A Multi-Constraint, Multi-Feature Collaborative Optimization Scheduling Method for Large-Scale Heat Treatment Production
By using a multi-constraint, multi-feature collaborative optimization scheduling method, and leveraging genetic algorithms and neighborhood search operators to optimize heat treatment scheduling, the problems of long equipment waiting time and unreasonable batch combinations in heat treatment production are solved. This results in an efficient and executable scheduling scheme, improving equipment utilization and delivery capabilities.
Patent Information
- Application Number
- CN202611127989.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-28
- Publication Date
- 2026-11-17
AI Technical Summary
Existing heat treatment scheduling methods are difficult to achieve efficient and executable scheduling under complex process and resource constraints. Especially in large-scale production and multi-constraint coupled scenarios, there are problems such as long equipment waiting time, unreasonable batch combination, and delivery delay. Moreover, existing algorithms are difficult to take into account multi-objective optimization.
A multi-constraint, multi-feature collaborative optimization scheduling method is adopted. A chromosome is constructed using a genetic algorithm. Combined with fitness evaluation and neighborhood search operators, a heat treatment scheduling scheme that satisfies multiple constraints is generated through chromosome decoding and iterative optimization. This includes batch idle forwarding and same-group continuous repair operators to optimize equipment processing sequence and process gaps.
It improved the overall quality of production scheduling results, reduced equipment waiting time, enhanced the feasibility and on-site adaptability of production scheduling results, improved equipment utilization and delivery capability, and reduced energy consumption.
Smart Images

Figure CN123117591A_ABST
Abstract
Description
Technical Field
[0001] This invention is a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production, belonging to the field of intelligent manufacturing and production scheduling optimization technology. Background Technology
[0002] In the steel production process, heat treatment is a crucial step that determines the microstructure, mechanical properties, and final product quality of steel. Heat treatment production typically involves multiple pieces of equipment, numerous processes, and complex technological requirements such as different temperature regimes, holding times, and cooling methods. Furthermore, different products exhibit significant differences in material, specifications, thickness, batch size, and delivery time. Therefore, rationally arranging the sequence of heat treatment operations, equipment allocation, and batch combinations while meeting process and production resource constraints is a critical issue for ensuring stable production operation, improving equipment utilization, and ensuring on-time delivery capabilities.
[0003] Traditional heat treatment scheduling methods largely rely on manual experience or simple rules for decision-making, such as prioritizing orders by delivery date, scheduling production based on similar steel grades or temperatures, or having schedulers manually assign equipment based on on-site experience. While these methods offer some flexibility, they have significant shortcomings overall: Firstly, heat treatment scenarios involve numerous complex and interdependent constraints, such as temperature regime matching constraints, process route sequence constraints, batch processing capacity constraints, thickness / specification continuous processing constraints, downtime maintenance calendar constraints, and delivery date constraints. Manual scheduling struggles to provide comprehensive coordination, easily leading to infeasible plans or frequent rework adjustments. Secondly, when there are many orders, many pieces of equipment, and frequent on-site disturbances, manual scheduling is inefficient, unable to respond promptly to actual production changes, and can easily result in increased equipment waiting time, unreasonable batch combinations, delivery delays, and even increased energy consumption.
[0004] As operations research algorithms are increasingly applied in manufacturing, heuristic algorithms, metaheuristic algorithms, and integer programming models have been gradually applied to heat treatment scheduling decisions. These methods can improve the automation level and solution quality of scheduling to a certain extent. However, existing technologies still have significant limitations. When using heuristic or rule-driven methods, the algorithms usually rely on manually designed local rules, which are easily affected by specific scenarios and have weak generalization and poor transferability. At the same time, in multi-objective and multi-constraint coupled scenarios, the algorithms are prone to getting trapped in local optima, making it difficult to simultaneously consider feasibility, equipment utilization, batch rationality, and delivery performance. When using integer programming or mixed-integer programming methods, if constraints such as batch combination, equipment allocation, job sequencing, downtime calendar, and continuous processing are modeled in a unified manner, the number of model variables and constraints will expand rapidly. Especially in large-scale order scenarios, the solution complexity increases significantly, and commercial solvers can hardly obtain high-quality executable solutions within an acceptable time, making it difficult to meet the business needs of on-site rolling scheduling and rapid rescheduling.
[0005] Therefore, this invention proposes a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production. This method can improve scheduling efficiency and scheme quality while satisfying complex process constraints and resource constraints, and enhance the executability and on-site adaptability of scheduling results. Summary of the Invention
[0006] To address the problems in existing technologies, this invention provides a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production.
[0007] The technical solution adopted by this invention to solve its technical problem is: a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production, comprising: Step S1: Obtain information on the processes to be scheduled, equipment information, and algorithm parameters; Step S2: Construct chromosomes and generate the initial population; Step S3: Decode the chromosomes in the initial population to determine the actual processing sequence of each process on each device and the start and end times of each process, and obtain the corresponding heat treatment production schedule. Step S4: Based on fitness evaluation and combined with preset strategy operators, use a genetic algorithm to iteratively search and continuously optimize the current heat treatment scheduling scheme; Step S5: When either the preset search time or the preset number of iterations is reached, stop the iteration and output the optimal heat treatment production schedule.
[0008] Furthermore, the information regarding the processes to be scheduled includes at least: Number of processes Total processing time for each process Single board processing time Delivery period Temperature point Distribution equipment Grouping identifiers And the process number of the process in its respective process path; The equipment information includes at least the heating rate and cooling rate of each device; The algorithm parameters include at least the maximum number of iterations, the maximum search time, the population size, and the parameters of the selected strategy operator.
[0009] Furthermore, in step S2, a process sequence coding method is used to construct the chromosome, with each coding bit corresponding to a process to be arranged; Several initial individuals are generated according to temperature sorting rules, delivery date sorting rules, and random generation rules, and then combined according to a preset ratio to form an initial population.
[0010] Furthermore, the heat treatment scheduling scheme satisfies constraints on process sequence, equipment availability time, downtime for maintenance, temperature connection, and continuous production. It also centrally arranges tasks at the same temperature point and tasks at similar temperature points by combining the temperature point information corresponding to the workpiece.
[0011] Furthermore, the decoding in step S3 includes: Based on the order of occurrence of processes in the chromosome, and combined with process path constraints, equipment pre-allocation constraints, equipment availability time constraints, downtime maintenance constraints, and temperature connection constraints, the processing sequence, start time, and completion time of each process on the corresponding equipment are determined step by step. For two consecutive processes on the same equipment, if the temperature points of the preceding and following processes are the same, the start time of the following process simultaneously satisfies the overlap constraint based on the single-board processing time of the preceding process and the time constraint based on the completion time of the preceding process. If the temperature points of the preceding and following processes are different, the start time of the following process shall not be earlier than the sum of the end time of the preceding process and the heating and cooling time of the equipment, wherein the heating and cooling time is determined based on the temperature difference between the preceding and following processes and one of the heating rate and cooling rate of the equipment. For processes with the same groupId, they are continuously merged according to their order of appearance on the corresponding device during the decoding process, so that they form a continuous production interval on the device and other processes are not allowed to be inserted into it.
[0012] Furthermore, in step S4, when evaluating the fitness of the current heat treatment production schedule, the fitness is determined by both on-time delivery performance and equipment idle ratio; the optimization objectives are to achieve the best on-time delivery performance and the lowest equipment idle ratio. The fitness value is used to select candidate parents by roulette wheel selection, and the offspring solution is generated by crossover to generate new candidate solutions; The candidate individuals are locally adjusted by combining the neighborhood search operator.
[0013] Furthermore, the neighborhood search operator includes a batch idle forward shift operator, which includes: Step S411: Based on the equipment sequence number of the process allocation, temperature point, start time of the statistical week to which the delivery period belongs, process path sequence number, and... The process is divided into different groups; and based on the current production schedule, the order in which the groups appear on each piece of equipment is extracted. The first occurrence position of the group on the corresponding device is taken as the order of the group on that device, thus obtaining the group-level processing order of each device; Step S412: For the process sequence on each machine, calculate the idle gap value between the start time of the process and the completion time of its preceding related process, and select processes with an idle gap value greater than zero as delayed candidate processes. Step S413: Using the idle gap value as a weight, perform a weighted selection of the candidate delayed processes to determine the target delayed process; Step S414: On the equipment where the target delayed process is located, randomly select a process from the process and its subsequent processes, and assign the process to... The corresponding group is identified as the group to be moved forward; Step S415: Insert the group to be moved forward into the group-level processing sequence of the corresponding equipment; Step S416: During the forward insertion process, determine whether the current insertion position will divide other continuous process groups. If it will divide, continue to adjust the insertion position forward until the insertion position is located at the boundary of the continuous process group. Step S417: If the selected process has an upstream preceding process on the process path, then the preceding process belongs to... The corresponding group moves forward synchronously in the group-level processing sequence of its respective equipment; Step S418: Reconstruct a new production schedule based on the adjusted processing sequence of each equipment group.
[0014] Furthermore, the neighborhood search operator also includes a same-group continuity repair operator, which includes: Step S421: Based on the process coding order in the current chromosome and the allocation equipment information corresponding to each process, construct the original process processing sequence on each equipment; Step S422: Traverse the original process sequence on each device and identify... Non-empty processes are designated as target processes for continuous repair within the same group; for For processes that are not completed, maintain their original order and do not participate in the repair process; Step S423: For each piece of equipment, collect the equipment with the same process sequence according to the original process order. The process forms The mapping relationship between the process set and the process set; the same The relative order of each process in the process set remains unchanged; Step S424: Iterate through the original process sequence on the equipment again. If the current process is not involved in the repair, output it in the original order; if the current process is involved in the repair and its... When it first appears, use that location as the anchor point to... All corresponding processes are output continuously in their original order; when the current process corresponds to When the output is complete, skip this step and do not repeat the output process. Step S425: Obtain the post-repair process sequence on each device, making them identical. The processes form a continuous group at their first occurrence position, while maintaining the relative order of non-repair processes and the original relative order of processes within the same group. Step S426: Based on the repaired process sequence of each equipment, reconstruct the process and equipment allocation relationship, and perform reverse decoding to obtain the repaired new production schedule solution.
[0015] Furthermore, the optimal heat treatment production schedule output in step S5 includes at least: the processing sequence of each process on each piece of equipment and the start time and completion time of each process.
[0016] The beneficial effects of this invention are: 1. This invention addresses the problems of long waiting time for processes, discrete distribution of processes in the same group, and non-compact processing sequence in heat treatment production scheduling. It proposes a production scheduling optimization method based on idle gap driven forward movement and continuous repair of the same group, which can effectively improve the overall quality of production scheduling results.
[0017] 2. This invention uses the idle gap as the basis for disturbance guidance, so that the adjustment process can be applied to key positions with long waiting time first. Combined with the upstream associated group linkage forward mechanism, it reduces unnecessary waiting time and enhances the coordination between equipment processing sequence and process path.
[0018] 3. This invention sets up a continuous group boundary protection mechanism during the group forward movement process, and restores the continuity of the same group to ensure that identical groups are protected. This allows for a more rational continuous processing structure, thereby improving the degree of continuous processing within the same group while maintaining the stability and controllability of the production scheduling structure.
[0019] 4. This invention comprehensively considers actual production factors such as multiple equipment, multiple processes, group continuity constraints, and delivery requirements in heat treatment production scheduling. It has strong engineering applicability and can still obtain a better and executable production scheduling scheme in large-scale, multi-constraint manufacturing scenarios. It has good practical value and promotion significance. Attached Figure Description
[0020] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to the present invention. Detailed Implementation
[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.
[0022] Please see Figure 1 This invention provides a multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production, comprising: Step S1: Obtain preset parameters such as the process to be scheduled, equipment information, and algorithm parameters; The information for the pending processes should at least include: the number of processes. Total processing time for each process Single board processing time Delivery period Temperature point Distribution equipment Grouping identifiers And the process number of the process in its respective process path; Equipment information should include at least each device heating rate and cooling rate ; Algorithm parameters: Genetic algorithm related parameters, such as maximum number of iterations, maximum search time, population size, and selected strategy operator parameters.
[0023] Step S2: Construct chromosomes and generate the initial population; Based on the information of the processes to be arranged obtained in step S1, a chromosome is constructed using a process sequence coding method, where each coding bit corresponds to a process to be arranged. The position of the coding bit in the chromosome indicates the priority scheduling order of this process in subsequent decoding; Several initial individuals are generated according to temperature sorting rules, delivery date sorting rules, and random generation rules, respectively, and then combined according to a preset ratio to form an initial population, which serves as the input for subsequent decoding and optimization.
[0024] Step S3: Decode the chromosomes in the initial population to determine the actual processing sequence of each process on each device and the start and end times of each process, and obtain the corresponding heat treatment production schedule. The heat treatment production scheduling scheme meets the constraints of process sequence, equipment availability time, downtime maintenance, temperature connection, and continuous production. It also combines the temperature point information of the workpiece to centrally arrange tasks at the same temperature point and tasks at similar temperature points, so as to improve production continuity, reduce changeover losses, and improve delivery performance.
[0025] Each chromosome in the initial population is decoded sequentially. The decoding includes: Based on the order of occurrence of processes in the chromosome, and combined with process path constraints, equipment pre-allocation constraints, equipment availability time constraints, downtime maintenance constraints, and temperature connection constraints, the processing sequence, start time, and completion time of each process on the corresponding equipment are determined step by step. For two consecutive processes on the same equipment, if the temperature points of the preceding and following processes are the same, the start time of the following process simultaneously satisfies the overlap constraint based on the single-board processing time of the preceding process and the time constraint based on the completion time of the preceding process. If the temperature points of the preceding and following processes are different, the start time of the following process shall not be earlier than the sum of the end time of the preceding process and the heating and cooling time of the equipment. The heating and cooling time is determined based on the temperature difference between the preceding and following processes and the heating or cooling rate of the equipment. For processes with the same groupId, they are continuously merged according to their order of appearance on the corresponding device during the decoding process, so that they form a continuous production interval on the device and other processes are not allowed to be inserted into it.
[0026] Step S4: Based on fitness evaluation and combined with preset strategy operators, use a genetic algorithm to iteratively search and continuously optimize the current heat treatment scheduling scheme; Based on the heat treatment production scheduling scheme obtained from step S3, when evaluating the fitness of the current heat treatment production scheduling scheme, the fitness is jointly determined by on-time delivery performance and equipment idle ratio; the optimization objectives are to achieve the best on-time delivery performance and the minimum equipment idle ratio. The fitness value is selected by roulette wheel selection, and the offspring solution is generated by crossover. The position or device is modified with a certain probability to perform mutation operation to generate new candidate solutions. Simultaneously, the neighborhood search operator is used to locally adjust candidate individuals to further improve the quality of the scheduling scheme. Newly generated individuals are further decoded and evaluated, and the population is iteratively updated to continuously retain and optimize better individuals, thereby gradually improving the overall solution performance.
[0027] Preferably, the neighborhood search operator includes a batch idle forward operator, which is used to advance the group-level processing sequence on the equipment according to the process waiting status. The batch idle forward operator includes: Step S411: Construct group-level processing sequence: Divide the processes into different groups based on the equipment number, temperature point, start time of the statistical week to which the delivery period belongs, process path number, and groupId; and extract the group appearance order on each equipment based on the current production scheduling results. The first appearance position of each groupId on the corresponding equipment is taken as the order of the group on that equipment, thus obtaining the group-level processing sequence of each equipment. Step S412: Identify delayed candidate processes: For the process sequence on each machine, calculate the idle gap value between the start time of the process and the completion time of its preceding related process, and filter processes with idle gap values greater than zero as delayed candidate processes. Step S413: Select the target delayed process: Using the idle gap value as a weight, the candidate delayed processes are selected by weighting to determine the target delayed process. The larger the idle gap value, the higher the probability of being selected. Step S414: Determine the group to be moved forward: On the equipment where the target delayed process is located, randomly select a process from the process and its subsequent processes, and determine the group corresponding to the groupId of the process as the group to be moved forward; Step S415: Perform forward insertion: Insert the group to be moved forward into the group-level processing sequence of the corresponding equipment. The insertion position is randomly generated before the original position of the group to achieve forward perturbation. Step S416: Perform group boundary protection: During the forward insertion process, determine whether the current insertion position will cut off other groups with the same boundary. If a continuous process group with a valid group mark is to be split, the insertion position will be adjusted forward until the insertion position is at the boundary of the continuous process group. If the insertion position divides the group into consecutive groups corresponding to the group to be moved forward, then the insertion is allowed.
[0028] Step S417, Link Upstream Related Group: If the selected process has an upstream preceding process on the process path, the group corresponding to the groupId of the preceding process is synchronously moved forward in the group-level processing sequence of the equipment. During the synchronous forward movement, group boundary protection is also performed to avoid splitting other continuous process groups.
[0029] Step S418: Reconstruct the production scheduling solution: Based on the adjusted processing sequence of each equipment group, restore the correspondence between the processes and equipment within the group, reconstruct the new production scheduling solution, and obtain the new solution after the disturbance.
[0030] The neighborhood search operator also includes a same-group continuity repair operator, which is used to group processes with the same sequence of operations without disrupting the relative order of non-grouped processes. The processes are stably aggregated on the corresponding equipment, and the continuous repair operators in the same group include the following steps: Step S421: Construct the equipment process sequence: Based on the process coding order in the current chromosome and the equipment information corresponding to each process, construct the original process processing sequence on each device; Step S422: Identify the processes involved in the repair: Traverse the original process processing sequence on each device, identify processes with non-empty groupId as target processes for continuous repair within the same group; for processes with empty groupId, maintain their original order and do not participate in the repair. Step S423: Collect the set of processes in the same group: For each machine, collect the processes with the same groupId according to the original process sequence to form a mapping relationship between groupId and process set; the relative order of each process under the same groupId in the process set remains unchanged; Step S424: Repair using the first occurrence position as the anchor point, and execute the following rules: The original process sequence on the equipment is traversed again. When the current process is not involved in the repair, it is output in the original order. When the current process is involved in the repair and its groupId appears for the first time, the position is used as the anchor point, and all processes corresponding to the groupId are output continuously in the original order. When the groupId corresponding to the current process has been completed, the process is skipped and will not be output again. Step S425: Obtain the repaired sequence: Through the above processing, the repaired process sequence on each device is obtained, so that processes with the same groupId form a continuous group at their first occurrence position, while keeping the relative order of non-repairing processes unchanged, and keeping the original relative order of processes within the same group unchanged. Step S426: Reverse Decoding and Reconstructing a New Solution: Based on the repaired sequence of each equipment process, reconstruct the process and equipment allocation relationship, and perform reverse decoding to obtain the repaired new production scheduling solution.
[0031] Step S5: When the preset search time or preset number of iterations is reached, stop the iteration and output the optimal heat treatment production schedule.
[0032] During the iterative search process of the genetic algorithm, the optimal individual in the current population and its corresponding production scheduling scheme are recorded in real time. When the preset maximum search time or maximum number of iterations is reached, the iterative calculation stops, and the heat treatment production scheduling scheme corresponding to the current optimal individual is output.
[0033] The optimal heat treatment production schedule output in this step includes at least: the processing sequence of each process on each piece of equipment, as well as the start time and completion time of each process.
[0034] Through the above steps, efficient scheduling optimization of heat treatment processes is achieved under the conditions of satisfying process path constraints, equipment constraints, downtime maintenance constraints, temperature connection constraints, and batch continuity constraints.
[0035] Preferably, the output heat treatment production schedule includes at least one, and also includes on-time delivery rate, equipment utilization rate, and total energy consumption. If all aspects of the output heat treatment production schedule are optimal, only one heat treatment production schedule can be output. Otherwise, heat treatment production schedules are output separately according to the optimal on-time delivery rate, equipment utilization rate, and total energy consumption. Based on the actual situation, such as emergency production, normal cost reduction, and equipment life extension, the staff selects the corresponding heat treatment production schedule. The total output plan includes a set of intelligent recommended execution plans and multiple sets of alternative non-inferior plans, which support manual scheduling and secondary selection, making it more practical.
[0036] The heat treatment scheduling problem is a multi-task, multi-constraint sorting optimization problem under the scenario of equipment pre-allocation, specifically: Given the known data of the workpieces to be processed, equipment resource data, process path data, and production constraint rules, and with the processing equipment for each workpiece to be processed already determined in advance, a comprehensive decision is made on the processing sequence, start time, and completion time of each workpiece to be processed, so as to form a heat treatment production scheduling plan covering all pending tasks. Production scheduling plans should not only meet constraints such as process sequence, equipment availability, downtime for maintenance, and process matching, but also incorporate temperature information for the workpieces, grouping tasks with similar or identical temperatures together to enhance production continuity and reduce changeover losses. Based on this, the production scheduling plan should be evaluated and optimized with delivery performance as the optimization objective.
[0037] Based on the characteristics of the problem, the heat treatment scheduling problem can be abstracted as a multi-task, multi-constraint sequencing optimization problem under a pre-allocation equipment scenario. In this problem, there are simultaneous time dependencies, process sequence relationships, and concentrated temperature point requirements among the workpieces to be processed. Given that the equipment allocation relationships are already determined, it is still necessary to further optimize the execution order and processing time of each task on its corresponding equipment. The resulting scheduling scheme should, while satisfying all constraints, reduce equipment idle time and on-time delivery rate to improve heat treatment scheduling.
[0038] It has There are 10 processes to be processed, denoted as the process set. ; has A heat treatment equipment set, denoted as [equipment name missing]. For any process Its processing equipment has been pre-allocated, denoted as ,in In addition, the known processes The processing time is The single-board processing time is Delivery period is Temperature point is For the same device Two consecutive processing steps and Let their corresponding temperature points be respectively and The start times are respectively and The end times are respectively and The single-board processing time is respectively and The total processing time is as follows: and .when = At the same time, to ensure the overlap between consecutive processes under continuous production conditions at the same temperature, the subsequent process... The start time should simultaneously satisfy the following two constraints: , ;when At that time, the next process The start time should not be earlier than the previous process. The sum of the end time and the equipment's heating / cooling duration, i.e. Among them, temperature switching time Defined as , Representing the equipment The rate of heating and the rate of cooling.
[0039] In addition, for some special processes that are subject to technological limitations, grouping identifiers are set in advance. In the processing sequence of the same equipment, those with the same The processes should form a continuous production zone, and other processes are not allowed to be inserted into it; at the same time, provided that existing process sequence constraints are not violated, the processing order of processes within the same group can be adjusted according to production scheduling needs. For those not set For certain processes, no continuous production constraints are imposed, and they are not included in batch processing; they can be arbitrarily sorted according to the overall production scheduling results.
[0040] Table 1: Comparison of the effects of this implementation case Please refer to Table 1. Eight test scenarios were designed. The on-time delivery rate, equipment utilization rate, and total duration of the entire plan for each scenario are shown in the table above. Example 1 shows that with manual scheduling, the on-time delivery rate is 88%, equipment utilization rate is 89%, and the total planned time is 3 units; with automatic scheduling, the on-time delivery rate is 92%, equipment utilization rate is 91%, and the total planned time is 1.2 units. The comparison shows that the delivery rate increased by 5%, the equipment utilization rate increased by 2%, and the total project duration was directly shortened by 60%. The invention has a significant effect on shortening the project duration.
[0041] In Example 7, the on-time delivery rate was 83%, equipment utilization rate was 84%, and the total planned time was 11 units when using manual scheduling; when using algorithmic automatic scheduling, the on-time delivery rate was 89%, equipment utilization rate was 89%, and the total planned time was 8 units. The comparison shows that the delivery rate increased by 7%, the equipment utilization rate increased by 6%, and the project duration was shortened by 27%. In this scenario, the present invention has a greater improvement in delivery and equipment utilization than manual scheduling in Example 1.
[0042] The average of all examples is as follows: the on-time delivery rate of the present invention is 90.25%, while that of manual labor is 86%, with the present invention showing an average improvement of 5 percentage points; the equipment utilization rate of the present invention is 90.5%, while that of manual labor is 86.875%, with the present invention showing an average improvement of 4 percentage points; the planned duration of the present invention is 3.5875, while that of manual labor is 5.7, with the present invention showing an average improvement and reduction of 41%.
[0043] A comparison of multiple sets of calculation examples shows that the present invention has a stronger ability to deliver orders on time; workshop equipment is utilized more fully, reducing idle time; the overall production planning cycle is significantly shortened, and the average lead time is reduced by more than 40%.
[0044] Although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. A multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production, characterized in that: include: S1. Obtain information on the processes to be scheduled, equipment information, and algorithm parameters; S2. Construct chromosomes and generate the initial population; S3. Decode the chromosomes in the initial population to determine the actual processing sequence of each process on each device and the start time and completion time of each process, and obtain the corresponding heat treatment production schedule. S4. Based on fitness evaluation and combined with preset strategy operators, use a genetic algorithm to iteratively search and continuously optimize the current heat treatment production scheduling scheme. S5. When either the preset search time or the preset number of iterations is reached, stop the iteration and output the optimal heat treatment production schedule.
2. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 1, characterized in that: The information on the pending processes includes at least: Number of processes N, total processing time of each process Single board processing time Delivery period Temperature point Distribution equipment Grouping identifiers And the process number of the process in its respective process path; The equipment information includes at least the heating rate and cooling rate of each device; The algorithm parameters include at least the maximum number of iterations, the maximum search time, the population size, and the parameters of the selected strategy operator.
3. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 1, characterized in that: In S2, chromosomes are constructed using a process sequence coding method, with each coding bit corresponding to a process to be arranged. Several initial individuals are generated according to temperature sorting rules, delivery date sorting rules, and random generation rules, and then combined according to a preset ratio to form an initial population.
4. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 1, characterized in that: The heat treatment production scheduling scheme satisfies constraints on process sequence, equipment availability time, downtime for maintenance, temperature connection, and continuous production. It also centrally arranges tasks at the same or similar temperature points based on the temperature point information of the workpiece.
5. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 1, characterized in that: The decoding in S3 includes: Based on the order of occurrence of processes in the chromosome, and combined with process path constraints, equipment pre-allocation constraints, equipment availability time constraints, downtime maintenance constraints, and temperature connection constraints, the processing sequence, start time, and completion time of each process on the corresponding equipment are determined step by step. For two consecutive processes on the same equipment, if the temperature points of the preceding and following processes are the same, the start time of the following process simultaneously satisfies the overlap constraint based on the single-board processing time of the preceding process and the time constraint based on the completion time of the preceding process. If the temperature points of the preceding and following processes are different, the start time of the following process shall not be earlier than the sum of the end time of the preceding process and the heating and cooling time of the equipment, wherein the heating and cooling time is determined based on the temperature difference between the preceding and following processes and one of the heating rate and cooling rate of the equipment. For processes with the same groupId, they are merged sequentially according to their order of appearance on the corresponding device during the decoding process, so that they form a continuous production interval on the device and other processes are not allowed to be inserted into it.
6. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 5, characterized in that: In S4, when evaluating the fitness of the current heat treatment production schedule, the fitness is determined by both on-time delivery performance and equipment idle ratio; the optimization objectives are to achieve the best on-time delivery performance and the minimum equipment idle ratio. The fitness value is used to select candidate parents by roulette wheel selection, and the offspring solution is generated by crossover to generate new candidate solutions; The candidate individuals are locally adjusted by combining the neighborhood search operator.
7. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 6, characterized in that: The neighborhood search operator includes a batch idle forward shift operator, which includes: S411, based on the equipment sequence number of the process allocation, temperature point, start time of the statistical week to which the delivery period belongs, process path sequence number, and The process is divided into different groups; and based on the current production schedule, the order in which the groups appear on each piece of equipment is extracted. The first occurrence position of the group on the corresponding device is taken as the order of the group on that device, thus obtaining the group-level processing order of each device; S412. For the process sequence on each machine, calculate the idle gap value between the start time of the process and the completion time of its preceding related process, and select processes with an idle gap value greater than zero as delayed candidate processes. S413. Using the idle gap value as a weight, the candidate delayed processes are weighted and selected to determine the target delayed process. S414. On the equipment where the target delayed process is located, randomly select one process from the process and its subsequent processes, and assign the process to... The corresponding group is identified as the group to be moved forward; S415. Insert the group to be moved forward into the group-level processing sequence of the corresponding equipment; S416. During the forward insertion process, determine whether the current insertion position will divide other continuous process groups. If it will divide, continue to adjust the insertion position forward until the insertion position is located at the boundary of the continuous process group. S417. If the selected process has an upstream preceding process on the process path, then the preceding process belongs to... The corresponding group moves forward synchronously in the group-level processing sequence of its respective equipment; S418. Based on the adjusted processing sequence of each equipment group, rebuild a new production schedule solution.
8. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 6, characterized in that: The neighborhood search operator further includes a same-group continuity repair operator, which includes: S421. Based on the process coding order in the current chromosome and the allocation equipment information corresponding to each process, construct the original process processing sequence on each equipment; S422. Traverse the original process sequence on each device and identify... Non-empty processes are designated as target processes for continuous repair within the same group; for For processes that are not completed, maintain their original order and do not participate in the repair process; S423. For each piece of equipment, collect data with the same process sequence according to the original workflow. The process forms The mapping relationship between the process set and the process set; the same The relative order of each process in the process set remains unchanged; S424. Iterate through the original process sequence on the equipment again. If the current process is not involved in the repair, output it in the original order; if the current process is involved in the repair and its... When it first appears, use that location as the anchor point to... All corresponding processes are output continuously in their original order; when the current process corresponds to When the output is complete, skip this step and do not repeat the output process. S425. Obtain the post-repair process sequence on each device, making them identical. The processes form a continuous group at their first occurrence position, while maintaining the relative order of non-repair processes and the original relative order of processes within the same group. S426. Based on the repaired process sequence of each equipment, reconstruct the process and equipment allocation relationship, and perform reverse decoding to obtain the repaired new production schedule solution.
9. The multi-constraint, multi-feature collaborative optimization scheduling method for large-scale heat treatment production according to claim 1, characterized in that: The optimal heat treatment production schedule output by S5 includes at least the following: the processing sequence of each process on each piece of equipment, as well as the start time and completion time of each process.
