JPH08287145A - Device and method for planning production schedule - Google Patents

Abstract

PURPOSE: To plan a semioptimum production schedule, which suppresses and stabilizes the maximum value of the volume of intermediate inventories between processes, in a time within a practical range by planning a production schedule in consideration of restrictions in a postprocess and a flow of materials and a product in the postprocess. CONSTITUTION: Information on a coil to be delivered within one week up to the specified date of delivery is gathered as information on a produced coil of one week (step 1), the obtained coil information on coils which are produced in one week are put together into the same steel kind, and coils of the same steel kind are sorted by plate width (step 2). On the basis of the sorted data, respective rolling lots are composed within restrictions of a rolling quantity (step 3), and the rolling quantity when a rolling lot is constituted within the same steel kind is checked (step 4); when the rolling quantity is not sufficient, rolling lot composition is performed including coils of another steel kind which can be rolled at the same time (step 5) and when the rolling quantity is sufficient, the restrictions of product plate thickness is checked (step 6).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a production schedule planning apparatus and method used for planning a rolling schedule extending from cold rolling process to continuous annealing process in a steel production line.

[0002]

2. Description of the Related Art A "schedule creating device" disclosed in Japanese Patent Application Laid-Open No. 4-146055 uses each processability degree that takes into consideration the priority of the current process of the product and the status of the subsequent process, and The present invention relates to a process candidate product that can be processed in a process, that is, scheduling for determining the order of materials to be added to the process.

The technique disclosed in Japanese Patent Application Laid-Open No. 3-50630 is equipped with a simulator that performs a simulation by using the contents of the conclusion part in the expert system, and an expert having an evaluation part that selects an optimal rule candidate from the result. Inference is performed by simulation.

[0004]

In Japanese Patent Laid-Open No. 4-146055, scheduling is performed according to the priority of the current process and the processability of the next process, so short-term and local optimization is possible. However, it is not always sufficient in terms of suppressing and stabilizing the maximum value of the intermediate stock over a long period of time.

Further, in JP-A-3-50630, optimization is performed by providing an expert system with an evaluation section, but the efficiency of optimization is not necessarily sufficient. In addition, obtaining an optimum schedule that stabilizes the amount of intermediate stock in consideration of the post-process is a burdensome and time-consuming and difficult task even for a skilled worker.

Therefore, the problem to be solved by the present invention is to formulate the production schedule by considering the restrictions in the post-process and further considering the processing state in the post-process, that is, the flow of materials and products, A suboptimal production schedule that suppresses and stabilizes the maximum value of the amount of intermediate stock between processes can be performed within a practical range of time.

[0007]

[Means and Actions for Solving the Problems] The production schedule planning method of the present invention is as follows: (1) A unit called "rolling lot information" in which "production coil information" is input in the preprocessing unit. The genetic information of the genetic algorithm (GA) can be treated as one piece of information, and the rolling lot information is arranged in a numerical sequence to form individual information indicating the order of rolling lot information. , Generate an initial solution. At this time, it is possible to avoid a contradiction with the lower process by organizing the lot that incorporates the constraint of the lower process.
Further, by consolidating lots, there is no restriction between gene information, and it is possible to freely rearrange the order of gene information. Further, even if the initial solution is appropriately given, there is no problem in the planned schedule, and the load on the operator can be reduced.

(2) The schedule planning section formulates a schedule for each process based on the obtained solutions including the initial solution. (3) In the simulation part, a simulation is performed to predict the amount of intermediate stock between the upper process and the lower process. (4) The evaluation department statistically calculates the changes in the inventory quantity from the simulation results. This makes it possible to evaluate the schedule based on the statistical characteristics of changes in the stock amount (it is possible to respond to changes in evaluation criteria).

(5) Calculation of genetic algorithm (rollover, mutation, generational change, etc.) in the rolling lot sequence changing section
To change the order of genetic information, that is, rolling lot information, to change the solution. By using the genetic algorithm, it is possible to search for a combination of orders as a solution quickly and efficiently, and a suboptimal solution can be obtained in a shorter calculation time.

The changed solution is re-scheduled by (2), simulated by (3), evaluated by (4), and (5) the order of solutions is changed.
The above cycle is repeated until the satisfactory schedule is reached.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A production schedule planning method according to an embodiment of the present invention will be described below with reference to FIGS. First, the outline of the production line that is the target of this production schedule planning method will be briefly described. FIG. 3 is a conceptual flow of steps that flow from the continuous pickling cold rolling step (CDCM) to the continuous annealing step (CAPL) and the electric cleaning step (ECL) in steel, which is the subject of this example.

1. Hot coils are brought into the hot coil yard from the outside and held for about one month. 2. In the front coil yard of the continuous pickling cold rolling process (CDCM), there is a coil for about one day, which is associated with the continuous pickling cold rolling process rolling instruction. 3. The coil rolled in the continuous pickling cold rolling process is distributed to the front coil yard in the continuous annealing process (CAPL) and the electric cleaning process (ECL) depending on the steel type. 4. In the front coil yard of the continuous annealing process (CAPL), tin plate (BP) and cold rolled steel plate (CRS) are stocked for one day as materials for production in the continuous annealing process. 5. Unannealed galvanized steel sheet (GI-NA) is stocked as a material for production in the electric cleaning process in the front coil yard of the electric cleaning process. 6. The material of the front coil yard in the continuous annealing process is passed through the continuous annealing process and the coil is sent to the packing line. 7. The material of the front coil yard in the electric cleaning process is passed through to the front coil yard in the coil treatment process (CPL).

The above is a schematic description of the production line targeted by the production schedule apparatus of this embodiment. next,
A brief overview of production will be given.

The production items targeted by this production schedule device are classified as follows according to the grade of the steel sheet.

All of the above items are rolled in the continuous pickling cold rolling process, and then distributed to the next process according to each steel type. GI-NA is rolled with BRIGHT rolls. Other steel grades are rolled with bright rolls if the product thickness is 0.4 mm or less, and the product thickness is 0.
If it is larger than 4 mm, it is rolled by DULL roll. In this rolling, it is necessary to replace the roll after rolling a predetermined amount of rolling. The unit rolled by this one roll is called a "rolling lot".

The same steel grade does not necessarily have to be rolled in the bright roll or dull roll. However, BP and CRS are sent to the continuous annealing process, and T4, T belonging to BP
5 is manufactured by the same heat cycle, CRS D
Q, DDQ and HORO are also manufactured by the same heat cycle. These heat cycles, in this example, three patterns, namely T4, T5 pattern, CQ pattern,
There are DQ, DDQ, and HORO patterns. When changing the heat cycle, it is necessary to insert a dummy coil and adjust the temperature inside the furnace.

Further, since the product production rate of the continuous annealing process is slower than the product production rate of the continuous pickling cold-rolling process, and the stripping amount of the continuous annealing process is considerably large, steel types of the same heat cycle as much as possible. It is possible to reduce the amount of inventory by continuously passing through and reducing the dead time. On the other hand, GI-NA is sent to the electric cleaning process without passing through the continuous annealing process after passing through the continuous pickling cold rolling process and passed through. In this embodiment, GI-NA has no particular restrictions on the production amount and the strip running, and the product production rate in the electric cleaning process is not as low as that in the continuous annealing process, so inventory does not become a problem. Therefore, the target of the inventory amount control in this embodiment is the front coil yard in the continuous annealing process.

Here, FIG. 1 shows a schematic view of a production schedule planning method for controlling the inventory quantity of the front coil yard in the continuous annealing step. The scheduler of the present embodiment applies a genetic algorithm (GA), and in order to apply this GA, it is necessary to associate production plan information with genetic information. In this embodiment, the production coil information, which is the production plan information, is integrated into rolling lot information, and each piece of rolling lot information is associated with genetic information. The procedure for collecting information from production coil information into rolling lot information is shown below.

First, as a premise, the production coil information for one month is given to the production coil information file. Table 1
The production coil information illustrated in (1) shows a part of the table in which the production coil information for one week is extracted from the production coil information for one month.

[Table 1] This file contains the coil number, which is the coil identifier. ,
Stores grade (steel type), product thickness information, material thickness information, product width information, product weight information, and delivery date information.
These pieces of information are necessary when the production coil information is organized into rolling lot information.

Based on the above-mentioned production coil information, rolling lot information is organized in consideration of each of the constraints of delivery date, strip width, product strip thickness, steel type and rolling amount. A series of flows of this method will be described with reference to the flowchart of FIG. 1) First, the information on the coils that will be delivered within one week from the file of the production coil information for one month to the specified delivery date will be displayed.
Collected as information on production coils for one week. 2) Collect the obtained information about the coils to be produced in one week into the same steel type, and sort the coils of the same steel type by plate width.

3) Based on the sorted data, each rolling lot is organized within the constraint of rolling amount. At that time, in order to make the steel types in the rolling lots the same, the rolling amounts of the rolling lots made of the same steel type should be averaged among the rolling lots. Also, in order to avoid uneven rolling schedules in each rolling lot, that is, when there are multiple products with the same steel type and the same strip width, those products are allocated to each rolling lot as evenly as possible. To do so.

4) Check the amount of rolling when a rolling lot is constructed in the same steel type. If the rolling amount is not sufficient, proceed to step 5), and if it is sufficient, proceed to step 6). 5) Perform rolling lot formation including coils of other steel types that can be rolled simultaneously. At that time, we will try to minimize the number of different steel lots. 6) Check the product thickness constraint. 7) If the product thickness does not satisfy the constraint, search for one that satisfies the constraint and satisfy the constraint.

The above procedure will be described with a simple example. For example, assume each constraint as follows. (1) Products with delivery time constraints of 1/2 to 1/8 (2) Rolling amount constraints Rolling with 50 to 60 tons per roll (3) Strip width constraints Product width that is rolled continuously from wide to narrow The amount of change is within 200 mm (4) Product thickness constraint Magnification of product thickness that is continuously rolled = Thick product / Thin product ≥ 2.5 (5) Steel type restriction Organize the same lot with the same steel type as much as possible

By the procedure of (1), the production schedule of the coil for one week from 1/2 to 1/8 is collected. This allows
The delivery time constraint can be satisfied.

By the procedure of (2), the same steel types are collected and sorted within the same steel type. Table 2 is a table showing sorted production coil information for simply explaining the aggregation of information into rolling lot information. As a result, with respect to the plate width constraint, it is possible to carry out the plate passing schedule in the rolling lot from wide to narrow.

[Table 2]

According to the procedure of (3), the temporary rolling lots are organized according to the sort order. The rolling lot organized by the procedure of (4) is checked. In this case, the average amount rolled by one roll is 47.5 tons, and the rolling amount is not sufficient, and the check in (4) does not pass. According to the procedure of (5), different steel plates will be combined to satisfy the restrictions on the rolling amount.

Table 3 is a table showing production coil information to be added to the sorted production coil information for simply explaining the aggregation of information into rolling lot information. In addition, Table 4 shows
It is the table which added Table 3 to Table 2.

[Table 3]

[Table 4]

As a result, the rolling lot becomes 60 tons each, and there is no problem regarding the restriction of the rolling amount, and the check of (4) can be passed. By the procedure of (3) again,
Organize a temporary rolling lot.

Table 5 shows the sorted production coil information for simply explaining the aggregation of information in the rolling lot information, which shows the rolling lot before the completion of checking the product sheet thickness.

[Table 5]

In Table 5, the product plate thickness ratios of the coil No. 13 and the coil No. 16 of Lot 3 exceed the constraint of 2.5 times, and there is a problem in operation, so the check in (6) is not passed.
Then, by the procedure of (7), coil No. and coil No. 19
By replacing and, it is possible to satisfy the constraint on the product plate thickness.

Through the above procedure, the final rolling lot is knitted. Table 6 is sorted production coil information for briefly explaining the aggregation of information into rolling lot information, which shows rolling lots corrected by checking the product sheet thickness.

[Table 6]

The rolling lot information organized in this way is treated as one piece of genetic information. Table 7 is a table showing rolling lot information in the case where data of production coil information different from the data used in the above-mentioned examples in this example are integrated into a rolling lot by the same procedure as above.

[Table 7]

Table 7 shows that the lot number of each rolling lot information, the weight to be rolled in each rolling lot, each rolling lot has each line (continuous pickling cold rolling step, continuous annealing step, electric The time required for passing the steel sheet through the cleaning step), the content of the steel type rolled in each rolling lot, and the rolls used for each rolling. This "order of rolling lot information"
(Define this as "solution") as one individual information,
Apply genetic algorithm (GA) to change and evaluate the solution, make a schedule, and make a sub-optimal production schedule.

Further, the front coil yard of the continuous annealing process is evaluated as an evaluation function for applying the genetic algorithm, and it is statistically well known for the stock amount of the front coil yard of the continuous annealing process. It was decided to evaluate with a value obtained by adding three times the standard deviation to the average value. As a result, it is possible to suppress the variation and the maximum inventory amount. Figure 1 shows the procedure for planning the production schedule.

1) Rolling lot information is organized in consideration of various restrictions. The lot number of this rolling lot information is used as the genetic information in applying the genetic algorithm, and the permutation of this genetic information is the individual information, that is, the order of rolling lots that are passed through the continuous pickling cold rolling process. Fig. 4 shows the individual information having the information in Table 7, and the numbers in the squares are the rolling lot N.
o. A plurality of individuals showing the initial solution (corresponding to the initial schedule) of the order of the organized rolling lots are randomly generated (50 in this embodiment), and the process proceeds to 2).

2) Develop each schedule for a plurality of generated solutions. The schedule of the continuous pickling cold-rolling step is obtained by inserting roll exchanges generated in Table 7 between each lot in Table 7. For the schedule of the continuous annealing step, insert a dummy coil by changing the steel type in Table 7, delete the electric cleaning step material, obtain the schedule, and proceed to 3).

3) According to the order of the knitted rolling lots,
Simulate the inventory of front coil yard in the continuous annealing process and proceed to 4).

4) The schedule is evaluated by statistically evaluating the transition of the stock amount of the simulation result.

5) Various operations of the genetic algorithm for each solution
By performing (crossover, mutation, generational change, etc.), the solution is recombined, that is, the rolling lot information is replaced to change the solution. An example of the structure of the individual information of the changed solution is as shown in FIG. It shows one of the individuals generated by recombination from the individual in FIG. 4 by the genetic algorithm. Go to 2).

According to this flow, a suboptimal solution is searched for. The search for the sub-optimal solution ends when the evaluation value of the best evaluation value among the schedules remains constant for a certain number of iterations (25 times in this embodiment).

FIG. 6 shows an execution result of the production schedule planning method having the above functions. The line before applying the genetic algorithm shown in Fig. 6 shows the transition of the stock quantity of the one with the best evaluation value in the schedule based on each of the multiple initial solutions, and the line after applying the genetic algorithm is the suboptimal. The transition of the stock quantity of the best evaluation value in the schedule based on each solution when the solution search is completed is shown.

FIG. 6 shows that a schedule based on some appropriately given initial solutions is improved as a result of being calculated according to the production schedule planning method. In this way, in order to use the genetic algorithm, the production coil information is aggregated into rolling lot information, and it is possible to suppress the maximum inventory amount with the production schedule planning device that uses this rolling lot information as the gene of the genetic algorithm. The amount of inventory can be stabilized.

Although the case where the production schedule planning method and apparatus of the present invention are applied to the rolling process in steel manufacturing has been illustrated above, the present invention can also be applied to other suitable product production lines.

[0044]

As described in detail above, by designing a production schedule according to the method of the present invention, a sub-optimal production schedule that stabilizes the amount of intermediate stock between processes for a long period of time is within the practical range. Can be done in time. It can also be applied to intermediate stock for general lines that meet the prerequisites.

[Brief description of drawings]

FIG. 1 is a flowchart illustrating the procedure and contents of a production schedule planning method in a rolling line according to an embodiment of the present invention.

FIG. 2 is a flow chart illustrating a procedure of information aggregation for performing genetic configuration of GA included in the processing content of step 1 of FIG.

FIG. 3 is a diagram showing an outline of a production line to which the above embodiment is applied.

FIG. 4 is a conceptual diagram showing an example in which the order of lot Nos. Is the gene constitution in the above example.

5 is a conceptual diagram showing an example in which the order of genes in FIG. 4 is changed by GA.

FIG. 6 is a conceptual diagram showing an example of a result of simulation in the above embodiment.

[Explanation of symbols]

 CDCM Continuous pickling cold rolling process CAPL Continuous annealing process ECL Electric cleaning process GA Genetic algorithm BP, CRS, GI-NA Steel grade

Claims (3)

[Claims] 1. In a production schedule planning device in a rolling process, (1) information about a given production coil (coil number, coil weight, etc.) is aggregated into rolling lot information in consideration of restrictions of each process, and the rolling lot information is collected. A pre-processing unit that treats information as genetic information for applying a genetic algorithm and generates multiple initial solutions, (2) Schedule planning unit that formulates each production schedule from each of the generated multiple solutions, (3) For each planned production schedule, the simulation part that simulates the inventory change between the upper process and the lower process and predicts the inventory, (4) From each simulation result, the inventory change between the processes (5) Based on the results obtained by the evaluation unit and the evaluation unit that performs the evaluation, the order of the rolling lot information is calculated by the operation of the genetic algorithm, which is a semi-optimization method. Production schedule planning apparatus comprising the rolling lot reordering unit for changing the Chi solutions. 2. In the method for planning a production schedule in a rolling process, (1) information is aggregated into rolling lot information from the given production coil information (coil number, coil weight, etc.) considering the constraints of each process, The information is treated as genetic information for applying the genetic algorithm, a plurality of initial solutions are generated, (2) each production schedule is drafted from each of the generated multiple solutions, and (3) each drafted production schedule For the above, a simulation of the inventory change between the upper process and the lower process was performed, and (4) the inventory change between the processes was evaluated from each simulation result, and (5) the result obtained by the evaluation section. Based on the above, the production schedule planning method is characterized in that the order of the rolling lot information, that is, the solution is changed by the operation of a genetic algorithm which is a semi-optimization method. (1) (1) In order to apply the genetic algorithm to the product lot information by aggregating the product lot information from the given product information (product number, product weight, etc.) considering the constraints of each production process. Generating multiple initial solutions by treating them as genetic information of (2) Planning each production schedule from each of the multiple generated solutions, and (3) Upper and lower processes for each planned production schedule. Simulation of inventory changes over time, (4) Evaluation of inventory changes between processes from each simulation result, and (5) Genetic optimization, which is a semi-optimization method, based on the results obtained by the evaluation section. A method of planning a production schedule, characterized in that the order of the rolling lot information, that is, the solution is changed by the operation of a statistical algorithm.