WO2012123526A1

WO2012123526A1 - Method for production scheduling in small-scale rolling mill and system thereof

Info

Publication number: WO2012123526A1
Application number: PCT/EP2012/054525
Authority: WO
Inventors: Nan Lin; Wan Gu DU; Yang Hua LIU
Original assignee: Siemens Aktiengesellschaft
Priority date: 2011-03-16
Filing date: 2012-03-15
Publication date: 2012-09-20
Also published as: CN102682353A

Abstract

Provided in the present invention is a method for production scheduling in a small-scale rolling mill, which method comprises the steps of: distributing customer orders to corresponding sequences and randomly selected passes according to steel grade groups and product size codes thereof, so as to generate an initial production schedule; sorting the initial production schedule in ascending order according to the delivery dates thereof, and combining two orders for the same steel grade on the basis of a combining algorithm, so as to generate a combined production schedule; arranging repeatedly a new pass for a combined production schedule when the corresponding total production yield thereof is less than a critical threshold, until the corresponding production yield in the production schedule is greater than or equal to a stop value, then outputting a pass-changed production schedule; and selecting a sequence to be treated from the pass-changed production schedule, and inserting another order into this sequence or using it to replace an order in this sequence, so as to generate an order-changed production schedule. Also provided in the present invention is a system for performing this method. The present invention generally solves the scheduling problems during the production of a rolling mill, such as the distribution, arrangement, and cutting optimization of the billets in stock.

Description

Method for production scheduling in small-scale rolling mill and system thereof Technical field

The present invention relates to a production scheduling method and production scheduling system, in particular, to a production scheduling method and production scheduling system in a small-scale rolling mill. Background art

A small-scale steel mill is usually equipped with production lines, such as casting, rolling, and finishing production lines, and the like, for producing long profiles, such as steel bars, steel segments, steel rails, and the like. A production scheduling system (PPS) used to generate a production arrangement in a small-scale rolling mill

according to a customer list is very important for an

effective rolling process, because the manufacturers need to comprehensively consider the raw material condition for the production and the customers' requirements on the products when laying a production schedule.

The raw material in a rolling mill can be billets which have been formed as cold materials, or hot materials from a continuous caster. When arranging a production schedule, the manufacturers need to know the steel grade of the billets in stock, cross-section size, the weight of each billet, number of billets, and so on; at the same time, the manufacturers also need to know the customers' requirements on supply, such as the delivery date of the products, product specification (the steel grade, cross-section size, length, allowance of the products, and so on), supply amounts, and so on.

A profile usually needs several rolling passes, each pass with different rolls, and each pass arrangement has a maximum tonnage that can be rolled before the rolls change, in order to improve the utilization efficiency of each pass. The billets to be produced which can be arranged into one pass must have the same cross-section size codes and must be in the same steel grade groups, and the products completed in this pass have the same cross-section size codes.

The rolled steel is cut into the length required by customers on a hot shearer, therefore the manufacturers need to take account of both the biggest length of a cooling bed and the diversity of the products, in order to make better use of the cooling bed.

Nowadays, there is a method employing operations research to solve the problems of production scheduling and scheduling in a rolling mill, which method is currently the most common approach to establish PPS . It models the whole problem to a single optimizing module or several related optimizing modules. The advantage of establishing PPS by using an operations research process is that the optimizing modules can express clearly the objects and limitations of these arrangements and treatments, and thus it is easy to check the accuracy and the precision of the modules; at the same time, with the development of the operations research, new special purpose packets are available to solve the optimizing

modules .

However, it is difficult for an operations research process to establish a single module to comprehensively consider all aspects of the arrangement treatment, for example, when laying a rolling schedule, the manufacturers need to consider many factors, such as the type, amount of the raw materials in stock, the production yield of each billet, the production yield of each production sequence, the maximum productivity of each pass rolling, delivery dates of the orders, and so on, and because the module is large scale, it is not easy for the established module to be solved and optimized. In

addition, in order to make the modules effective, the input data of the modules need a very high precision and an

accurate time. Many studies summarize these problems as the modules of knapsack constraint problems or prize collecting travelling salesman problem. Because these modules are the difficulties of non-deterministic polynomial (NP) problems, they are usually solved by means of an heuristic algorithm.

Some studies try to preferably distribute billets among orders. They form multi-dimensional knapsack constraint problems, which are solved by the heuristic algorithm but are different from the restrictive conditions in practical problems needed to be solved.

Some further studies are based on the context of the steel production schedule, and solve a part of the whole process. Among them, some studies focus on how to distribute orders to the billets in stock and increase the production yield of the billets; other studies focus on generating the algorithm for solving the scheduling problems, and ensuring the performance of the algorithm even in the worst case; and still other studies focus on the whole processing system, providing the flow of process and describing this flow of process by diagrams. But the production scheduling method and system thereof which can solve the whole schedule and arrangement of rolling production are yet to be found.

Contents of the invention

An object of the present invention is to provide a production scheduling method and system to integrally solve these problems of, for example, distribution, arrangement, and cutting optimization of the billets in stock, and the like.

Another object of the present invention is to provide a production scheduling method and system which can improve the customers' satisfaction degree and can both achieve a high production yield and a high sequence utilization rate.

Provided in the present invention is a method for production scheduling in a small-scale rolling mill, which method comprises the steps of: (a) distributing customer orders to corresponding sequences and randomly selected passes based on steel grade groups and product size codes, so as to generate an initial production schedule ; (b) sorting the sequences in the initial production schedule according to delivery dates thereof, and combining any two of said orders for the same steel grade according to a combining algorithm, to obtain treated sequences, so as to generate a combined production schedule; (c) selecting repeatedly and randomly one of the sequences in the combined production schedule if a corresponding total production yield of the current combined production schedule is less than a critical threshold, and arranging a new pass therefor, until the combined production schedule of the changed pass is judged by means of an evaluation function as being greater than or equal to a stop value, then outputting the production schedule at this moment as a pass-changed production schedule; and

(d) selecting a sequence to be treated from the pass-changed production schedule, inserting another order into the

sequence, or using the other order to replace an order in the sequence, so as to generate an order-changed production schedule with the orders therein changed.

In another exemplary embodiment of a method for production scheduling in a small-scale rolling mill, the above-mentioned step (a) comprises:

(al) dividing the customer orders into a plurality of order groups of the same steel grade groups and product size codes ; (a2) selecting one of the order groups;

(a3) setting a new sequence, and randomly selecting a (a4) selecting an order from the order group in step (a2), and judging whether the order can be arranged into the sequence set in step (a3), if yes, then arranging the order into the sequence; if no, then creating a new sequence and arranging the order therein;

(a5) repeating step (a4), until each of the orders in the order group as selected in step (a2) has been arranged into a suitable sequence;

(a6) repeating steps (a2) to (a5) , until all the orders in all of the order groups obtained in step (al) have been arranged; and

(a7) arranging billets for all the orders in the order groups, so as to generate an initial production schedule.

And the billets therein include the billets in stock and the billets from a continuous caster.

In yet another exemplary embodiment of a method for

production scheduling in a small-scale rolling mill, step (b) further comprises:

(bl) for the initial production schedule obtained in step (a) , sorting said sequences according to the earliest

delivery date of all said orders in each of said sequences;

(b2) for all the sequences for which the treatments have not yet been completed, combining two of the orders with the same steel grade group in said sequence according to the combining algorithm;

(b3) and outputting the schedule treated by the above steps as the combined production schedule.

In yet another exemplary embodiment of a method for

production scheduling in a small-scale rolling mill,

characterized in that the combining step comprises:

(i) inputting the product length x of a first order, the product length y of a second order, and the length z of a current billet;

(ii) calculating the remaining billet length 1 obtained by arranging said first order to the current billet;

(iii) if y > x, setting the billet length allocated to the second order as: kx + l

y-x in which, k is an integer and satisfies

(y-x)[z]-yl

0 < k≤

xy and the selection of k is to minimize kx + l kx + l

y - x y - x at the same time, setting the billet length allocated to the first order as:

X

(iv) if y < x, setting the billet length allocated to the second order as : kx + x I

x -y in which, k is an integer and satisfies

(x - y) [z] + yl

0≤k <

xy and the selection of k is to maximize h+x-l h+x-l

x-y x-y at the same time, setting the billet length allocated to the first order as: kx + x-l

y

x-y

In a further exemplary embodiment of a method for production scheduling in a small-scale rolling mill, step (c) comprises:

(cl) in the case that the total production yield of the combined production schedule is less than a critical

threshold, calculating the evaluation value fo of the

combined production schedule according to an evaluation function, and setting a stop value;

(c2) selecting randomly a sequence i in the combined production schedule, arranging a new pass therefor, and calculating the evaluation value f± of the production

schedule of the changed pass according to the evaluation function;

(c3) using the evaluation values fo and f± obtained above to calculate

if the value calculated is less than the stop value, then going to step (c4), and if the value calculated is greater than or equal to the stop value, then going to step (c5) ;

(c4) repeating steps (c2) to (c3) ;

(c5) and outputting the current production schedule with the changed pass as a pass-changed production schedule. In this method, the evaluation function is expressed as: f = sequence_number + (1 - total_yield) , in which: sequence_number represents the number of sequences in the current production schedule, total_yield represents the total production yield of the current production schedule. For instance, the stop value can be 0.05.

In a further exemplary embodiment of a method for production scheduling in a small-scale rolling mill, step (d) comprises: (dl) inputting the pass-changed production schedule, dividing all of the sequences into a plurality of pass groups, and with the sequences in each pass group being of the same product size and steel grade group;

(d2) selecting a pass group for which the treatment is not yet completed, and selecting a sequence from the pass group;

(d3) judging whether another order can be inserted into said sequence or an order can be replaced by using the other order, if it is judged as yes, then inserting said other order into said sequence or using said other order to replace an order in the sequence;

(d4) selecting the next sequence from said pass group, repeating step (d3), until there is no more sequences to be treated in this pass group; and

(d5) repeating steps (d2) to (d4), until the treatment of all of the pass groups has been completed, so as to generate said order-changed production schedule.

Also provided by the present invention is a system for performing the production scheduling method of the present invention : module for generating an initial production schedule, used ' distribute the customer orders into the corresponding sequences and the randomly selected passes on the basis of the same steel grade groups and production size codes

inputted therein, so as to output an initial production schedule ; a module for generating a combined production schedule, used to sort the sequences in the initial production schedule according to the delivery dates, to combine any two of the orders with the same steel grade according to a combining algorithm, to obtain the sequences thus treated, and to output the combined production schedule; a module for generating a pass-changed production schedule, used to select repeatedly and randomly one of the sequences in the combined production schedule, to arrange a new pass therefor in the case that the total production yield of the combined production schedule is less than a critical

threshold, until the production schedule of the changed pass is judged by means of an evaluation function as being greater than or equal to a stop value, and then to output the

production schedule at this moment as the pass-changed production schedule; and a module for generating an order-changed production schedule, used to select a sequence to be treated from the pass-changed production schedule, to insert another order into said sequence or use another order to replace an order in said sequence, so as to output the order-changed production schedule with the changed order, characterized in that, the output of the module for

generating the initial production schedule is connected to the input of the module for generating the combined

production schedule, the output of the module for generating the combined production schedule is connected to the input of the module for generating the pass-changed production

schedule, and the output of the module for generating the pass-changed production schedule is connected to the input of the module for generating the order-changed production schedule .

The production scheduling method and production scheduling system of the present invention follow the delivery date needs of the customer orders by sorting the rolling sequence from the earliest to the latest, and can make sure that the long profiles are sent to customers before the delivery dates, thus improving the customers' satisfaction degree. These production scheduling method and system combine the orders into the same billet as much as possible, improves the utilization rate of billets, reduce the waste of billets as much as possible, and improve the total production yield of the whole procedure by changing passes of a certain sequence. In addition, the production scheduling method and system described herein increase orders in a sequence as much as possible or use a more suitable order to replace an original order in the sequence, so as to increase the utilization rate of the sequence.

Description of drawings

The following drawings only are used to schematically

describe and explain the present invention, and do not limit the scope of the invention.

Fig. 1 is a schematic diagram of a schematic flow of a method for production scheduling in a small-scale rolling mill.

Fig. 2 is a schematic diagram of a schematic flow for

generating an initial schedule in the production scheduling method .

Fig. 3 is a schematic diagram of a schematic flow for

generating a combined production schedule in the production scheduling method. Fig. 4 is a schematic diagram of a schematic flow for

generating a pass-changed production schedule in the

production scheduling method. Fig. 5 is a schematic diagram of a schematic flow for

generating an order-changed production schedule in the production scheduling method.

Fig. 6 is a schematic diagram of an embodiment of a system for production scheduling in a small-scale rolling mill.

Particular embodiment

In order to make the technical features, objects, and effects of the present invention clearer, now the particular

embodiments of the present invention are described in

reference to the accompanying drawings.

Fig. 1 shows a schematic flow of a method for production scheduling in a small-scale rolling mill.

At step S102: distribute customer orders to corresponding sequences and randomly selected passes according to steel grade groups and product size codes, so as to generate an initial production schedule.

At step S103: sort the rolling sequences of the initial production schedule formed in step S102 according to delivery dates, and combine any two orders having the same steel grade groups on the basis of a combining algorithm and distribute the combination into a billet, and to obtain the rolling sequences treated, and so as to generate a combined

production schedule.

At step S104: when a corresponding total production yield of the combined production schedule is less than a critical threshold, select repeatedly and randomly one of the

sequences in the combined production schedule and arrange a new pass therefor, until the production schedule of the changed pass is judged by means of an evaluation function as being greater than or equal to a stop value, then output the production schedule at this moment as a pass-changed

production schedule. At step S105: select a sequence to be untreated from the pass-changed production schedule, insert another order into the sequence, or use said other order to replace an order in the sequence, so as to generate an order-changed production schedule with the orders therein changed.

In such a production scheduling method, in order to ascertain whether or not the customer orders can be arranged into the corresponding rolling sequence and how to distribute the billets in stock and billets from a continuous caster into the corresponding orders, step S102 can use the three following modules.

An initial arrangement module:

The initial arrangement module focuses on how to arrange all the customer orders into the sequences. This can be expressed as knapsack constraint problems: OSi_j can be used to

represent a decision-making variable.

1 O_t e S_J

in which: OSv,

o o,e s

Oi represents the ith customer order, and S_j represents jth sequence . The object here is to use the least sequences in the

production schedule. Since at the beginning of the schedule period the weight of the product is fixed, this goal can achieve the highest complete spot supply rate of a small- scale steel-casting foundry. A distributing module:

The distributing module focuses on how to distribute billets in a storage to the orders. This is a matching problem that makes the orders match with their most suitable billets. The decision-making variable, goal function and limit are listed hereinafter.

Assume, OB_j represents a decision-making variable. OBi_j can be 0 or a positive integer. If it is equal to 0, this indicates that the customer order i will not use the billet j for production. Otherwise, it represents the product number of the customer order i produced by the billet j. The object of this module is to make every billet have the highest yield, and in turn lead to the highest output rate of RS (rolling sequence) . This is a matching problem that can be solved by multi-dimensional knapsack constraint problems.

A remaining order treating module: This module is used to distribute the orders not being matched with suitable billets in the distributing module to a continuous caster. The decision-making variable herein is the length of each billet produced by the continuous caster, and the object of the module is to save the billets produced from the continuous caster to the greatest extent. Compared with the above two modules, the problem of this module is

relatively simple. After completing the distribution of the modules, the length of the billets can be calculated by an algebraic method. The three modules are not separated from each other, with the results of the initial arrangement module inputted into the distributing module and the results of the distributing module inputted into the remaining order treating module. In addition, in order to obtain a better resolution, the results of the remaining order treating module is returned to the initial arrangement module. The information of these modules is related with each other, and this can be achieved by using the heuristic algorithm known as the local search.

Fig. 2 schematically describes the flow chart for generating the initial production schedule by using the above three modules .

At step S201: input the job parameters corresponding to the flow line of the rolling mill, the information of customer orders (including the delivery dates, the steel grades, and the product size codes), the stock billet amount, and so on.

At step S202: initialize the above parameters, and select the customer orders needed to be treated by the production schedule at this time. At step S203: divide the orders into a plurality of order groups for the products with the same steel grade groups and product size codes (in the following and drawings, called as "SGGS group" or "order group") , and sort them in an ascending order according to the orders' delivery dates in the order group.

At step S204: select an order group.

At step S205: set a new sequence, and randomly select a pass. At step S206: select an order.

At step S207: judge whether the order can be distributed into the sequence set currently. If yes, then go to step S208; if no, then go to step S209.

At step S208: distribute the orders into the sequence set currently .

At step S209: create a new sequence, and distribute the order into the created new sequence.

At step S210: judge whether there is an order needed to be allocated. If yes, then return to step S206, and continue on the operation of distributing the order into the

corresponding sequence; if no, then go to step S211. At step S211: judge whether there is an order group which still needs to be distributed. If yes, then return to step S204, and continue on the above operation; if no, then go to step S212.

At step S212: distribute the billets in stock to the above order, and make good use of the billets in stock. At step S213: distribute the remaining order to the billets from the continuous caster.

At step S214: sort all of the above sequences.

At step S215: set the hot shearing parameters, in order to make better use of the cooling bed, taking the length of the biggest cooling bed into consideration when setting the hot shearing parameters .

At step S216: set the cooling saw parameters.

At step S217: at the end of the flow, obtain the initial production schedule.

Fig. 3 schematically describes the flow chart of the

combining model used to generate the combined production schedule .

At step S301: input the initial production schedule. At step S302: set the delivery date of each sequence, and use the earliest delivery date of all of the orders in a sequence as the delivery date of the sequence when there are a

plurality of orders in the sequence.

At step S303: sort these sequences according to the delivery dates, and the particular sorting order being performed according to the delivery date from the first to the last, thus preferentially conduct the production scheduling for the sequence having an earlier sequence, so as to meet the customer needs. At step S304: judge whether all of the sequences have been treated, if so, then go to S305; if not, then go to step S305.

At step S305: end this flow, and output the treated schedule as the combined production schedule. At step S306: select a sequence untreated by combination. At step S307: set the total number of the orders in the sequence as N, and set i = 1, j = 2.

At step S308: judge whether j is bigger than N, if j > N, then go to step S309, otherwise go to step S310. At step S309: treat the order i, and return to step S304.

At step S310: judge whether the order i and order j have the same grade steel group, if both have the same grade steel group, then go to step S311, otherwise go to step S312.

At step S311: combine and incorporate the order i and the order j according to the combining algorithm and then use one billet to produce them, reset i and j on the basis of the distribution regulation, i.e., no longer consider the orders combined in the sequence, set the orders remained in the sequence as the order i and the order j, respectively, in which i = l_r j = 2, and return to step S308, and repeat the abovementioned flow operation again.

At step S312: re-arrange the orders, assume i = j, j = j + 1, return to step S308, conduct the operation of the above- mentioned flow again. An exemplary embodiment of the combining algorithm mentioned in step S311 is described. According to the practical

situation of the work station, normally at most two orders are distributed into the same billet:

(i) input the product length x, the product length y of the order j, and the length z of a current billet;

(ii) calculate the remaining billet length 1 obtained by arranging the order i to the current billet;

(iii) if y>x, setting the billet length allocated to the order j as : kx + l

,

_ y - x _ in which, k is an integer and satisfies:

₁

xy and the selection of k is to minimize: kx + l kx + l

y-x y-x at the same time, setting the billet length allocated to the order i as :

X

IV if y<x, the setting the billet length allocated to the order j as : kx ~\~ x I

x-y in which, k is an integer and satisfies:

(y-x)[z]-yl

0≤k≤- xy and the selection of k is to maximize kx + x - I kx + x - I

x-y x-y at the same time, setting the billet length allocated to the order i as : kx + x-l

x-y y Fig. 4 schematically describes the flow chart for generating the pass-changed production schedule.

At step S401: input the combined production schedule obtained by the flow shown in the Fig. 3. At step S402: judge whether the total production yield (i.e. product weight/billet weight) of the combined production schedule is greater than or equal to a critical threshold set according to empirical values, if the judgment result is positive, then go to step S403, if the judgment result is negative, then go to step S404.

At step S403: the flow is over, and output the combined production schedule as a pass-changed production schedule.

At step S404: set i = 0, and then calculate the evaluation value : f = sequence_number + (1 - total_yield) , in which: sequence_number represents the number of sequences in the current production schedule, total_yield represents the total production yield in the current production schedule, calculating f from the current combined production schedule, referred as fo, setting a stop value, for example, setting the stop value being 0.05 in an exemplary embodiment. The setting of the stop value is determined according to the particular requirement of the total production yield, and also can use other stop values.

At step S405: increase i by 1, and randomly select a rolling sequence .

At step S406: select a new pass arrangement for this rolling sequence . At step S407: invoke the combined model in Fig. 3 (see, steps S304-S312), and incorporate the orders in the rolling

sequence according to the pass requirements again.

At step S408: calculate f± of the current schedule according to the above evaluation function. At step S409: compare the values of f± and f±-i, if f± < fi-i, then it is considered that at this moment it is more possible to produce a higher total production yield, and go to step S410; if f >= fi-i, then it is considered that at this moment it is difficult to produce a higher total production yield, and go to step S405.

At step S410: by substitution of f± into:

/. '

judge whether the calculated result is greater than or equal to the stop value set in step S404, if it is greater than or equal to the stop value, then go to step S403, and end the flow; if it is less than the stop value, then go to step S405, and repeat the above operation.

The treatment flow described by Fig. 4 will improve the total production yield, and even in the case that the total

production yield of the combined production schedule produced according to the flow in Fig. 3 is less than the critical threshold, it is also able to obtain a certain improvement.

Fig. 5 schematically describes the flow chart for generating the order-changed production schedule. At step S501: input the pass-changed production schedule obtained by the flow in Fig. 4, divide all of the rolling sequence into a plurality of pass groups (in the following and drawings, called an "SQG group" or "pass group") , with the sequences in each pass group being of the same product size and the steel grade group. At step S502: judge whether all of the pass groups in this schedule have been treated, if they have been treated, then go to step S503, if they have not been treated, then go to step S504. At step S503: the flow is over; output the production

schedule at this moment as an order-changed production schedule .

At step S504: select a pass group for which the treatment is not yet completed. At step S505: calculate the number of sequences in the pass group and set it to N, and sort the N sequences in the pass group according to the order of 1 to N, respectively.

At step S506: set the initial value of i to 1, the number of i after each cycle will increase by 1 until it reaches N, and select the sequence i.

At step S507: judge whether there is still a sequence to be treated in this pass group, if all sequences therein have been treated, then go to step S502, otherwise go to step S508. At step S508: judge whether another order can be inserted into the sequence i, if it is judged as yes, then go to step S509, if it is judged as no, then go to step S510.

At step S509: insert the other order into the sequence i.

At step S510: judge whether the order can be used to replace a certain order in the sequence i, if it is judged as yes, then go to step S511, if it is judged as no, go to step S506.

At S511, use this order to replace that order in the sequence i. After steps S509, S510, S511, the flow goes to step S506, performing a new cycle. After treatment through the flow described in Fig. 5, the insertion of a new order into the existing rolling sequence or the use of another more suitable order to replace a certain order in the sequence improves the utilization rate of sequences.

Fig. 6 schematically shows an example of a system for

production scheduling in a small-scale rolling mill. The production scheduling system 600 encompasses a module 602 for generating an initial schedule, a module 604 for generating a combined production schedule, a module 606 for generating a pass-changed production schedule, and a module 608 for generating an order-changed production schedule. The module 602 for generating the initial production schedule can perform the treatment flow shown in Fig. 2, the module 604 for generating the combined production schedule can perform the treatment flow shown in Fig. 3, the module 606 for generating the pass-changed production schedule can perform the treatment flow shown in Fig. 4, and the module 608 for generating the order-changed production schedule can perform the treatment flow shown in Fig. 5. The output of the module 602 for generating the initial production schedule is

connected to the input of the module 604 for generating the combined production schedule, the output of the module 604 for generating the combined production schedule is connected to the input of the module 606 for generating the pass- changed production schedule, and the output of the module 606 for generating the pass-changed production schedule is connected to the input of the module 608 for generating the order-changed production schedule.

The abovementioned production scheduling method and

production scheduling system follow the delivery date needs of the customer orders by sorting the rolling sequence from the earliest to the latest, and can transport long profiles to customers before the delivery dates, thus improving the satisfaction degree of customers. These production scheduling method and system combine the orders into the same billet as much as possible, improve the use rate of billets, and raise the total production yield of the whole procedure by changing passes of a certain sequence, and increase the calculation speed. In addition, the production scheduling method and system described herein increase orders in a sequence as much as possible or use a more suitable order to replace an existing order in the sequence, so as to increase the

utilization rate of the rolling sequence.

It should be understood that although the description is described according to the examplary embodiments, each individual example may not contain only one independent technical scheme. This manner of statement in the description is only for clarity, and the skilled person in the art should regard the description as a whole, and the technical scheme in each of the exemplary embodiments can be combined properly to form other embodiments which can be understood by the skilled person in the art.

These detailed descriptions listed above are only directed to the particular description of feasibility exemplary

embodiments of the present invention, which are not used to limit the scope of the invention, and all equivalent

exemplary embodiments or alterations made without departing from the technical spirit of the invention are included in the protection scope of the present invention.

Claims

Patent claims

1. A method for production scheduling in a small-scale rolling mill, comprising the steps of: (a) distributing customer orders to corresponding sequences and randomly selected passes on the basis of the same steel grade groups and product size codes, so as to generate an initial production schedule;

(b) sorting said sequences in said initial production

schedule according to the delivery dates of the orders, and combining two of said orders with the same steel grade group based on a combining algorithm, to obtain a treated sequence, so as to generate a combined production schedule;

(c) selecting repeatedly and randomly one of the sequences in said combined production schedule if a corresponding total production yield of said combined production schedule is less than a critical threshold, and arranging a new pass therefor, until it is judged by means of an evaluation function that the combined production schedule with the changed pass is greater than or equal to a stop value, and then outputting the production schedule at this moment as a pass-changed production schedule; and

(d) selecting a sequence to be treated from said pass-changed production schedule, inserting another order into said sequence or using said other order to replace an order in said sequence, so as to generate an order-changed production schedule with the orders therein changed.

2. A method for production scheduling as claimed in claim 1, characterized in that said step (a) comprises: (al) dividing the customer orders into a plurality of order groups of the same steel grade groups and product size codes ;

(a2) selecting one of the order groups; (a3) setting a new sequence, and randomly selecting a pass ;

(a4) selecting an order from said order group in step (a2), and judging whether the order can be arranged into said sequence set in step (a3), if yes, then arranging the order into said sequence; if no, then creating a new sequence and arranging the order therein;

(a5) repeating step (a4), until each of the orders in said order group as selected in step (a2) has been arranged into a suitable sequence;

(a6) repeating steps (a2) to (a5) , until all orders in all said order groups obtained in step (al) have been arranged; and

(a7) arranging billets for all the orders in said order group, so as to generate said initial production schedule.

3. A method for production scheduling as claimed in claim 2, characterized in that said billets include billets in stock or billets from a continuous caster.

4. A method for production scheduling as claimed in claim 1, characterized in that said step (b) further comprises:

(bl) for said initial production schedule obtained in step (a) , sorting said sequences according to the earliest

delivery date of all said orders in each of said sequences;

(b2) for all the sequences for which the treatments have not yet been completed, combining two of said orders with the same steel grade group in said sequence according to the combining algorithm; and

(b3) outputting the schedule treated by the above steps as said combined production schedule.

5. A method for production scheduling as claimed in claim characterized in that the step of combining two of said orders with the same steel grade group in said sequence according to said combining algorithm comprises:

(iii) if y > x, setting the billet length allocated to the second order as : kx + l

y - x in which, k is an integer and satisfies

(y - x)[z] - yl

0≤k≤

xy and the selection of k is to minimize

at the same time, setting the billet length allocated to the first order as:

X

(iv) if y < x, setting the billet length allocated to the second order as : kx + x -l

x -y in which, k is an integer and satisfies: (x - y)\z] + yl

0≤k < ^y '^{v J y} - 1 ,

xy and the selection of k is to maximize: kx + x— l kx + x - l

x - y x - y at the same time, setting the billet length allocated to the first order as: . kx + x - l

y

x - y

6. A method for production scheduling as claimed in claim 1, characterized in that said step (c) comprises:

(cl) in the case that the total production yield

corresponding to said current combined production schedule is less than a critical threshold, calculating the evaluation value fo of said combined production schedule according to an evaluation function, and setting said stop value;

(c2) selecting randomly a sequence i from said combined production schedule, arranging a new pass therefor, and calculating the evaluation value f± of the production

schedule of the changed pass according to the evaluation function;

(c3) using the evaluation values fo and f± obtained above to calculate

if the values calculated are less than the stop value, then going to step (c4), and if the values calculated are greater than or equal to the stop value, then going to step (c5) ;

(c4) repeating steps (c2) to (c3) ; and (c5) outputting the current production schedule with the changed pass as said pass-changed production schedule.

7. A method for production scheduling as claimed in claim 6, characterized in that said evaluation function is expressed as follows: f = sequence_number + (1 - total_yield) , in which: sequence_number represents the number of sequences in the current production schedule, total_yield represents the total production yield

corresponding to the current production schedule.

8. A method for production scheduling as claimed in claim 7, characterized in that said stop value is 0.05.

9. A method for production scheduling as claimed in claim 1, characterized in that said step (d) comprises: (dl) inputting the pass-changed production schedule, dividing all of the sequences into a plurality of pass groups, with the sequences in each pass group being of the same product size and steel grade group;

(d2) selecting a pass group for which the treatment is not yet completed, and selecting a sequence from said pass group;

(d3) judging whether another order can be inserted into said sequence or an order in the sequence can be replaced by using said other order, if it is judged as yes, then

inserting said other order into said sequence or using said other order to replace an order in the sequence;

10. A system for performing the method as claimed in any one of claims 1-9, comprising: a module for generating an initial production schedule, used to distribute the customer orders into the corresponding sequences and the randomly selected passes on the basis of the same steel grades and production size codes inputted therein, so as to output the initial production schedule; a module for generating a combined production schedule, used to sort said sequences in said initial production schedule according to the delivery dates and combine any two of said orders with the same steel grade according to a combining algorithm, to obtain said sequences thus treated, so as to output the combined production schedule; a module for generating a pass-changed production schedule, used to select repeatedly and randomly one of the sequences from said combined production schedule and arrange a new pass therefor in the case that the total production yield

corresponding to said combined production schedule is less than a critical threshold, until the production schedule of the changed pass is judged by means of an evaluation function as being greater than or equal to a stop value, then

outputting the production schedule at this moment as the pass-changed production schedule; and a module for generating an order-changed production schedule, used to select a sequence to be treated from said pass- changed production schedule, to insert another order into said sequence or use said other order to replace an order in said sequence, so as to output the order-changed production schedule with the changed order, characterized in that, the output of the module for

generating said initial production schedule is connected to the input of the module for generating said combined

production schedule, the output of the module for generating said combined production schedule is connected to the input of the module for generating said pass-changed production schedule, and the output of the module for generating said pass-changed production schedule is connected to the input of the module for generating said order-changed production schedule .