Production takt-based distribution estimation algorithm for solving steelmaking continuous casting scheduling problem

Abstract

The invention provides a production takt-based distribution estimation algorithm for solving a steelmaking continuous casting scheduling problem, and belongs to the field of hybrid flow shop scheduling problems. The improved algorithm is used for solving the actual manufacturing problem and mainly comprises the following steps that in the initial stage, the rhythm of flow production is adjusted by establishing a probability model between a continuous casting machine and the production takt and adjusting the production takt of the continuous casting machine, and probability model sampling is conducted according to the adjusted rhythm to produce individuals; and local search and probability model updating are carried out on the generated excellent individuals, and an objective function value is reduced fastest within the shortest time. According to the method, the behavior that the target function value is increased due to the fact that workpieces of the same casting time are spaced too long in steelmaking continuous casting production scheduling is solved, the machining process is adjusted in a targeted mode, and the problem that energy consumption is too much in the actual process is optimized.

Description

Technical Field

[0001] The invention relates to a distribution estimation algorithm based on production beat for solving steelmaking continuous casting scheduling problems, and belongs to the field of mixed flow shop scheduling problems. Background Art

[0002] In modern manufacturing, production efficiency and resource optimization are the key to enterprise competitiveness. As a complex scheduling problem, the hybrid flow shop scheduling problem, especially for the special hybrid flow shop scheduling problem of steelmaking continuous casting production scheduling, involves the arrangement of multiple production equipment and processes. The goal is to minimize the production cycle, reduce costs, and improve resource utilization under the premise of meeting production constraints. Traditional scheduling methods often cannot adapt to the changing production environment, resulting in uneven production rhythm, which in turn affects the overall efficiency. By analyzing historical data and production characteristics, the processing time and machine availability of each process can be more accurately predicted. By describing the production rhythm of the continuous casting machine, the method can optimize the scheduling strategy and improve the response speed of the production line. Therefore, a distribution estimation algorithm based on the production rhythm for solving the steelmaking continuous casting scheduling problem is particularly important. The scheduling method proposed in the present invention can effectively integrate various information, avoid the workpieces of the same casting from waiting too long, thereby improving the overall production efficiency, reducing the delivery time, and achieving the optimal allocation of resources. At the same time, this method is different from the previous method of only considering the production of workpieces in one day, but considering the workload of one week for scheduling. Large-scale problem research is of great significance to such problems.

[0003] The main research methods for this type of problem include mathematical optimization methods, metaheuristic algorithms, heuristic methods, etc. For mathematical optimization methods, the efficiency of this method is relatively low when considering the actual constraints and the size of the problem. It is difficult for heuristic methods to solve the problem from a global perspective, and the quality of the obtained scheduling solutions is not high. Metaheuristic algorithms are prone to premature phenomena, and the quality of the initial population has a greater impact on the final solution. Therefore, better results can be achieved by searching and improving the initial probability model.

[0004] With the rise of intelligent manufacturing, the distribution estimation algorithm, as a new evolutionary algorithm in the meta-heuristic method, samples through the search space and predicts the optimal area through statistical learning. However, for the special production scheduling problem of steelmaking continuous casting, random sampling may cause the workpieces in the same casting to be too far apart, which consumes a lot of resources waiting. Therefore, in this context, the distribution estimation algorithm based on the production cycle for solving the steelmaking continuous casting scheduling problem shows superiority. By adjusting the probability of the interval between the same continuous casting machines, it shows excellent results, reduces energy consumption, and quickly generates high-quality scheduling plans. Summary of the invention

[0005] The present invention provides a distribution estimation algorithm based on the production rhythm for solving the steelmaking continuous casting scheduling problem. The algorithm is applied in the field of mixed flow workshop scheduling problems. The rhythm of flow production is adjusted by adjusting the production rhythm of the continuous casting machine. Probabilistic model sampling is performed according to the adjusted rhythm to produce individuals. The excellent individuals generated are used to guide the search direction of the algorithm through local search and updated probability models to make the search faster and reduce the objective function value as quickly as possible in the shortest time. The present invention solves the problem of other workpieces in the same pouring being waiting due to the workpieces being too far apart in the same pouring in the steelmaking continuous casting production scheduling. It specifically prevents the workpieces from being too far apart, shortens the optimization time, and optimizes the problem of excessive energy consumption in the actual process. It is implemented on a computer according to the following steps:

[0006] : Build the production information of the workpieces to be scheduled and build the task scheduling architecture. Different from the previous method of only considering the production of workpieces for one day, the scheduling is based on a week's workload. The research on large-scale problems is of great significance to such problems.

[0007] : Initialization, set the following basic variables;

[0008] Set the problem variables:

[0009] :Every workpiece that needs to be processed, ;

[0010] : The number of castings to which the workpiece belongs, ;

[0011] : Number of processing machines on the production line, ;

[0012] : Number of production stages, , where the first stage is the steelmaking stage, the second stage is the refining stage, and the third stage is the continuous casting stage;

[0013] : Workpiece In stage Processing time in

[0014] : Setting time between pouring times;

[0015] : Workpiece In stage Start processing time;

[0016] Set algorithm related parameters:

[0017] : learning rate;

[0018] : Maximum number of intervals;

[0019] : Number of outstanding individuals;

[0020] : Initialize the probability model to generate individuals;

[0021] :A probability model describing the production cycle is established according to the characteristics of the problem. Due to the uniqueness of the problem, the probability matrix of the model is based on the number of continuous casting machines in the continuous casting stage and the number of workpieces processed by other continuous casting machines between two adjacent workpieces processed by the same continuous casting machine. Individuals are generated by sampling the probability matrix;

[0022] : If there is no workpiece at the sampled position, place the workpiece directly at this position;

[0023] :When the sampled position has been placed with other artifacts, compare the probability of sampling this position with the probability of sampling this position before;

[0024] : If the new probability is greater than the old probability, place the new artifact here and move the old artifact to the nearest free position;

[0025] : Set the mark position of the moved workpiece. To maintain stability, each workpiece is moved only once;

[0026] : If the workpiece being placed has been moved, then place the new workpiece directly in the nearest free position;

[0027] : Place all workpieces according to the above method until all workpieces are placed;

[0028] :Individual coding, according to the characteristics of the problem, a coding method between continuous casting machines is proposed to adjust the production rhythm;

[0029] : Select some excellent individuals for local search;

[0030] :Update the probability model through excellent individuals. Update the probability model according to the excellent individuals after the local search, and update the sampling function of each continuous casting machine separately;

[0031] : Iterative process, after each generation of resampling, the newly sampled individuals and the individuals of the previous generation are sorted by the objective function value, and the number of individuals with the smallest objective function value is retained to the next generation. Local search is continuously performed and the probability model is updated until the termination condition is met;

[0032] : If the algorithm reaches the termination condition, the optimal individual is output as the final solution.

[0033] According to the above-mentioned distribution estimation algorithm based on production cycle for solving the steelmaking continuous casting scheduling problem, the present invention has carried out a large number of simulation experiments. From the results, it can be seen that for the same scheduling problem, the proposed algorithm can obtain better results and significantly reduce the objective function value. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Schematic diagram of the hardware system structure of the distribution estimation algorithm based on production cycle to solve the steelmaking continuous casting scheduling problem.

[0035] Figure 2 Algorithm flow chart of the distribution estimation algorithm based on production cycle for solving the steelmaking continuous casting scheduling problem.

[0036] Figure 3 Convergence diagram of the results of 100 iterations of Example 2 based on the distribution estimation algorithm of production cycle to solve the steelmaking continuous casting scheduling problem.

[0037] Figure 4 Gantt chart of the calculation results of Example 1 based on the distribution estimation algorithm of production cycle to solve the steelmaking continuous casting scheduling problem. DETAILED DESCRIPTION

[0038] The scheduling method of the present invention relies on the relevant data acquisition system. In the actual workshop scheduling, the software and hardware architecture diagram of the present invention is as follows: Figure 1 As shown, the following are specific embodiments of the present invention:

[0039] : Construct the production information of the workpieces to be scheduled, build the task scheduling framework, and form an instance of the scheduling problem to be solved. Different from the previous consideration of only one day's production of workpieces, the scheduling is based on a week's workload. Large-scale problem research is of great significance to such problems;

[0040] : Set algorithm related parameters;

[0041] : learning rate;

[0042] : Maximum number of intervals;

[0043] : Number of outstanding individuals;

[0044] : Initialize the probability model to generate individuals and set a 4-line Column probability matrix, each position represents the number of other workpieces between two adjacent workpieces processed on the same continuous casting machine;

[0045] :Individual coding, according to the characteristics of the problem, a coding method between continuous casting machines is proposed to adjust the production rhythm;

[0046] : Select some excellent individuals for local search;

[0047] :Update the probability model according to the dominant individuals, update the probability model according to the excellent individuals after the local search, and update the sampling function of each continuous casting machine separately;

[0048] : Iterative process, after each generation of resampling, the newly sampled individuals and the individuals of the previous generation are sorted by the objective function value, and the number of individuals with the smallest objective function value is retained to the next generation. Local search is continuously performed and the probability model is updated until the termination condition is met;

[0049] : If the algorithm reaches the termination condition, the optimal individual is output as the final solution;

[0050] The learning rate is 0.5;

[0051] The maximum number of intervals is 7;

[0052] The number of excellent individuals is 10;

[0053] Through simulation experiments, the improved algorithm of the present invention ( ) to solve the steel production scheduling problem and verify it with the distribution estimation algorithm ( )、Fruit Fly Algorithm( ) for comparison, ten examples 1-10 were run, and each example was run 10 times to compare the average value.

[0054] Table 1 , , Results comparison: IEDA EDA FOA 1 86547 104155 110032 2 94344 103678 103792 3 98110 124800 113548 4 92398 114448 103791 5 84412 122158 132577 6 74220 102477 113247 7 85138 113820 104813 8 101749 155328 132795 9 89794 103542 93251 10 82705 123544 89722

[0055] By analyzing Table 1 and Figure 3 We can see that in the example, the method proposed in the present invention can achieve better standards in terms of convergence speed and the degree of excellent solutions. This is because the improved algorithm has strong local and global search capabilities.

[0056] The above is a detailed description of the specific implementation of the present invention in conjunction with the accompanying drawings, but the present invention is not limited to the above implementations. Any modification or equivalent replacement of the technical solution of the present invention falls within the protection scope of the present invention. The technical parts not described in detail in the present invention are all well-known technologies.

Claims

1. Solving the steelmaking continuous casting scheduling problem The distribution estimation algorithm based on the production cycle is characterized by: A scheduling method for solving the steelmaking continuous casting production scheduling problem based on a distribution estimation algorithm, and the method is implemented on a computer in accordance with the following steps in sequence: : Build the production information of the workpieces to be scheduled and build the task scheduling architecture. Different from the previous method of only considering the workpieces produced in one day, the scheduling is based on the workload of one week. The research on large-scale problems is of great significance to such problems. : Initialization, set the following basic variables; Set the problem variables: :Every workpiece that needs to be processed, ; : The number of castings to which the workpiece belongs, ; : Number of processing machines on the production line, ; : Number of production stages, , where the first stage is the steelmaking stage, the second stage is the refining stage, and the third stage is the continuous casting stage; : Workpiece In stage Processing time in : Setting time between pouring times; : Workpiece In stage Start processing time; Set algorithm related parameters: : learning rate; : Maximum number of intervals; : Number of outstanding individuals; :Initialize the probability model to generate individuals. According to the characteristics of the problem, a probability model describing the production cycle is established. Due to the uniqueness of the problem, the probability matrix of the model is based on the number of continuous casting machines in the continuous casting stage and the number of other workpieces between two adjacent workpieces processed by the same continuous casting machine. Individuals are generated by sampling the probability matrix; :Individual coding, according to the characteristics of the problem, a coding method between continuous casting machines is proposed to adjust the production rhythm; : Select some excellent individuals for local search; :Update the probability model according to the dominant individuals, update the probability model according to the excellent individuals after the local search, and update each continuous casting machine separately; : Iterative process, in each generation, some excellent individuals are retained to the next generation, and then the remaining individuals are generated by sampling through the updated probability model, and all individuals are merged into a new generation population to continue iterating until the termination condition is met; : If the algorithm reaches the termination condition, the optimal individual is output as the final solution.

2. The method according to claim 1, characterized in that: In the process of generating individual samples, the corresponding production rhythm is adjusted by adjusting the individual sampling intervals between the same continuous casting machines to find the most suitable individual.

3. The method according to claim 1, characterized in that: In the individual coding process, the coding method is effectively simplified by generating codes between continuous casting machines.

4. The method according to claim 1, characterized in that: In order to better reflect the characteristics of this problem, large-scale problem optimization is adopted, that is, modeling is carried out through one week's workload.

Images