the cast plan created in step S207 so that it becomes the minimum or maximum within the range satisfying (until the minimum value or maximum value does not change) (step S208).
Next, the cast plan calculated in step S208 is displayed by the cast plan creation result display means 109 including a display device such as a display (step S209).

  The planner checks the cast plan drafting result (step S210), and if the cast plan drafting result is favorable, the cast plan drafting result registration means 110 registers the cast plan drafting result (step S211), and the process ends. If the result is not preferable, the planning policy setting unit 105 resets the planning policy and repeats the processing steps S204 to S210.

Hereinafter, the process executed by the first optimization calculation means 106 (step S206), the process executed by the initial cast plan creation means 107 (step S207), and the second optimization calculation means 108. The process (step S208) to be performed will be described in detail.
The production types listed below mean a group of orders with the same steel composition in the converter and similar load on the finishing process. In the production line for thick plate products and steel pipe products, the load on the finishing process Because leveling is important, the production varieties consist of a group of orders that have the same steelmaking process and finishing process pattern. For example, a production type having a steel output component A and a passing process pattern of 100... 1 is represented by a symbol “A_100. In this way, the production varieties have a structure in which the steel output components are subdivided into a refined passage process pattern (classification of the steel output components as a large classification and a passage process pattern as a small classification). It is sufficient if orders with similar loads can be put together. For example, if the load of the finishing process is almost the same for each field (shipbuilding field, building field, industrial machine field, etc.) Production varieties may be used in which the components are classified into large categories, the fields are classified into medium categories, and the presence or absence of painting is classified into small categories. Moreover, said passage process pattern is a code in which predicted values (0: difficult to pass, 1: easy to pass) of passage of each finishing process are arranged in the order of the finishing process. As a method for calculating a predicted value, a model for predicting whether or not each refining process has passed from a manufacturing specification of an order may be constructed using a method such as a decision tree, using past manufacturing performance data as learning data. However, in the above description, the passing process pattern is a symbol in which the predicted values of the passing or not of each refining process are arranged in order. However, a symbol in which the passing frequency is discretized may be used instead of the passing or not. For example, the passing frequency of the refining process is represented by three symbols A (low frequency), B (medium frequency), and C (high frequency), and past manufacturing performance data is used as learning data, and each refining process A model that predicts the symbol (A, B, C) of the passing frequency from the manufacturing specifications of the order may be constructed, and the predicted values of the predicted model may be arranged in the order of the refinement process to form a passing process pattern.

  Table 2 shows an example of the manufactured product model (r [i] [j] [l]) stored in the manufactured product model storage unit 104. Further, the manufactured product model r [i] [j] [l] is the processing occurrence probability of the refining process for each manufactured product. As the product category of the production product model, there are I types of steel output components i, and there are J [i] types of passing process patterns j for each of the steel output components i. The number of refining steps l is L, and the order is gas (l = 1), correction (l = 2), care (l = 3),..., Test (l = L). . The production type is a combination of the steel output component and the passing process pattern, and the number of types of the production type is the sum of the values from J [1] to J [I]. The incidence of the refining process is different for each product type. For example, the production type of the steel output component A (i = 1) and the passing process pattern 100... 1 (j = 2) (hereinafter A_100... 1) has a gas process occurrence probability of 0.3, The occurrence probability is 0.1, and the occurrence probability of the care process is 0.1.

Table 3 shows an example of the order matrix (x _r [i] [j] [k]) 103 created by the order matrix creating means 102. The order matrix 103 is requested to return the production process (i, j) and how much steel on which day (k) by going back from the refining process to the lower process on the basis of the delivery date. The amount of steel required is expressed by arranging it in a grid of production type (row) and date of request for steel production (column). Here, k is a number representing a date, where the current planning date is 1, and the planning period is K days (k = 1 to K).

Table 4 shows an example of the process capability upper limit value (y _r [l] [k]) that is one of the planning policy parameters set by the planning policy setting means 105. In Table 4, the capacity of all processing days of all processes is 300 tons, but the processing capacity decreases when the process stops due to regular repair or the like.

  The first optimization calculation means 106 determines a production type-specific allocation frame x [i] [j] [k] as shown in Table 5 using a linearly described constraint equation and an evaluation function. The production type-specific allocation frame x [i] [j] [k] is the amount of production of the production type of the passage pattern j of the steel output component i on date k. Since the delivery date and the passing process of orders are various, in the order matrix 103 shown in Table 3, steels with various steel output components must be cast on various days, but the part surrounded by the thick frame in Table 5 As in the case of the steel output component A, April 2 and April 3, if the steel output component B is April 2 and April 14, if the steel output component X is April As shown on the 1st and April 14th, it can be seen that the steel components to be cast by day are gathered. As described above, the first optimization calculation means 106 is to calculate the allocation frame for each production type in which the number of types of outgoing steel components to be cast on a daily basis is small (the number of the same steel types is large). However, there are constraints and evaluation functions that must be taken into consideration in addition to the number of types of steel output components when calculating the production type allocation frame, which will be described.

  First, there is a limit to the amount of steel output, and the daily steel output S [k] on the kth day is subject to the amount of steel output indicated by equation (1). Here, S_max represents the upper limit of steel capacity for one sunrise.

  Approval by production type for the amount of steel output C [i] [k] of the steel output component i on the kth day and the process pattern j (= 1, 2... J [i]) of the steel output component i on the kth day. The total value of the frames x [i] [j] [k] and the billing surplus material β [i] [k] of the k-th day and the steel output component i are expressed by the relationship of the formula (2). Here, J [i] is the number of types of passing process patterns of the steel output component i.

It is not necessary to output steel as per the steel output request date of the order matrix 103, and it is sufficient to output steel in a range where the deviation from the steel output request date is not large. A certain amount of steel at sunrise includes those for different steelmaking dates. Approval frame x [i] [j] [k] for each production type of the k-th day, the steel output component i, and the passing process pattern j is the kth day, the steel output component i, the passing process pattern j, and the steel output request date t As a cumulative value within the planning period (K days) of the steelmaking request-specific allocation frame x _t [i] [j] [t] [k] for each production type, it is expressed by equation (3).

Since the order quantity and the production quantity as a whole are balanced, within the planning period (K days), the order matrix x _r [i] [j] [ The relationship between t], the kth day, the steel output component i, and the production type allocation frame x [i] [j] [k] of the passing process pattern j is expressed by Equation (4).

  Since the process load depends on the rate of occurrence of the process, the process load y [l] [k] on the kth day and the process number l is allocated to the kth day, the outgoing steel component i, and the passing process pattern j by production type. Formula (5) using the frame x [i] [j] [k], the production rate model r [i] [j] [l], which is the process occurrence rate of the steel output component i, the passing process pattern j, and the process number l Are related by Here, I is the number of types of steel output components.

The lot size LOT_SIZE is the processing amount of one lot (one charge) in the converter, and the amount of steel output C [i] [k] of the steel component i at the k-th sunrise and the number of lots δ _{L of the} steel component i at the k-th sunrise. Using [i] [k], the relationship is given by equation (6).

Expression (7) represents whether or not there is a steel output of the steel output component i on the kth day. It is assumed that δ _c [i] [k] is 1 if there is steel output of the steel output component i on the kth day, and 0 if not.

  However, the maximum value of C [i] [k] is M. That is, Formula (7) formulates the following Formula (8).

  Using the minimum interval days span [i] set for each steel output component, the constraint on the steel output planned date for each steel output component is expressed as shown in Equation (9).

  A certain number or more of devices for the refining process after the number of days achieve_day [l] set for each process l are secured. The relationship between the initial in-process stock_0, the k-th day, and the in-process stock [l] [k] of the process l is expressed by Expression (10).

  The constraint equation for ensuring the safety mechanism is expressed as equation (11) using safety_stock [l].

  Using the number of cast steel cups H [i] and cast number h [i] [k] per cast set for each steel output component, the constraint for steel output at an integral multiple of the cast is as shown in Equation (12) It is expressed in

  Equation (13) is a constraint on the daily steel production schedule by amount of steel production waku [i] [k] set by the planner.

  Next, evaluation functions (14) to (19) are defined. Formula (14) is an evaluation index that aims at minimizing the amount of preceding steel output and the amount of delayed steel output, that is, minimizing the difference between the steel output request date and the steel output date.

However, ref [i] [j] [k] is the cumulative value Σx _t [i] [j] [t] of the steel output equivalent for the production type of the steel output component i and the passing process pattern j up to the kth day. This is a target value for [q] (q = 1 to k), and is represented by the following equation (15).

Equation (16) is an evaluation function aimed at minimizing the difference between the total steel output S [k] on the k-th day and the steel capacity target value S _r [k] on the k-th sunrise.

  Expression (17) is an evaluation function aimed at leveling the process load, and is an evaluation function aimed at minimizing the difference between the moving average of the process load for 3 days and the process capability upper limit value.

  Equation (18) is an evaluation function aimed at minimizing bills remaining.

  Equation (19) is an evaluation function aimed at minimizing the number of dissimilar steel joints during casting (expansion of steelmaking lots).

  Further, in order to suppress excessive preceding steel and delayed steel in the minimum amount of delayed steel and the minimum amount of advanced steel, a weight function as shown in FIG. 14) to the evaluation function.

  For example, a = 2, b = 5, and c = 100. This weight function is represented as W (k, t). When W (k, t) is added to Expression (14) and a weighted linear sum of Expressions (14) to (19) is taken, a comprehensive evaluation index (21) in which each index is balanced is obtained.

Here, W ₁ , W ₂ , W ₃ , W ₄ , and W ₅ are respectively the degree of compliance with the date of steelmaking request, that is, the degree of difference between the date of steelmaking request and the date of steelmaking, the degree of achievement of the steel output target amount, and the process load. It is a relative evaluation weight with respect to leveling, minimum billing surplus, and lot summary achievement, that is, the degree of steelmaking lot expansion. That is, the setting of the priority of each evaluation index in the planning policy setting unit 105 is to set the evaluation weights W _{1 to} W ₅ .

In summary, the first optimization calculation means 106 calculates the upper limit value of the process capability and the evaluation weights W _{1 to} W ₅ of each evaluation index set by the order matrix 103, the production type model and the planning policy setting means 105. The constraint formulas (1) to (13) and the evaluation functions (14) to (19) are created, and the first optimization calculation time and the first optimization calculation convergence set in the planning policy setting means 105 are used. Optimization calculation is performed by the mixed integer programming according to the conditions, and the allocation frame x [i] [j] [k] for each product type is calculated. In addition, the optimization calculation by the mixed integer programming may use a commercially available mathematical programming solver or the like as appropriate.

  As an evaluation function for leveling the process load, the absolute value of the difference between the moving average of the process load for three days and the process capability upper limit value is used as in the formula (17). It may be the square of the difference between the moving average of the process load for 3 days and the upper limit of the process capability as in '), or the moving average of the process load in 3 days is the process as in equation (17' '). It is good also as a value exceeding the capacity upper limit. The moving average days are not limited to 3 days, and may be longer or shorter than 3 days. The moving average days are not limited to using the moving average, and the difference between the daily process load and the process capacity upper limit. An evaluation function according to

  Similar to the process load leveling, the evaluation index aimed at minimizing the amount of prior steel production and delayed steel production, that is, minimizing the difference between the steel production request date and the steel production date is limited to Equation (14). For example, it may be a quadratic function as shown in equation (14 ′). Further, W (k, t) is symmetrical in FIG. 3, but it may be asymmetrical (a large weight for the delayed steel amount and a small weight for the preceding steel amount).

  In addition, although the five evaluation indexes of the formulas (14) to (19) are minimized, among them, the formulas (14), (17), and (19) are necessary for developing a good cast plan. However, equations (16) and (18) can be omitted. This is because, with regard to the equation (16), in reality, the steel is often produced up to the steel production capacity upper limit value, so that only the constraint equation of the equation (1) needs to be considered. In addition, with regard to equation (18), since the order quantity is often equal to or greater than the steel output capacity, there is a plan to reduce the amount of billing surplus by minimizing the preceding steel output and delayed steel output. It is because it is obtained. Further, the evaluation index is not limited to the expressions (14) to (19), and an evaluation index other than the above may be added to the expression (21).

Through the above-described procedure, it is possible to calculate the production type appropriation frame x [i] [j] [k], which is the amount of steel output according to the daily production type, and the daily output by the steel output component. Steel quantity C [i] [k] is also obtained.
However, the order of casting for each steel output component and the cast (from where to where is the same cast) are not determined, and it is not possible to lead to production instructions as it is. In many cases, the production control personnel manually determine the casting order and casting so that the casting constraints are satisfied. However, the first optimization calculation means 106 optimizes the delivery date and leveling load leveling. In many cases, the cast is planned only considering the casting restrictions and the casting lot expansion, and the optimization effect of the corner is often lost.
Therefore, an initial value of the cast plan is created by the initial cast plan creation means 107, the process load leveling is performed by the second optimization calculation means 108, the difference between the desired steel output date and the output date is minimized, and the output steel lot Develop a casting plan that comprehensively considers expansion and casting restrictions.

The initial cast plan creation unit 107 creates an initial value of the cast plan based on the production type allocation frame x [i] [j] [k] calculated by the first optimization calculation unit 106.
First, from the number of charges D [k] on the k-th day calculated by the equation (22), the number of casts E [k] on the k-th day and the number of charges F [k] [m] of each cast are expressed by the equation (23). And is calculated by the equation (24).

Here, Ceiling (•) is a function that rounds up decimals to an integer, Round (•) is a function that rounds off, CAST_SIZE is the maximum number of charges that can be consecutively performed in one cast, and F [k] [m] is the first It means the number of charges of the mth cast on day k (m = 1,..., E [k]). When the total number of charges from F [k] [l] to F [k] [E [k]] is different from D [k] as a result of calculating the number of charges for all casts using Equation (24), F [k] k] [l] are adjusted by adding or subtracting 1 in order. For example, when D [k] = 28 and CAST_SIZE = 12, E [k] = 3 and F [k] [1] = F [k] [2] = F [k] [3] = 9. , F [k] [1] + F [k] [2] + F [k] [3] = 27 (≠ D [k]), so that F [k] [1] = 10 is adjusted.
The method represented by the formulas (22) to (24) is a calculation method in the case where there is one continuous casting machine, but can be easily extended even when there are a plurality of continuous casting machines. For example, if there are two continuous casting machines and the difference in capacity is 3: 4, the charge number D [k] calculated by Equation (22) is divided into 3: 4, and the number of casts and the number of charges are divided. May be calculated by Equation (23) and Equation (24).

As described above, after determining the number of casts E [k] for each day in the planning period and the number of charges F [k] [m] for each cast, each charge (the number of the mth cast on the kth day) The production type is assigned to the nth charge) (n = 1,..., F [k] [m]). As shown in FIG. 4, the allocation of each charge and the production type is a process of determining a steel output component of each charge, a steel output amount for each passing process pattern, and an order amount for each steel output request date.
First, the amount of steel output and the amount of steel output for each passing process pattern are determined in order of date, manufacturing product number, and passing process pattern number from the allocation frame for each manufacturing type (x [i] [j] [k]). For example, in the case of allocation of only the steel output component X on April 1 in the allocation frame by production type in Table 5, the allocation is as shown in FIG. Here, the processing amount (LOT_SIZE) of one lot in the converter is 200 tonnes. According to the allocation frame by production type in Table 5, the amount of steel output of the production type of the passage process pattern 000... 0 in the steel component X on April 1 is 440 tons. As shown in the column, 200 ton is assigned to the first charge of cast 1, 200 ton is assigned to the second charge, and 40 ton is assigned to the third charge. Moreover, according to the allocation frame by production type in Table 5, the amount of steel output of the production type of the passing component pattern 100... 0 with the steel component X on April 1 is 230 tons. As shown in the column “”, the third charge of cast 1 is assigned to 160 tons so as to be LOT_SIZE 200 tons together with 40 tons in the passing process pattern 000.
Next, the order quantity according to the desired steel production date corresponding to the production quantity according to the production type is assigned in the order from the early steel production request date. For example, as shown in FIG. 4, in the first charge of cast 1, 200 ton steel is produced for the production type of passing process pattern 000. According to the other order quantity, the order of the production date of the production process of the production process component 000... As shown, 200 tons are allocated to the first charge of cast 1. The second charge will also produce 200 tons of the production type, but the remaining 30 tons of the order quantity of the production type on April 1 will be allocated, and the remaining 170 tons will be placed on orders on April 2 and April 3 , And allocated to 50ton and 120ton respectively.
As a result, it is possible to assign the charge and the production type. However, using the production type-specific steel request date allocation frame x _t [i] [j] [t] [k] You may determine the order quantity according to the steel production request day according to steel quantity.
In this way, the casting of each day in the planning period and the number of charges, the steel output composition of each charge and the amount of steel output by production type, and the order amount by date of steel output request are determined, and this is the initial value of the cast plan. Become.

The second optimization calculation means 108 uses the search method to create a cast plan that takes into account complicated casting constraints.
FIG. 5 shows processing in the second optimization calculation means 108. In the following, as a casting constraint related to product quality, three heavy constraints that should be strictly observed and two light constraints that should be preferably observed will be described as an example. Note that the content of casting constraints and the distinction between heavy constraints and light constraints differ for each continuous casting facility and product type, and the following is just an example.
(Casting Constraint 1: Heavy Constraint) The specified steel component charge must not be cast at the beginning of the cast (not startable).
(Casting constraint 2: Heavy constraint) The specified steel output component charge must be cast at the end of casting (last designation).
(Casting restriction 3: Heavy restriction) The specified steel output component must be less than or equal to the number of charges specified in one cast (restricted number of consecutive).
(Casting constraint 4: light constraint) It is better not to continuously cast the specified two steel output components (dissimilar steel joint compatibility).
(Casting restriction 5: Light restriction) It is better to cast the charge in descending order of specific gravity (in order of specific gravity).

First, the initial cast plan created by the initial cast plan creation unit 107 is used as a current solution and an optimal solution, and an evaluation value is calculated (step S501). The evaluation function is the following equation (26) that is a function that returns the total value of the evaluation value J _S calculated by the equation (21) and the following equation (25).

Here, _JH is an evaluation index for drafting a cast plan that complies with heavy constraints on product quality. _JL is an evaluation index for drafting a cast plan that complies with light constraints on product quality. v _{1 to} v ₅ are the numbers of charges that violate the casting constraints 1 to 5 (the number of times the constraint is violated), and W _{6 to} W ₁₀ are their evaluation weights. The magnitudes of the values of W _{6 to} W ₁₀ are determined depending on the importance of casting constraints, and the evaluation weights of heavy constraints that need to be strictly observed are set larger than the light constraints. The values of these evaluation weights are tested for optimization calculation under the conditions of several cases, adjusted so as to obtain a desirable plan result, and the specified steel output components of casting constraints 1 to 4 and the output steel components of casting constraint 5 are adjusted. Are read by the planning policy setting means 105.

Casting constraints 1 to 5 are constraints related to the casting order of charges, and although it is difficult to represent each in a linear format (not impossible, the number of variables is enormous), but as shown in FIG. If the charge to be cast is determined, it can be easily determined whether or not the casting constraints 1 to 5 are violated, and the evaluation value J _S of the equation (25) can be calculated. If heavy constraints that need to be strictly observed are incorporated in the evaluation function, a cast plan that violates the heavy constraints may be drafted depending on the evaluation weight value. Is essential, instead of incorporating the heavy constraint into the evaluation function (equivalent to not adding the heavy constraint evaluation value J _H when obtaining the evaluation value J _S in equation (25)), It is also possible to make a cast plan that always obeys heavy constraints by prohibiting exchange of charge groups that violate heavy constraints. Here, among the casting constraints 1 to 3, the casting constraint related to at least one product quality needs to be considered by the second optimization calculation means 108 as described above. Any one or both may be added to the formula (25) as an evaluation index as necessary. In addition, an evaluation index related to the manufacturing cost may be added to Expression (25). Similarly to the first optimization calculation means 106, the evaluation function of the equation (21) requires at least the equations (14), (17), (19), but the equations (16), (18 ) Can be omitted. Moreover, you may add evaluation indexes other than Formula (14)-(19) to Formula (21) as an evaluation index.

  Next, a neighborhood solution is created by correcting a part of the current solution, and its evaluation value is calculated (step S502). There are various methods for creating a neighborhood solution, and one example is shown in FIG. As shown in FIG. 6, it is only necessary to create a neighborhood solution by selecting two charge groups from the current solution by random numbers, order, etc., and exchanging (swapping) the two. The two charge groups may be selected from different casts as shown in FIG. 6 (a), or may be selected from the same charge as shown in FIG. 6 (b) (however, the two charge groups cannot overlap). ). Further, FIG. 6 shows a case where there is one continuous casting machine, but in the case where there are a plurality of continuous casting machines, two charge groups may be selected and exchanged from different continuous casting machines. In FIG. 6, only the charge groups having the same number of charges are exchanged, but charge groups having different numbers of charges may be exchanged.

The neighborhood solution is generated by such a method, and the evaluation value is calculated. It is determined whether or not the neighborhood solution is accepted based on the magnitude relationship between the neighborhood solution evaluation value and the current solution evaluation value (step S503). In the case of local search (mountain climbing method), it is accepted only when the evaluation value of the neighborhood solution is better than that of the current solution, but the simulated annealing method (SA method) etc. Even if the evaluation value of is worse than that of the current solution, it is accepted with a certain probability.
When the neighborhood solution is not accepted, the process proceeds to the convergence determination (step S506). When the neighborhood solution is accepted, the neighborhood solution is set as the current solution (step S504). If the evaluation value of the neighborhood solution is better than that of the optimum solution, the neighborhood solution is set as the optimum solution (step S505).

  Since it is not possible to determine whether or not a true optimum solution has been obtained by the search method, convergence determination is performed according to a predetermined rule, and when it is determined that the convergence has been completed, the search is stopped (step S506). As a rule for determining the convergence, whether the specified number of repetitions has been calculated, whether the specified time has been calculated, whether the update of the optimum solution has not been performed more than the predetermined number of repetitions, or the like is used. If it is determined that it has not converged, the process returns to step S502 and the search is repeated. If it is determined that it has converged, the optimal solution obtained so far is output as a cast plan (step S507).

In this way, a cast plan that takes into account detailed casting constraints is obtained, and the calculated cast plan is output by the cast plan formulation result display means 109 (step S209). The planner can confirm the plan result, and if there is a problem, it is possible to formulate a highly accurate cast plan by re-designating the plan policy and repeating the optimization calculation.
In the above description, the unit of the adjustment load is the weight. However, an average number of sheets per unit weight may be set in advance for each manufactured product, and the unit of the adjustment load may be converted into the number.

  A cast plan was drafted under the conditions shown in Table 6 by the cast plan drafting apparatus according to the embodiment described above. The second optimization calculation means 108 performs the optimization calculation in consideration of the casting constraints 1 to 5.

  The cast plan finally obtained by applying the present invention (referred to as the cast plan of the present invention) is shown in FIG. FIG. 7 shows the cast, charge, and steel type for each day. In the cast plan of the present invention, a cast plan that satisfies all of the casting constraints 1 to 5 is drawn up, and is therefore an executable plan. In addition, from the evaluation value of “excess amount of finishing processing capacity” in Table 7, the finishing load is within twice the process capacity. Among them, the ratio of steel that could not be steeled by the date of request for steelmaking was 95% or more, and the amount of steel in advance (inventory) was about 10,000 tons, so a good casting plan was created. Yes. In this embodiment, the number of charges in one cast is fixed to 8 charges, so when exchanging two charge groups as shown in FIG. Creating a solution.

  On the other hand, FIG. 8 shows a cast plan (referred to as a conventional cast plan) obtained by the initial cast plan creation means 107 without using the second optimization calculation means 108, and Table 8 shows the evaluation values. FIG. 8 shows the cast, charge and steel type for each day. In the conventional cast plan, the charge in the thick line portion violates the casting constraint, and it cannot be connected to the production instruction as it is. On the other hand, due to the effect of the first optimization calculation means 106, the evaluation values in Table 8 are equal to or slightly better than the cast plan of the present invention, but there is almost no difference from the cast plan of the present invention. It was confirmed that the casting constraints were not satisfied at the sacrifice of the evaluation value. Also, the calculation time is 227 seconds, which is only 12 seconds (= 227-215 seconds) longer than the conventional cast plan, and the cast plan can be drawn up in a practical time.

  As described above, the mathematical optimization method can calculate a strict optimum solution, but is limited to a problem in which constraint equations and evaluation functions are expressed in a linear form. On the other hand, the search method can be applied to problems with complicated constraints and evaluation functions that cannot be expressed in a linear form, but there is a problem that the initial value is highly dependent and an optimal solution cannot be obtained if the initial value is bad. . Therefore, first, the major constraints (steeling capacity constraints) that can be expressed in a line format by mathematical optimization methods and the main evaluation functions (leveling of process load, difference between requested date and output date) The amount of steel to be allocated by production type, which is the amount of steel output by production type by day, is calculated by a mathematical planning method. By calculating the initial value of the cast plan for the entire planning period from the allocation frame by production type obtained, it does not consider complex casting restrictions that can not be expressed in linear form, but it is a major restriction such as steel production capacity, Considering major indicators such as process load, delivery date / inventory (difference between steel production request date and steel production date), steel production lot, etc., it is a good quality initial for the planning of the next search method. A value is obtained. Then, by using such a good initial value to obtain a cast plan by a search method, it is possible to satisfy a complicated casting constraint, and a cast plan with an excellent evaluation value over the entire planning period is a practical time. Can be planned.

By applying the present invention, the following effects can be obtained.
1) Since a strict optimal solution is calculated by mathematical optimization, the skeleton of the cast plan is a plan in which optimality is guaranteed.
2) Using this cast plan as an initial value, a cast plan that satisfies complex casting constraints by a search method is established, so that a plan that satisfies complex casting constraints and that is excellent in optimality can be obtained.
3) The search method has a strong dependency on the initial solution (initial plan) and is likely to fall into a local optimal solution with a poor evaluation value. However, since the cast plan obtained by the mathematical optimization method is used as the initial solution, the evaluation value is good. Since an initial solution (close to the optimal solution) is used, a plan with excellent optimality can be obtained.
4) In order to obtain an optimal solution only by the search method, optimization calculation is performed using a plurality of initial solutions (for example, 100), and the best evaluation value is obtained among the obtained local optimal solutions. In many cases, the multi-start search method of adopting a new plan is used, but the calculation time is increased by the number of initial solutions, and the time until a cast plan is obtained is enormous. By applying the present invention, since a plurality of initial solutions are not used, a cast plan with excellent optimality can be obtained in a short time.

Each means of the cast planning apparatus to which the present invention is applied is realized by a computer device including, for example, a CPU, a ROM, a RAM, and the like.
As mentioned above, although this invention was demonstrated with various embodiment, this invention is not limited only to these embodiment, A change etc. are possible within the scope of the present invention.
The present invention also provides software (program) that implements the functions of the present invention to a system or apparatus via a network or various storage media, and the system or apparatus computer reads out and executes the program. It is feasible.

  101: Order database storage means, 102: Order matrix creation means, 104: Production type model storage means, 105: Planning policy setting means, 106: First optimization calculation means, 107: Initial cast plan creation means, 108: First 2 optimization calculation means, 109: cast plan planning result display means, 110: cast plan planning result registration means

Claims (7)

A cast planning device for drafting a casting plan in a steelmaking process using equipment including a converter and continuous casting equipment,
Order database storage means for storing order information including at least the weight, the production type, and the steel extraction request date for a plurality of orders in the order database;
Based on the information in the order database, the steel product types with similar production specifications are aggregated as one production type, and the orders in which the production type and the date of steel output request are the same are aggregated by weight as the same order group Order matrix creating means for creating a matrix;
Production type model storage means for storing a production type model that is the probability of process processing occurrence for each type of production;
At least the process capacity upper limit, the weight related to leveling of the process load, the weight related to the difference between the steel output request date and the steel output date, the weight related to the expansion of the steel output lot, and the weight related to the product quality are set as planning policy parameters. Planning policy setting means,
Using the order matrix, the production model, and the planning policy parameters, at least an evaluation value related to leveling of the process load, an evaluation value related to a difference between a steel output request date and a steel output date, and an evaluation related to expansion of a steel output lot The evaluation function expressed by a weighted linear sum with the value is minimized or maximized within a range satisfying the steel output amount constraint that the amount of steel output is equal to or less than the upper limit of the steel output capacity. A first optimization calculating means for calculating an allocation frame by production type, which is the amount of steel output, by a mathematical optimization method;
An initial cast plan creation means for creating an initial value of a cast plan based on the production type allocation frame calculated by the first optimization calculation means;
A weighted linear sum of at least the evaluation value for leveling the process load, the evaluation value for the difference between the date of requesting steel production and the date of steel production, the evaluation value for expanding the steel production lot, and the number of violations of casting constraints related to product quality The initial value of the cast plan is searched so that the evaluation function represented by can be minimized or maximized within a range that satisfies the constraint conditions including the steel output amount constraint that the steel output amount is equal to or less than the steel output capacity upper limit value. A second optimization calculating means for calculating a cast plan by a technique;
Output means for outputting the cast plan calculated by the second optimization calculation means,
The initial cast plan creation means determines the number of casts for each day in the planning period and the number of charges for each cast, and assigns a production type to each charge.   The cast planning apparatus according to claim 1, wherein the first optimization calculation unit uses multi-objective mixed integer programming as a mathematical optimization method. In the second optimization calculation means, the initial value of the cast plan created by the initial cast plan creation means as the current solution and the optimal solution,
A neighborhood solution is created by correcting a part of the current solution, a predetermined evaluation value is calculated, it is determined whether the neighborhood solution is made the current solution according to the predetermined evaluation value, and the neighborhood solution is In the case of a solution, if the predetermined evaluation value of the neighborhood solution is better than that of the optimum solution, the neighborhood solution is repeated as an optimum solution under a predetermined convergence determination rule. The cast planning apparatus according to claim 1 or 2.   4. The cast planning apparatus according to claim 3, wherein the second optimization calculation means creates a neighborhood solution by exchanging charges from the current solution.   The second optimization means adds the number of violations to the constraint condition instead of adding the number of violations as a weighted linear sum to the evaluation function for casting constraints that must be observed among casting constraints related to the product quality. The cast planning apparatus according to any one of claims 1 to 4, characterized in that: A cast planning method for creating a casting plan in a steelmaking process using equipment including a converter and continuous casting equipment,
The order matrix creation means manufactures one type of steel material with similar manufacturing specifications based on order database information that stores order information including at least the weight, production type, and date of request for steel production for multiple orders. A step of creating an order matrix in which orders that are aggregated as varieties and that have the same production varieties and steel output request dates are aggregated by weight as the same order group;
The planning policy setting means includes at least a process capacity upper limit value, a weight related to leveling of the process load, a weight related to the difference between the steel output request date and the steel output date, a weight related to the expansion of the steel output lot, and a weight related to the product quality. Setting as a planning policy parameter;
The first optimization calculation means uses the order matrix, the production type model that is the probability of occurrence of process processing for each production type, and the planning policy parameter, and at least an evaluation value related to leveling of the process load, The amount of steel output in which the amount of steel output is below the upper limit of the steel output capacity, using an evaluation function expressed by a weighted linear sum of the evaluation value for the difference between the requested date and the date of steel output and the value for the expansion of the steel output lot. Calculating the allocation limit by production type, which is the amount of steel output according to the production type by day, by a mathematical optimization method with the minimum or maximum within a range that satisfies the constraints;
An initial cast plan creation means creating an initial value of a cast plan based on the calculated appropriation frame by production type;
The second optimization calculation means includes at least an evaluation value relating to leveling of the process load, an evaluation value relating to a difference between the steel output request date and the steel output date, an evaluation value relating to the expansion of the steel output lot, and a casting constraint relating to product quality. The evaluation function represented by the weighted linear sum with the number of violations of the above, so as to be the minimum or maximum within the range that satisfies the constraint condition including the steel output amount constraint that the steel output amount is below the upper limit of the steel output capacity Calculating a cast plan by a search method using an initial value of the cast plan;
An output means for outputting the calculated cast plan;
In the step of creating the initial value of the cast plan, the number of casts of each day in the planning period and the number of charges of each cast are determined, and a production type is assigned to each charge. A program for creating a casting plan in a steelmaking process using equipment including a converter and continuous casting equipment,
Order database storage means for storing order information including at least the weight, the production type, and the steel extraction request date for a plurality of orders in the order database;
Based on the information in the order database, the steel product types with similar production specifications are aggregated as one production type, and the orders in which the production type and the date of steel output request are the same are aggregated by weight as the same order group Order matrix creating means for creating a matrix;
Production type model storage means for storing a production type model that is the probability of process processing occurrence for each type of production;
At least the process capacity upper limit, the weight related to leveling of the process load, the weight related to the difference between the steel output request date and the steel output date, the weight related to the expansion of the steel output lot, and the weight related to the product quality are set as planning policy parameters. Planning policy setting means,
Using the order matrix, the production model, and the planning policy parameters, at least an evaluation value related to leveling of the process load, an evaluation value related to a difference between a steel output request date and a steel output date, and an evaluation related to expansion of a steel output lot The evaluation function expressed by a weighted linear sum with the value is minimized or maximized within a range satisfying the steel output amount constraint that the amount of steel output is equal to or less than the upper limit of the steel output capacity. A first optimization calculating means for calculating an allocation frame by production type, which is the amount of steel output, by a mathematical optimization method;
An initial cast plan creation means for creating an initial value of a cast plan based on the production type allocation frame calculated by the first optimization calculation means;
A weighted linear sum of at least the evaluation value for leveling the process load, the evaluation value for the difference between the date of requesting steel production and the date of steel production, the evaluation value for expanding the steel production lot, and the number of violations of casting constraints related to product quality The initial value of the cast plan is searched so that the evaluation function represented by can be minimized or maximized within a range that satisfies the constraint conditions including the steel output amount constraint that the steel output amount is equal to or less than the steel output capacity upper limit value. A second optimization calculating means for calculating a cast plan by a technique;
A program for causing a computer to function as output means for outputting a cast plan calculated by the second optimization calculation means,
The initial cast plan creation means determines the number of casts for each day of the planning period and the number of charges for each cast, and assigns a production type to each charge.

Images