A multi-variety variable batch intelligent scheduling and dispatching method, system, medium and device

Abstract

This invention relates to the field of high-end electronic manufacturing and discloses a multi-variety, variable-batch intelligent production scheduling method, system, medium, and equipment. It includes: generating an initial production scheduling scheme that satisfies process logic and static resource constraints based on a resource-constrained project scheduling model, obtaining an initial resource occupancy curve; performing initial production scheduling on the initial resource occupancy curve using a subspace optimization-based genetic production scheduling algorithm; determining overdue projects and bottleneck resources based on the average resource utilization rate and occupancy curve; dynamically expanding capacity for expandable resources other than site availability, updating the resource availability curve, and inputting it into the genetic production scheduling algorithm for iterative optimization until all projects are on time or the maximum number of iterations is reached. The optimization objective is to minimize the weighted sum of the time differences for all projects, ensuring on-time delivery. This invention significantly improves the feasibility, robustness, and delivery accuracy of the production scheduling scheme.

Description

Technical Field

[0001] This invention relates to the field of high-end electronic manufacturing technology, and in particular to a method, system, medium and equipment for intelligent scheduling of multi-variety variable batch production. Background Technology

[0002] In the field of high-end electronics manufacturing, product production is characterized by high variety, small batch size, variable batch size, and high complexity. Planning requires coordinating hundreds of parallel projects, each containing hundreds of highly dependent processes, consuming a large amount of scarce resources such as specialized equipment, highly skilled personnel, and critical sites. However, current scheduling often employs a manual, block-based approach: each scheduler schedules independently, lacking a global resource view, leading to frequent resource conflicts after plan aggregation, resulting in low efficiency and delivery delays. This problem essentially falls under the Resource-Constrained Project Scheduling Problem (RCPSP), which aims to optimize project duration or cost while satisfying process sequence relationships and limited resource constraints. This problem has been proven to be a classic NP-hard problem. As the complexity of manufacturing systems increases, research has gradually expanded from single-project to multi-project parallel scheduling, further giving rise to high-level collaborative scheduling paradigms at the enterprise planning level. In recent years, given the frequent disturbances in actual production, dynamic scheduling has become a research hotspot. The Vieira system has proposed a rescheduling strategy framework that includes complete rescheduling, local repair, and rolling optimization. Meanwhile, many scholars have applied digital twins to dynamic scheduling, and the system has explored the real-time response capability to disturbances in a "physical-information" fusion environment, providing an important technical path for building a closed-loop, adaptive scheduling system.

[0003] The main methods for solving RCPSP problems can be categorized into three main types: exact methods, heuristic algorithms, and metaheuristic algorithms. Early methods relied primarily on exact approaches such as branch and bound and integer programming. However, when faced with problems involving numerous steps and complex procedures, their effectiveness drops drastically, and their cost becomes prohibitively high. Therefore, heuristic algorithms have gradually become the mainstream approach for solving RCPSP problems. These are solution strategies based on experience or intuitive rules, aiming to obtain feasible solutions within a reasonable computation time. However, heuristic algorithms only guarantee feasible solutions and do not guarantee solution quality. To pursue high-quality solutions, metaheuristic algorithms have emerged. Common metaheuristic algorithms include genetic algorithms, particle swarm optimization, and simulated annealing.

[0004] Among numerous metaheuristic algorithms, genetic algorithms are the mainstream of scholarly research. In solving genetic algorithms, existing technologies include improved meme algorithms, which significantly enhance local search performance by introducing a similar block sequential crossover operator and a variable insertion neighborhood search strategy; adaptive genetic algorithms with S / P and F / B dual-gene encoding, which outperform existing activity list-based GA methods on the RCPSP standard test set by dynamically selecting decoding strategies and scheduling directions; existing literature proposes a novel and concise RCPSP solution algorithm by integrating evolutionary strategies into the genetic algorithm framework; existing nested ant colony-genetic hybrid algorithms solve the dual-resource-constrained production scheduling problem for aerospace structural components under three different resource dimensions; and addressing the problem that most models only focus on minimizing the construction period as a single objective and fail to reflect actual resource usage needs, a multi-objective model that balances construction period and resource equilibrium is constructed, proposing the NSGA-IIs algorithm, which effectively reduces infeasible solutions and improves solution efficiency through subsetation operations.

[0005] However, existing research still has significant limitations: On the one hand, most genetic algorithm (GA) methods focus on static RCPSP models, which not only assume fixed resource capacity and do not support batch splitting, but also typically pursue the optimization objective of minimizing the maximum makespan, making it difficult to reflect the multi-project collaborative scheduling needs centered on contract delivery time in high-end electronics manufacturing; on the other hand, research generally relies on standard datasets such as PSPLIB for validation, lacking real-world applications at high-complexity planning levels (hundreds of projects, thousands of highly dependent processes, and multi-dimensional coupled resources of equipment, manpower, and site). More importantly, traditional methods usually only generate one-time static scheduling schemes, lacking a closed-loop response mechanism for the risk of overdue deadlines, and cannot adapt to the challenges of dynamic resource constraints and rigid delivery in actual production. Summary of the Invention

[0006] To address the aforementioned issues, the present invention aims to provide a multi-variety, variable-batch intelligent production scheduling method, system, medium, and equipment, which can significantly improve the feasibility, robustness, and delivery accuracy of the production scheduling scheme.

[0007] To achieve the above objectives, in a first aspect, the technical solution adopted by the present invention is as follows: a multi-variety variable batch intelligent production scheduling method, comprising: generating an initial production scheduling scheme that satisfies process logic and static resource constraints based on a resource-constrained project scheduling model, and obtaining an initial resource occupancy curve; performing initial production scheduling on the initial resource occupancy curve using a genetic production scheduling algorithm based on subspace optimization; determining overdue status for the scheduling results, and identifying overdue projects and bottleneck resources through the average resource utilization rate and occupancy curve; dynamically expanding capacity for expandable resources other than site capacity based on the actual capabilities of the enterprise, updating the resource availability curve, and inputting it into the genetic production scheduling algorithm for iterative optimization until all projects are not overdue or the maximum number of iterations is reached, with the optimization objective being to minimize the weighted difference value of all projects to ensure on-time delivery.

[0008] Furthermore, a genetic scheduling algorithm based on subspace optimization is used to perform initial production scheduling on the initial resource occupancy curve. This includes: using subspaces as the basic optimization unit, with each subspace corresponding to one or more packaged model project production plans; each individual in the population representing a complete scheduling scheme; and using a two-segment real-number encoding of process sequence + periodic coefficient to ensure solution feasibility. The two-segment real-number encoding consists of: the first half being a randomized arrangement of the global indices of the processes, representing the scheduling execution order, which must satisfy the condition that all preceding processes of any process z must be completed before it; the encoding process embeds a priority verification and adjustment function, which performs topological correction on the input sequence based on a pre-constructed preceding-follower relationship table; and the second half being the periodic coefficient of process z. The normalized periodicity coefficient is used to dynamically determine the process. Actual execution cycle After the coding is completed, the start and end times are set sequentially according to the order of the processes in the coding. Then, the fitness of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding. The coded individual is transformed into a specific time arrangement that meets the dynamic resource constraints. The fitness of the plan is evaluated. Under the premise of meeting the process logic and dynamic resource constraints, the feasible production plan that minimizes the time difference of all projects is generated through iterative evolution.

[0009] Furthermore, the adaptability of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding, including: The input consists of complete individual codes and subspace information. The codes are split into priority sequences and periodic sequences. It is determined whether the sequences are inverted. If they are inverted, the resource occupancy curve is inverted, the immediate predecessor and immediate successor lists are swapped, and the time complements of the front and back walls are swapped. If there is no inverted index, then traverse the priority sequence and process each step in the order of the priority sequence. Calculate the actual cycle time and daily resource consumption of the current process z, and determine the earliest start time of process z. It is determined by the end time of all its immediate preceding processes; Call the resource limit validation and time delay function to validate the time range. Does the resource consumption exceed the instantaneous capacity limit of equipment, manpower, and site? If resource conflicts exist, the earliest start time will be used. The work is postponed until the first feasible window that meets the resource constraints is found, which is recorded as the actual start time. If there are no resource conflicts, the earliest start time will be used. Record as the actual start time ; Based on actual start time Calculate completion time : And update the resource usage curve; After all the processes are arranged, calculate the fitness P of the solution. The smaller the fitness P, the better the solution.

[0010] Furthermore, the production scheduling results are assessed for overdue status, and overdue projects and bottleneck resources are identified through average resource utilization and occupancy curves, including: After the initial production scheduling plan is generated, resources are initially screened based on average utilization rate, and a first threshold is set. The average utilization rate of the markers exceeds Resources; By scanning the resource utilization curve, all continuous time periods with utilization rates consistently equal to or close to 100% are identified as high-load operation periods for the resources. The number of days of high-load operation is then calculated, and a second threshold is set. The number of days marked as high-load operation is higher than Resources; Filter out those with an average utilization rate higher than the first threshold And the number of days operating at high load exceeds the second threshold. The resource is marked as a bottleneck resource.

[0011] Furthermore, for expandable resources other than the site, dynamic capacity expansion is implemented based on the company's actual capabilities, updating the resource availability curve and inputting it into the genetic scheduling algorithm for iterative optimization until all projects are on time or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​for all projects, ensuring on-time delivery, including: Based on the identified bottleneck resources and combined with the company's actual dynamic adjustment principles, an initial resource occupancy curve that can be directly used by the genetic algorithm is generated. For different resource types, the established dynamic expansion principles are as follows: human resources can be expanded by adding overtime or additional personnel to increase the working hours limit for a set time period; equipment resources can be expanded by changing the shift system; and site resources cannot be expanded. Based on the principle of dynamic expansion, the following adjustment suggestions are given for different types of bottleneck resources: Filter out non-scalable resources. If the bottleneck resource is a site type, skip it directly without making any adjustments. For expandable human or equipment resources, adjustment suggestions are generated based on the length and urgency of their high-load operation periods; Calculate the daily increase in resource capacity within the adjustment window and quantify the adjustment amount; A new resource occupancy curve is constructed, and the incremental information of all bottleneck resources is superimposed on the original static resource availability curve as input for the next round of genetic algorithm. The optimization objective of the algorithm is still to minimize the difference value of all items.

[0012] Furthermore, after each iteration, an overdue analysis will be performed on the current production schedule, and a decision will be made as to whether to terminate the iteration. The iteration termination condition can be any of the following: (a) There are no overdue projects; (b) If the set of overdue projects and the total number of overdue days do not change after three consecutive rescheduling iterations, it is determined that the remaining overdue projects are not caused by resource bottlenecks, but by unreasonable production plans. This situation cannot be resolved by increasing resources.

[0013] Furthermore, after completing all rescheduling iterations, the latest resource usage curve will be analyzed again, and the specific implementation method will be adjusted accordingly: Resource It was given an increased capacity during a certain adjustment. The increment was originally planned to be applied to the time period. This interval is divided into a set of discrete time points. Each of them This corresponds to a scheduling time step; set up This indicates that in the final production schedule, resources... At any moment The actual amount occupied; At any moment Determine whether the increased capacity is being used. If not, adjust the available resource capacity at that moment. Back to Conversely, if the increase in capacity is not used, it indicates that the increased capacity has been utilized, thus yielding the final optimized resource availability curve. .

[0014] Secondly, the technical solution adopted by the present invention is: a multi-variety variable batch intelligent production scheduling system, which includes: The data preprocessing module generates an initial production schedule that satisfies both process logic and static resource constraints based on the resource-constrained project scheduling model, and obtains the initial resource occupancy curve. The initial production scheduling module uses a subspace-optimized genetic production scheduling algorithm to perform initial production scheduling on the initial resource occupancy curve; The bottleneck identification, dynamic capacity expansion, and rescheduling modules determine overdue production results and identify overdue projects and bottleneck resources through average resource utilization and occupancy curves. For expandable resources other than the site, dynamic capacity expansion is carried out based on the company's actual capabilities, updating the resource availability curve and inputting it into the genetic production scheduling algorithm for iterative optimization until all projects are not overdue or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​of all projects to ensure on-time delivery.

[0015] Thirdly, the technical solution adopted by the present invention is: a computer-readable storage medium for storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any of the methods described above.

[0016] Fourthly, the technical solution adopted by the present invention is: a computing device comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0017] The present invention has the following advantages due to the adoption of the above technical solutions: This invention aims to minimize the weighted difference period and constructs a closed-loop framework of "initial production scheduling - bottleneck identification - dynamic capacity expansion - rescheduling". Through a two-stage coded genetic algorithm and subspace optimization strategy, while explicitly modeling actual constraints such as batch splitting, batch interval and whole batch matching, it can accurately identify and limit the expansion of equipment and manpower bottlenecks, effectively eliminate resource conflicts and overdue delivery, and significantly improve the feasibility, robustness and delivery accuracy of the production scheduling scheme. Attached Figure Description

[0018] Figure 1 This is a flowchart of the intelligent scheduling method for multi-variety variable batch production in an embodiment of the present invention; Figure 2 This is included in the embodiments of the present invention. A diagram illustrating the individual coding of each process step; Figure 3 This is a flowchart of the priority verification and adjustment function in an embodiment of the present invention; Figure 4 This is an example diagram of encoding function verification in an embodiment of the present invention; Figure 5 This is a flowchart of fitness calculation (decoding function) in an embodiment of the present invention; Figure 6 These are two schematic diagrams of intersections in an embodiment of the present invention; Figure 7 These are schematic diagrams illustrating two variations in embodiments of the present invention. Detailed Implementation

[0019] Addressing the limitations of existing technologies, this invention provides a multi-variety, variable-batch, and highly complex intelligent production scheduling method, system, medium, and equipment for high-end electronic manufacturing planning scenarios. This method supports intelligent scheduling that coordinates batch production logic with delivery time optimization. Based on a genetic algorithm, a closed-loop framework of "initial scheduling—bottleneck identification—dynamic capacity expansion—rescheduling" is constructed. For the first time, the goal is to minimize the weighted sum of time differences for all projects, aligning with delivery-oriented management needs. Explicit modeling of batch flow synchronous scheduling logic, through process batching, batch intervals, and whole-batch matching mechanisms, accurately reflects the actual production process. A two-stage encoding method combining process sequence and cycle coefficients is used to generate feasible solutions, and subspace optimization is introduced to mitigate premature convergence in large-scale problems. Simultaneously, overdue projects and bottleneck resources are automatically identified, and limited dynamic capacity expansion of expandable resources is implemented based on enterprise capabilities. Iterative rescheduling ensures on-time delivery. Verification by an electronics manufacturing company shows that this invention can effectively eliminate resource conflicts and delivery delays, and significantly improve the feasibility, robustness, and delivery accuracy of production scheduling schemes.

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0022] To describe the aforementioned production scheduling problem, a mathematical model is constructed from four dimensions: scheduling object, time dimension, resource type, and process relationship. The symbolic meanings of some parameters involved in this invention are as follows: Represents a collection of items; This indicates a project index, representing a specific production batch. Indicates project A set of processes; This represents the process index, indicating the sequence number of the process. Indicates project The One process; Indicates project The number of units put into production; Indicates project Priority; Indicates project The earliest start time, Indicates project The latest delivery time; Indicate process Number of pieces to be processed; Indicate process The set of preceding processes; Indicate process The number of pieces produced in a single batch; Indicate process The standard cycle time for a single piece (batch) Indicate process Expedited turnaround time for single items (batch); Indicate process Multiple batch intervals; Indicate process The core equipment occupied; Indicate process Equipment occupancy time, unit: unit Hour; Indicate process Required personnel; Indicate process Single-piece man-hours, unit: person Hour / piece; Indicate process The core area occupied; Indicates the amount of space occupied; Indicate process The actual production cycle; Indicate process start time, Indicate process The completion time.

[0023] In one embodiment of the present invention, a multi-variety, variable-batch intelligent production scheduling method is provided. This invention aims to minimize the time lag, integrates a dynamic resource adjustment mechanism, constructs a closed-loop scheduling framework, and verifies its effectiveness using real enterprise data. This addresses the issues of dynamically changing resource constraints and rigid delivery deadlines in multi-project parallel manufacturing scenarios. For the annual production planning scenario of electronic manufacturing enterprises, it is necessary to coordinate the parallel production of multiple multi-variety, variable-batch projects under a unified scheduling framework. Each project consists of a series of sequential processes, some of which consume internal equipment, manpower, and site resources. The scheduling objective is to minimize the sum of weighted time lag values ​​for all projects while satisfying the following three types of constraints and closely aligning with the enterprise's batch production logic, ensuring that no project exceeds its contract delivery deadline: process logic constraints, dynamic resource capacity constraints, and project time window constraints.

[0024] In this embodiment, as Figure 1 As shown, this intelligent scheduling method for multi-variety variable batch production includes the following steps: 1) Generate an initial production schedule that satisfies both process logic and static resource constraints based on the Resource Constrained Project Scheduling Model (RCPSP), and obtain the initial resource occupancy curve; 2) A genetic scheduling algorithm based on subspace optimization is used to perform initial scheduling on the initial resource occupancy curve to alleviate premature convergence of large-scale problems; 3) Determine if the production schedule is overdue, and identify overdue projects and bottleneck resources by using the average resource utilization rate and occupancy curve; for expandable resources other than the site, dynamically expand capacity according to the actual capacity of the enterprise, update the resource availability curve, and input it into the genetic production scheduling algorithm to iteratively optimize until all projects are not overdue or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference value of all projects to ensure on-time delivery.

[0025] In step 1) above, the method for constructing the resource-constrained project scheduling model includes the following steps: 1.1) Actual production cycle of the process: The scheduling cycle of a process is a set of high-precision discrete time points. (where time step) (0.1 days may be taken). Actual production cycle of the process. The constraint of equation (1) should be satisfied.

[0026]

[0027] In the formula, For a single piece (or a single batch) of standard production cycle; For single-piece (or single-batch) expedited processing, it indicates the minimum continuous execution time required for the process due to physical or technological limitations, which cannot be further reduced by increasing manpower or equipment.

[0028] 1.2) Process constraints: In multi-project parallel manufacturing scenarios, the execution of each process is not only constrained by resources but also must strictly adhere to process logic and timing relationships. The processes must satisfy the following constraints: (a) Batch Interval Constraint: Batch production processes must meet the batch interval constraint, that is, a certain process of Greater than It needs to be split into By constraining the process cycle and interval, batch production of processes can be achieved, closely mimicking the actual production process.

[0029] (b) Precedence constraints: For all processes, production can only begin after all preceding processes are completed. For processes whose preceding processes are all single-batch production processes, the constraints to be satisfied are shown in equation (2).

[0030]

[0031] In the formula, For the preceding process The completion time.

[0032] Two special cases should be considered further: (1) No preceding process: If the process is the first process of the project, its start time shall not be earlier than the earliest start time of the project, as shown in formula (3).

[0033]

[0034] (2) Case where the preceding process includes a batch production process: If a certain preceding process For batch production processes, let it be executed in total. Batch, No. The completion time of the batch is .

[0035] In actual production, to balance efficiency and material availability, companies require that the start time of subsequent processes not be earlier than the maximum of the completion time of the first batch and the completion time of other non-batch production pre-processes, and that their completion time not be earlier than the completion time of the last batch plus 1 / Multiply by the single-batch production cycle to ensure the complete flow of the entire batch of products.

[0036] Therefore, the definition includes pre-batch production processes. For subsequent processes The earliest startup time is shown in (4).

[0037] (4) In the formula, for The set of preceding processes excluding batch production processes.

[0038] The complete set constraint is shown in equation (5).

[0039]

[0040] In the formula, This is the single-batch cycle time for this process, used as the minimum buffer time.

[0041] 1.3) Resource Constraints: (a) Equipment resource constraints: Assume process A set of core equipment is required. For any of the devices Throughout its entire production cycle The cumulative equipment usage hours must equal the process quota value. (Unit: unit * hour), and the operating hours of the equipment must not exceed the maximum limit at any time.

[0042] First, the time requirement constraint of the main station is shown in equation (6).

[0043]

[0044] In the formula, , indicating process At any moment For equipment Instantaneous occupancy rate.

[0045] Secondly, the equipment capacity constraint at any given time is shown in equation (7).

[0046] (7) In the formula, For the collection of all core devices, This is the maximum available time for the equipment at any given moment, and it is related to the number of equipment and the working duration.

[0047] (b) Human resource constraints: setting up work processes Requires the use of specific types of personnel Throughout its entire production cycle The total manpower hours required are a fixed value. (Unit: person * hour), and the maximum number of working hours shall not be exceeded at any time.

[0048] First, the total manpower hours requirement is constrained as shown in equation (8).

[0049] (8) In the formula, , indicating process At any moment For job types The instantaneous rate of manpower occupancy.

[0050] Secondly, the human resource capacity constraint at any given time is shown in equation (9).

[0051] (9) In the formula For the collection of all job types, This represents the maximum available working hours at any given time, and is related to the number of workers and the duration of their work.

[0052] (c) Site resource constraints: design process Occupy core area The number of units typically occupied is fixed. Unlike equipment and manpower, site resources have continuous exclusivity: once activated, they must be held for the entire production cycle. It is continuously occupied and cannot be used by other processes during this period.

[0053] For any site type Let the number of available parallel units be . Then the venue allows Multiple processes use this type of space simultaneously.

[0054] To model this constraint, a binary indicator variable is introduced, as shown in Equation (10).

[0055] (10) For any site type and at any time The site resource constraints are shown in equation (11).

[0056] (11) 1.4) Optimization objective: Under the premise of satisfying all process logic constraints, dynamic resource capacity constraints and project time window constraints, minimize the sum of the difference values ​​of all projects (it is considered that completing 10 days in advance is optimal), as shown in Equation (12).

[0057]

[0058] In the formula, For the project The final completion time, For the project The latest delivery time.

[0059] In step 2) above, the present invention uses a genetic algorithm as the core algorithm for production scheduling optimization. Its goal is to generate a feasible production plan that minimizes the time difference of all projects through iterative evolution, under the premise of satisfying process logic and dynamic resource constraints.

[0060] Specifically, in this embodiment, a genetic scheduling algorithm based on subspace optimization is used to perform initial scheduling of the initial resource occupancy curve, including the following steps: 2.1) Using subspaces as the basic optimization unit, each subspace corresponds to one or more packaged model project production plans; each individual in the population represents a complete production scheduling scheme.

[0061] 2.2) Individual solutions employ a two-segment real-number encoding method combining process sequence and periodic coefficients to ensure solution feasibility; its encoding structure is as follows: Figure 2 As shown.

[0062] The two-segment real-number encoding consists of: the first half being a randomized arrangement of the global indices of the processes, representing the scheduling execution order. This sequence must satisfy the condition that all preceding processes of any process z must be completed before it. To ensure feasibility, a priority verification and adjustment function is embedded in the encoding process. This function performs topological correction on the input sequence based on a pre-built preceding-following relationship table; the second half is the periodic coefficient of process z. The normalized periodicity coefficient is used to dynamically determine the process. Actual execution cycle .

[0063] Specifically, the individual coding setting method is as follows: Assume the subspace contains Each production process within the facility is assigned a unique global index. And establish mapping relationships , The item number to which the process belongs. Assign its process number within the project. Individuals are encoded using a two-segment real number, with a total length twice the total number of processes in the subspace, as shown in equation (13).

[0064]

[0065] In the formula, the first half The global index is arranged in random order, following a priority sequence; the latter half It is a periodic sequence composed of the periodic coefficients of process z.

[0066] In the first half, the encoding process embeds priority checking and adjustment functions, such as Figure 3 As shown, the specific steps include: 2.2.1) Input any priority sequence, calculate the number of preceding processes i for each process, and create an empty priority list h; 2.2.2) Place the operations with i=0 in the sequence into the priority list h in order, traverse the priority list h, and take out the operations z in the sequence in order and put them into the result list result; 2.2.3) Traverse the list of preceding operations for operation z, and set the number of preceding operations pointed to by operation z to i-1; 2.2.4) Determine if the number of preceding operations i is 0. If it is not 0, return to step 2.2.3) and re-traverse the preceding operation list of operation z. If it is 0, put the operation pointed to by operation z into the priority list, and then return to step 2.2.3) and re-traverse the preceding operation list of operation z.

[0067] 2.2.5) After traversing the priority list h, we obtain the result list result after the order has been adjusted.

[0068] In the first half, such as Figure 4 As shown, there are predetermined pre- and post-task constraints between tasks: Task 1 is a pre-task of Task 2, Task 2 is a pre-task of Tasks 3 and 4, and Tasks 3 and 4 together constitute a pre-task of Task 5. During the data preprocessing stage, the encoding function uses the task pre- and post-task relation matrix to verify and adjust the randomly generated initial priority sequence, ultimately outputting a valid priority sequence that satisfies the process logic constraints.

[0069] In the latter half, the process Actual execution cycle As shown in equation (14).

[0070]

[0071] In the formula, For the original process Standard production cycle For the original process Based on the shortest expedited processing cycle and the pre-calculated average daily resource usage, the daily equipment hours, manpower hours, and site usage for process z within the cycle can be further derived.

[0072] 2.3) After the coding is completed, the start and end times are set sequentially according to the order of the processes in the coding. Then, the fitness of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding. The coded individual is transformed into a specific time arrangement that meets the dynamic resource constraints, and the merits of the plan are evaluated based on the fitness. 2.4) Under the premise of satisfying the process logic and dynamic resource constraints, generate a feasible production plan that minimizes the time difference of all projects through iterative evolution.

[0073] In step 2.3) above, the decoding process is implemented by a fitness calculation function. Its core is to transform the encoded individual into a specific time schedule that meets resource constraints, and to evaluate the merits of the solutions based on their fitness. Specifically, the fitness of the production plan is calculated based on the time of the project to which the process belongs to complete the decoding, such as... Figure 5 As shown, it includes the following steps: 2.3.1) The input consists of complete individual encoding and subspace information. The encoding is then split into a priority sequence. Given a periodic sequence, determine if the sequence is inverted. If it is inverted, invert the resource occupancy curve, swap the immediate preceding and immediate following lists, and swap the time complements of the front and rear walls.

[0074] 2.3.2) If there is no inverted index, then traverse the priority sequence and sort by priority. Process each step sequentially Calculate the actual cycle time and daily resource consumption of the current process z, and determine the earliest start time of process z. It is determined by the end time of all its immediate preceding processes.

[0075] 2.3.3) Call the resource limit check and time delay function to check the time range. Whether the resource occupation exceeds the instantaneous capacity limit of equipment, manpower and site, as shown in equations (6)-(11).

[0076] 2.3.4) If resource conflicts exist, the earliest start time will be used. The work is postponed until the first feasible window that meets the resource constraints is found, which is recorded as the actual start time. If there are no resource conflicts, the earliest start time will be used. Record as the actual start time .

[0077] 2.3.5) Based on actual commencement time Calculate completion time : And update the resource usage curve.

[0078] 2.3.6) After all the processes are arranged, calculate the fitness P of the scheme. The smaller the fitness P, the better the scheme.

[0079] Wherein, fitness P is:

[0080] In the formula, For the project The final completion time, For the project The latest delivery time.

[0081] This embodiment also includes step 2.5 of genetic crossover and mutation processing. A genetic operation strategy combining roulette wheel selection, conditional crossover, and Gaussian mutation is employed. The specific implementation method is as follows: 2.5.1) Selection operation: The roulette wheel selection is used to select parent individuals from the current population according to their fitness ratio.

[0082] 2.5.2) Crossover Operation: For the two-segment structure of chromosomes (the first half is the process priority sequence, and the second half is the normalized periodic coefficient), a conditional segmented crossover mechanism is designed: based on the crossover probability... Trigger crossover; if triggered, further probability is applied. The crossing-over method is determined (only one crossing-over method is selected per chromosome). It is divided into priority sequence crossing-over and cycle crossing-over. The crossing-over mechanisms of both are as follows: Figure 6 As shown.

[0083] 2.5.3) Mutation Operation: Design a segmented probability mutation mechanism for the same chromosome's two-segment structure: using mutation probability (Trigger mutation; if triggered, further mutation is carried out with probability.) The mutation mode is determined (only one mutation mode is selected per chromosome). It is divided into priority sequence mutations and periodic mutations. The mutation mechanisms of both are as follows: Figure 7 As shown.

[0084] 2.5.4) Elite Preservation Strategy: After each generation of evolution, the best individual from the parent generation (i.e. the one with the highest fitness) is directly copied to the new generation of the population to replace the worst individual.

[0085] Through the above encoding-decoding-iterative evolution, the genetic algorithm can efficiently search for near-optimal production scheduling solutions in complex constraint spaces, providing reliable input for subsequent bottleneck identification and dynamic resource adjustment.

[0086] In step 3) above, the production scheduling results are judged to be overdue, and overdue projects and bottleneck resources are identified by the average resource utilization rate and occupancy curve, including the following steps: 3.1.1) After the initial production scheduling plan is generated, resources are initially screened based on the average utilization rate, and a first threshold is set. The average utilization rate of the markers exceeds Resources.

[0087] Specifically, after the initial production schedule is generated, the project production plan, average resource utilization rate, and resource usage curve are read. If there is a project overdue (i.e., When identifying bottleneck resources that cause project delays, it is necessary to identify the type, specific name, and full-load operating time of the resource.

[0088] 3.1.2) By scanning the resource utilization curve, identify all instances where the utilization rate is consistently equal to or close to 100% (i.e., reaching the maximum instantaneous available time limit of the equipment). Or the maximum available working hours for manpower at any given time A continuous time period is defined as the period of high resource load operation. The number of days of high resource load operation is calculated, and a second threshold is set. The number of days marked as high-load operation is higher than Resources.

[0089] 3.1.3) Select those with an average utilization rate higher than the first threshold. And the number of days operating at high load exceeds the second threshold. The resource is marked as a bottleneck resource.

[0090] Specifically, the screening formula is:

[0091] In the formula, For resources Average utilization rate For resources The number of days of high-load operation. In this embodiment, the first threshold is used. The second threshold is 40%. It lasts for 30 days.

[0092] In step 3) above, for expandable resources other than the site, dynamic capacity expansion is carried out based on the company's actual capabilities, the resource availability curve is updated, and the genetic scheduling algorithm is input for iterative optimization until all projects are on time or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​of all projects to ensure on-time delivery, including the following steps: 3.2.1) Based on the identified bottleneck resources and combined with the company's actual dynamic adjustment principles, an initial resource occupancy curve that can be directly used by the genetic algorithm is generated.

[0093] 3.2.2) For different resource types, the dynamic expansion principle is as follows: human resources can be expanded by working overtime or adding people to increase the working hours limit for a set time period; equipment resources can be expanded by changing the shift system; site resources cannot be expanded.

[0094] 3.2.3) Based on the principle of dynamic expansion, the following adjustment suggestions are given for different types of bottleneck resources: 3.2.3.1) Filter out non-scalable resources. If the bottleneck resource is a site type, skip it directly without making any adjustments. 3.2.3.2) For expandable human or equipment resources, generate adjustment suggestions based on the length and urgency of their high-load operation periods; 3.2.3.3) Calculate the daily increase in resource capacity within the adjustment window and quantify the adjustment amount. As shown in equation (17).

[0095]

[0096] In the formula, To adjust the daily working hours limit for the window period; This indicates the duration of normal daily work.

[0097] 3.2.3.4) Construct a new resource occupancy curve and overlay the incremental information of all bottleneck resources onto the original static resource availability curve as input for the next round of genetic algorithm. The optimization objective of the algorithm is still to minimize the difference value of all items.

[0098] Meanwhile, at the end of each rescheduling iteration, a historical list of bottleneck resources is maintained. Before the next round of adjustments, the currently identified bottleneck list is compared and merged with this historical list, and the bottlenecks are divided into three categories: (1) Persistent bottlenecks: Resources that appear in both the historical and current lists. For these resources, more aggressive adjustment measures should be taken, such as directly recommending a three-shift system or increasing the number of staff on top of overtime. (2) Emerging Bottlenecks: Resources that only appear in the current list. A relatively conservative approach should be taken to adjust these resources. (3) Historical bottlenecks: Resources that only appear in the historical list but not in the current scheduling. Such resources usually do not need to be adjusted and can be directly retained.

[0099] In this embodiment, to avoid the rescheduling process getting stuck in an invalid loop, an overdue analysis of the current production schedule is performed after each iteration, and the decision to terminate the iteration is based on this analysis. The iteration termination condition is any one of the following: (a) There are no overdue projects, i.e. ; (b) If the set of overdue projects and the total number of overdue days do not change after three consecutive rescheduling iterations (as shown in Equation (18)), it is determined that the remaining overdue projects are not caused by resource bottlenecks, but by unreasonable production plans. This situation cannot be resolved by increasing resources.

[0100] (18) in, .

[0101] In this embodiment, after completing all rescheduling iterations, the latest resource usage curve will be analyzed again, and the specific implementation method for analysis and adjustment includes the following steps: (1) Assume resources It was given an increased capacity during a certain adjustment. The increment was originally planned to be applied to the time period. This interval is divided into a set of discrete time points. Each of them This corresponds to a scheduling time step (e.g., 0.1 days). (2) Let This indicates that in the final production schedule, resources... At any moment The actual amount occupied; (3) At time Determine whether the increased capacity is being used. If not, adjust the available resource capacity at that moment. Back to Conversely, if the increase in capacity is not used, it indicates that the increased capacity has been utilized, thus yielding the final optimized resource availability curve. .

[0102] Equation (19) can be used to determine whether the capacity expansion increment has been used: (19) If equation (19) holds, then Not used.

[0103] Resource availability curve for: (20) in, .

[0104] This rollback mechanism effectively prevents resource waste caused by excessive conservatism or misjudgment, ensuring that the final production schedule meets delivery requirements while maintaining the economy and rationality of resource utilization.

[0105] In summary, this invention uses a genetic algorithm as its core and generates an initial production schedule that satisfies both process logic and static resource constraints based on the Resource Constrained Project Scheduling Model (RCPSP). Individual projects employ a two-stage encoding (process sequence + cycle coefficient) to ensure the feasibility of the solution. Subsequently, the scheduling results are judged for overdue status, and bottlenecks in equipment or manpower causing delays are identified through average resource utilization and occupancy curves. For scalable resources (excluding site space), dynamic capacity expansion is implemented based on the company's actual capabilities, updating the resource availability curve. Finally, the new constraints are fed back to the genetic algorithm for rescheduling, iteratively optimizing until all projects are within their due dates or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​for all projects, striving to complete delivery 10 days before the contract deadline. By combining a static scheduling model with dynamic resource decision-making, this invention not only retains the efficiency of genetic algorithms in large-scale combinatorial optimization but also endows the scheduling system with adaptability and robustness to actual production disturbances, providing a feasible technical path for high-end manufacturing enterprises to achieve precise delivery.

[0106] In one embodiment of the present invention, a multi-variety variable batch intelligent production scheduling system is provided, comprising: The data preprocessing module generates an initial production schedule that satisfies both process logic and static resource constraints based on the resource-constrained project scheduling model, and obtains the initial resource occupancy curve. The initial production scheduling module uses a subspace-optimized genetic production scheduling algorithm to perform initial production scheduling on the initial resource occupancy curve; The bottleneck identification, dynamic capacity expansion, and rescheduling modules determine overdue production results and identify overdue projects and bottleneck resources through average resource utilization and occupancy curves. For expandable resources other than the site, dynamic capacity expansion is carried out based on the company's actual capabilities, updating the resource availability curve and inputting it into the genetic production scheduling algorithm for iterative optimization until all projects are not overdue or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​of all projects to ensure on-time delivery.

[0107] In the above embodiments, a genetic scheduling algorithm based on subspace optimization is used to perform initial scheduling on the initial resource occupancy curve, including: The subspace is used as the basic optimization unit, and each subspace corresponds to one or more packaged model project production plans; each individual in the population represents a complete production scheduling scheme. Individual processes employ a two-segment real-number encoding method, combining process sequence and periodic coefficient, to ensure solution feasibility. The two-segment real-number encoding consists of: the first half, a randomized arrangement of global indices of the processes, representing the scheduling execution order. This sequence must satisfy the condition that all preceding processes of any process z must be completed before it. The encoding process embeds a priority check and adjustment function, which performs topological correction on the input sequence based on a pre-constructed preceding-following relationship table; the second half represents the periodic coefficient of process z. The normalized periodicity coefficient is used to dynamically determine the process. Actual execution cycle ; After coding is completed, start and end times are set sequentially according to the order of the processes in the code. Then, the fitness of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding. The coded individual is transformed into a specific time arrangement that meets the dynamic resource constraints, and the merits of the plan are evaluated based on the fitness. Under the premise of satisfying process logic and dynamic resource constraints, a feasible production plan that minimizes the time difference of all projects is generated through iterative evolution.

[0108] In the above embodiments, the fitness of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding, including: The input consists of complete individual codes and subspace information. The codes are split into priority sequences and periodic sequences. It is determined whether the sequences are inverted. If they are inverted, the resource occupancy curve is inverted, the immediate predecessor and immediate successor lists are swapped, and the time complements of the front and back walls are swapped. If there is no inverted index, then traverse the priority sequence and process each step in the order of the priority sequence. Calculate the actual cycle time and daily resource consumption of the current process z, and determine the earliest start time of process z. It is determined by the end time of all its immediate preceding processes; Call the resource limit validation and time delay function to validate the time range. Does the resource consumption exceed the instantaneous capacity limit of equipment, manpower, and site? If resource conflicts exist, the earliest start time will be used. The work is postponed until the first feasible window that meets the resource constraints is found, which is recorded as the actual start time. If there are no resource conflicts, the earliest start time will be used. Record as the actual start time ; Based on actual start time Calculate completion time : And update the resource usage curve; After all the processes are arranged, calculate the fitness P of the solution. The smaller the fitness P, the better the solution.

[0109] In the above embodiments, the production scheduling results are overdue, and overdue projects and bottleneck resources are identified through the average resource utilization rate and occupancy curve, including: After the initial production scheduling plan is generated, resources are initially screened based on average utilization rate, and a first threshold is set. The average utilization rate of the markers exceeds Resources; By scanning the resource utilization curve, all continuous time periods with utilization rates consistently equal to or close to 100% are identified as high-load operation periods for the resources. The number of days of high-load operation is then calculated, and a second threshold is set. The number of days marked as high-load operation is higher than Resources; Filter out those with an average utilization rate higher than the first threshold And the number of days operating at high load exceeds the second threshold. The resource is marked as a bottleneck resource.

[0110] In the above embodiments, for expandable resources other than the site, dynamic capacity expansion is carried out based on the actual capabilities of the enterprise, the resource availability curve is updated, and the genetic scheduling algorithm is input for iterative optimization until all projects are on time or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​of all projects to ensure on-time delivery, including: Based on the identified bottleneck resources and combined with the company's actual dynamic adjustment principles, an initial resource occupancy curve that can be directly used by the genetic algorithm is generated. For different resource types, the established dynamic expansion principles are as follows: human resources can be expanded by adding overtime or additional personnel to increase the working hours limit for a set time period; equipment resources can be expanded by changing the shift system; and site resources cannot be expanded. Based on the principle of dynamic expansion, the following adjustment suggestions are given for different types of bottleneck resources: Filter out non-scalable resources. If the bottleneck resource is a site type, skip it directly without making any adjustments. For expandable human or equipment resources, adjustment suggestions are generated based on the length and urgency of their high-load operation periods; Calculate the daily increase in resource capacity within the adjustment window and quantify the adjustment amount; A new resource occupancy curve is constructed, and the incremental information of all bottleneck resources is superimposed on the original static resource availability curve as input for the next round of genetic algorithm. The optimization objective of the algorithm is still to minimize the difference value of all items.

[0111] In the above embodiments, after each iteration, the current production schedule will be overdue for analysis, and the iteration will be terminated accordingly. The iteration termination condition can be any of the following: (a) There are no overdue projects; (b) If the set of overdue projects and the total number of overdue days do not change after three consecutive rescheduling iterations, it is determined that the remaining overdue projects are not caused by resource bottlenecks, but by unreasonable production plans. This situation cannot be resolved by increasing resources.

[0112] In the above embodiment, after completing all rescheduling iterations, the latest resource usage curve will be analyzed again, and the specific implementation method for analysis and adjustment is as follows: Resource It was given an increased capacity during a certain adjustment. The increment was originally planned to be applied to the time period. This interval is divided into a set of discrete time points. Each of them This corresponds to a scheduling time step; set up This indicates that in the final production schedule, resources... At any moment The actual amount occupied; At any moment Determine whether the increased capacity is being used. If not, adjust the available resource capacity at that moment. Back to Conversely, if the increase in capacity is not used, it indicates that the increased capacity has been utilized, thus yielding the final optimized resource availability curve. .

[0113] The system provided in this embodiment is used to execute the above-described method embodiments. For specific processes and details, please refer to the above embodiments, which will not be repeated here.

[0114] In one embodiment of the present invention, a computing device is provided. This computing device can be a terminal and may include a processor, a communication interface, memory, a display screen, and an input device. The processor, communication interface, and memory communicate with each other via a communication bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. When the computer programs are executed by the processor, they implement the methods described in the above embodiments. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, a management network, NFC (Near Field Communication), or other technologies. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad mounted on the casing of the computing device, or an external keyboard, touchpad, or mouse. The processor can call logical instructions stored in the memory.

[0115] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0116] In one embodiment of the present invention, a computer program product is provided, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to perform the methods provided in the above-described method embodiments.

[0117] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, which stores server instructions that cause a computer to perform the methods provided in the above embodiments.

[0118] The computer-readable storage medium provided in the above embodiments has a similar implementation principle and technical effect to the above method embodiments, and will not be described again here.

[0119] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0120] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0121] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-variety, variable-batch intelligent production scheduling method, characterized in that, include: An initial production schedule that satisfies both process logic and static resource constraints is generated based on the resource-constrained project scheduling model, resulting in an initial resource occupancy curve. A genetic scheduling algorithm based on subspace optimization is used to perform initial scheduling of the initial resource occupancy curve; The scheduling results are judged to be overdue, and overdue projects and bottleneck resources are identified by the average resource utilization rate and occupancy curve. For expandable resources other than the site, the capacity is dynamically expanded according to the actual capacity of the enterprise, the resource availability curve is updated, and the genetic scheduling algorithm is input. Iterative optimization is carried out until all projects are not overdue or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference value of all projects to ensure on-time delivery.

2. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 1, characterized in that, A genetic scheduling algorithm based on subspace optimization is used to perform initial scheduling on the initial resource occupancy curve, including: The subspace is used as the basic optimization unit, and each subspace corresponds to one or more packaged model project production plans; each individual in the population represents a complete production scheduling scheme. Individual processes employ a two-segment real-number encoding method, combining process sequence and periodic coefficient, to ensure solution feasibility. The two-segment real-number encoding consists of: the first half, a randomized arrangement of global indices of the processes, representing the scheduling execution order. This sequence must satisfy the condition that all preceding processes of any process z must be completed before it. The encoding process embeds a priority check and adjustment function, which performs topological correction on the input sequence based on a pre-constructed preceding-following relationship table; the second half represents the periodic coefficient of process z. The normalized periodicity coefficient is used to dynamically determine the process. Actual execution cycle ; After coding is completed, start and end times are set sequentially according to the order of the processes in the code. Then, the fitness of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding. The coded individual is transformed into a specific time arrangement that meets the dynamic resource constraints, and the merits of the plan are evaluated based on the fitness. Under the premise of satisfying process logic and dynamic resource constraints, a feasible production plan that minimizes the time difference of all projects is generated through iterative evolution.

3. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 2, characterized in that, The suitability of the production plan is calculated based on the time of the project to which the process belongs in order to complete the decoding, including: The input consists of complete individual codes and subspace information. The codes are split into priority sequences and periodic sequences. It is determined whether the sequences are inverted. If they are inverted, the resource occupancy curve is inverted, the immediate predecessor and immediate successor lists are swapped, and the time complements of the front and back walls are swapped. If there is no inverted index, then traverse the priority sequence and process each step in the order of the priority sequence. Calculate the actual cycle time and daily resource consumption of the current process z, and determine the earliest start time of process z. It is determined by the end time of all its immediate preceding processes; Call the resource limit validation and time delay function to validate the time range. Does the resource consumption exceed the instantaneous capacity limit of equipment, manpower, and venue? If resource conflicts exist, the earliest start time will be used. The work is postponed until the first feasible window that meets the resource constraints is found, which is recorded as the actual start time. If there are no resource conflicts, the earliest start time will be used. Record as the actual start time ; Based on actual start time Calculate completion time : And update the resource usage curve; After all the processes are arranged, calculate the fitness P of the solution. The smaller the fitness P, the better the solution.

4. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 1, characterized in that, The production scheduling results are assessed for overdue status, and overdue projects and bottleneck resources are identified through average resource utilization and occupancy curves, including: After the initial production scheduling plan is generated, resources are initially screened based on average utilization rate, and a first threshold is set. The average utilization rate of the markers exceeds Resources; By scanning the resource utilization curve, all continuous time periods with utilization rates consistently equal to or close to 100% are identified as high-load operation periods for the resources. The number of days of high-load operation is then calculated, and a second threshold is set. The number of days marked as high-load operation is higher than Resources; Select those with an average utilization rate higher than the first threshold. And the number of days operating at high load exceeds the second threshold. The resource is marked as a bottleneck resource.

5. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 1, characterized in that, For expandable resources other than the site, dynamic capacity expansion is implemented based on the company's actual capabilities. The resource availability curve is updated, and the data is input into a genetic scheduling algorithm for iterative optimization until all projects are on time or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​for all projects, ensuring on-time delivery. This includes: Based on the identified bottleneck resources and combined with the company's actual dynamic adjustment principles, an initial resource occupancy curve that can be directly used by the genetic algorithm is generated. For different resource types, the established dynamic expansion principles are as follows: human resources can be expanded by adding overtime or additional personnel to increase the working hours limit for a set time period; equipment resources can be expanded by changing the shift system; and site resources cannot be expanded. Based on the principle of dynamic expansion, the following adjustment suggestions are given for different types of bottleneck resources: Filter out non-scalable resources. If the bottleneck resource is a site type, skip it directly without making any adjustments. For expandable human or equipment resources, adjustment suggestions are generated based on the length and urgency of their high-load operation periods; Calculate the daily increase in resource capacity within the adjustment window and quantify the adjustment amount; A new resource occupancy curve is constructed, and the incremental information of all bottleneck resources is superimposed on the original static resource availability curve as input for the next round of genetic algorithm. The optimization objective of the algorithm is still to minimize the difference value of all items.

6. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 5, characterized in that, After each iteration, an overdue analysis will be performed on the current production schedule, and a decision will be made as to whether to terminate the iteration. The iteration termination condition can be any of the following: (a) There are no overdue projects; (b) If the set of overdue projects and the total number of overdue days do not change after three consecutive rescheduling iterations, it is determined that the remaining overdue projects are not caused by resource bottlenecks, but by unreasonable production plans. This situation cannot be resolved by increasing resources.

7. The intelligent production scheduling method for multiple product varieties and variable batch sizes as described in claim 5, characterized in that, After completing all rescheduling iterations, the latest resource usage curve will be analyzed again, and the specific implementation method will be adjusted accordingly: Resource It was given an increased capacity during a certain adjustment. The increment was originally planned to be applied to the time period. This interval is divided into a set of discrete time points. Each of them This corresponds to a scheduling time step; set up This indicates that in the final production schedule, resources... At any moment The actual amount occupied; At any moment Determine whether the increased capacity is being used. If not, adjust the available resource capacity at that moment. Back to ; Conversely, if the increase in capacity is not used, it indicates that the increased capacity has been utilized; thus, the final optimized resource availability curve is obtained. .

8. A multi-variety, variable-batch intelligent production scheduling system, characterized in that, include: The data preprocessing module generates an initial production schedule that satisfies both process logic and static resource constraints based on the resource-constrained project scheduling model, and obtains the initial resource occupancy curve. The initial production scheduling module uses a subspace-optimized genetic production scheduling algorithm to perform initial production scheduling on the initial resource occupancy curve; The bottleneck identification, dynamic capacity expansion, and rescheduling modules determine overdue production results and identify overdue projects and bottleneck resources through average resource utilization and occupancy curves. For expandable resources other than the site, dynamic capacity expansion is carried out based on the company's actual capabilities, updating the resource availability curve and inputting it into the genetic production scheduling algorithm for iterative optimization until all projects are not overdue or the maximum number of iterations is reached. The optimization objective is to minimize the total weighted difference in period values ​​of all projects to ensure on-time delivery.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 7.

Images