Production scheduling visualization method based on big data

Abstract

The invention relates to the technical field of production scheduling, and discloses a production scheduling visualization method based on big data. According to the method, multi-dimensional production data such as equipment real-time states, material inventory, personnel skills and order delivery periods are collected, and a four-dimensional dynamic constraint model is generated through a dynamic constraint modeling algorithm; and obtaining an initial scheduling scheme by using a mixed integer programming and improved genetic algorithm dual-drive optimization engine, inputting the initial scheduling scheme into a visual interaction module, displaying a result by using a drag-and-drop Gantt chart, and detecting resource conflicts in real time. The feasibility of the adjusted scheduling scheme is verified through a dynamic re-calculation algorithm, and then the model is updated. According to the method, multi-dimensional data are integrated, production scheduling optimization and visualization are realized, production resource conflicts can be effectively solved, the production efficiency is improved, the cost is reduced, and an efficient solution is provided for enterprise production management.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling, and in particular to a production scheduling visualization method based on big data. Background Art

[0002] As modern manufacturing industry is booming, the production environment is becoming more and more complex. Enterprises are facing unprecedented challenges, and efficient production scheduling has become a key factor in improving the competitiveness of enterprises. Traditional production scheduling methods mainly rely on manual experience, which has many disadvantages. First of all, manual scheduling is difficult to fully and accurately consider the many complex factors in the production process. In terms of equipment, it is impossible to grasp the operating status of the equipment, the available time window, and the potential risk of failure in real time, resulting in low equipment utilization, frequent equipment idleness or overuse, which not only increases production costs, but may also affect the production progress and quality of products.

[0003] In terms of material management, it is difficult for manual labor to accurately control material inventory data, and there is a lack of timely and accurate judgment on material completeness indicators and replenishment cycles, which can easily lead to material shortages or backlogs. Material shortages can cause production lines to stagnate and delay order delivery; material backlogs take up a lot of funds and storage space, increasing corporate operating costs. Manual scheduling is often difficult to achieve scientific and reasonable matching of personnel skills with production tasks, and it is impossible to give full play to the professional skills of employees, reducing production efficiency.

[0004] With the diversification and personalization of market demand, the requirements for order delivery are becoming more and more stringent. Manual scheduling is difficult to quickly respond to changes in order delivery while ensuring product quality, which is prone to delays in order delivery, damaging the company's reputation and customer satisfaction. In addition, traditional scheduling methods lack visualization methods, making it difficult for production managers to intuitively understand the progress and resource allocation of the entire production process, which is not conducive to timely discovery and resolution of production problems.

[0005] Although some companies have introduced some simple production management software, these software have great functional limitations. They can often only analyze and process a single factor, and cannot integrate multi-dimensional production data such as equipment capacity, material completeness, personnel skills and order delivery time, making it difficult to build a comprehensive and accurate production constraint model. In terms of scheduling optimization algorithms, most of these software use simple rules or traditional algorithms, and cannot find the optimal or near-optimal scheduling solution in a complex production environment. Moreover, the visual interface of these software is not flexible enough and has poor interactivity, and cannot meet the needs of users to adjust scheduling and detect resource conflicts in real time.

[0006] The rapid development of big data and artificial intelligence technologies has brought new opportunities to the field of production scheduling. However, the relevant solutions on the market have not yet fully utilized the advantages of these advanced technologies to achieve intelligent, efficient and visualized production scheduling. Therefore, developing a production scheduling visualization method and system based on big data to solve the shortcomings of traditional scheduling methods and meet the growing production management needs of modern enterprises has become an important issue that needs to be solved in this field. Summary of the invention

[0007] The purpose of the present invention is to provide a production scheduling visualization method based on big data to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a production scheduling visualization method based on big data, the method comprising: Collect multi-dimensional production data sets; the multi-dimensional production data includes real-time equipment status data, material inventory data, personnel skill data and order delivery date data; the real-time equipment status data includes equipment available time window and fault warning information, and the material inventory data includes material completeness index and replenishment cycle; Based on the multi-dimensional production data, a four-dimensional dynamic constraint model is generated by integrating equipment capacity constraints, material completeness constraints, personnel skill constraints and order delivery constraints through a dynamic constraint modeling algorithm; According to the four-dimensional dynamic constraint model, a mixed integer programming and improved genetic algorithm dual-driven optimization engine are used to generate an initial scheduling plan; Input the initial scheduling plan into the visual interaction module, display the scheduling results through a drag-and-drop Gantt chart, and detect resource conflicts in the adjustment operation in real time; Based on the adjusted scheduling scheme, the feasibility is verified through a dynamic recalculation algorithm, and the four-dimensional dynamic constraint model is updated.

[0009] Preferably, the dynamic constraint modeling algorithm integrates equipment capacity constraints, material completeness constraints, personnel skill constraints and order delivery constraints to generate a four-dimensional dynamic constraint model, including: Performing availability segmentation processing on the real-time status data of the device to generate a device time occupancy rate curve; Calculate the completeness gap based on the material inventory data, and generate a dynamic constraint chain for material supply in combination with the replenishment cycle; Match the order process complexity according to personnel skill data and build a skill-task association matrix; The order delivery date data is decomposed into process-level time nodes, and integrated with the equipment time occupancy rate curve, the material supply dynamic constraint chain and the skill and task association matrix to generate a four-dimensional dynamic constraint model.

[0010] Preferably, the construction method of the hybrid integer programming and improved genetic algorithm dual-driven optimization engine includes: Initialize the objective function of the hybrid integer programming model to minimize the total production cost, and load the four-dimensional dynamic constraint model as the constraint condition; Adopt an improved genetic algorithm to design a variable-length chromosome coding mechanism, and encode equipment allocation, material consumption, and personnel scheduling as gene segments; Through an alternating iteration mechanism, inject the local optimal solution of the hybrid integer programming model into the population mutation operation of the genetic algorithm to generate a global optimization solution set; Screen the initial scheduling plan according to the preset convergence threshold.

[0011] Preferably, the method for real-time conflict detection through a drag-and-drop Gantt chart includes: Analyze the operation time series in the scheduling plan to generate a task dependency tree with a hierarchical relationship; When listening to the user's drag operation, extract the spatio-temporal coordinates of the adjusted operation and the associated resource identifier; Traverse the resource occupancy status of adjacent nodes in the task dependency tree, and calculate the time overlap degree and resource conflict coefficient; Dynamically mark the conflict area through color coding and trigger a conflict warning prompt.

[0012] Preferably, the verification method of the dynamic recalculation algorithm includes: Capture the changed operation parameters in the adjusted scheduling plan, extract the local recalculation area, construct a sub-problem solution space based on the four-dimensional dynamic constraint model, and retain the scheduling results of the unadjusted area; Adopt a heuristic backtracking algorithm to traverse the feasible solutions of the sub-problems, perform consistency verification with the global scheduling plan, merge the feasible solutions that pass the verification into the global scheduling plan, and update the display data of the visual interaction module.

[0013] Preferably, the generation method of the equipment time occupancy rate curve includes: Predict the maintenance time window according to the equipment failure warning information, and divide the available time period and unavailable time period of the equipment; Adopt a time slice rotation algorithm to calculate the time occupancy weight of each operation on the equipment; Combine historical production data to fit the equipment load balancing curve and optimize the time occupancy rate segmentation accuracy.

[0014] Preferably, the parameter optimization method of the improved genetic algorithm includes: Adaptively adjust the crossover probability and mutation probability according to the complexity of the four-dimensional dynamic constraint model; Design an elite retention strategy to force the local optimal solution output by the hybrid integer programming to be retained in the next generation population; By integrating the production cost, delivery achievement rate and resource utilization rate indicators through a dynamic fitness function, the chromosome screening mechanism is optimized.

[0015] Preferably, the method for constructing the task dependency tree includes: Analyze the precedence relationship of processes in the order data to generate a directed acyclic graph structure; Map the process nodes to the equipment timeline based on the equipment allocation result to form a spatio-temporal dependency link; Traverse the dependency link using the depth-first search algorithm to construct a task dependency tree with weight relationships.

[0016] Preferably, the execution steps of the heuristic backtracking algorithm include: Define the resource conflict priority of the local recomputation area, generate a backtracking search path, and prune the invalid solution space through the constraint propagation algorithm to narrow the backtracking range; Adopt the memoization search technique to cache the states of the traversed solutions, select the optimal local solution according to the verification result, and trigger the update of the global scheduling scheme.

[0017] Preferably, the present invention further includes a production scheduling visualization system based on big data, and the system includes: Data acquisition module: used to acquire a multi-dimensional production data set, and the multi-dimensional production data includes equipment real-time status data, material inventory data, personnel skill data and order delivery date data; Dynamic modeling module: configured to integrate equipment production capacity constraints, material completeness constraints, personnel skill constraints and order delivery date constraints through a dynamic constraint modeling algorithm to generate a four-dimensional dynamic constraint model; Optimization engine module: used to generate an initial scheduling scheme according to the four-dimensional dynamic constraint model by using a hybrid integer programming and improved genetic algorithm dual-driven optimization engine; Visualization interaction module: display the scheduling result through a drag-and-drop Gantt chart and detect resource conflicts in the adjustment operation in real time; Real-time verification module: based on the adjusted scheduling scheme, verify the feasibility through a dynamic recomputation algorithm and update the four-dimensional dynamic constraint model.

[0018] Compared with the prior art, the beneficial effects of the present invention are: In terms of data collection and integration, by comprehensively collecting real-time status data of equipment, material inventory data, personnel skill data, and order delivery date data, multi-dimensional key information in the production process can be obtained. The in-depth integration of these data breaks the situation of isolated data in each link of traditional production management. Enterprises can plan equipment maintenance in advance based on the real-time status data of equipment, reduce the impact of equipment failures on production, and improve equipment utilization rate; according to the material inventory data, accurately control the material availability and replenishment time, avoid material shortages or overstocking, and reduce inventory costs; by combining personnel skill data and order process complexity, achieve precise matching of personnel and tasks, give full play to the professional capabilities of employees, and improve production efficiency.

[0019] Generating a four-dimensional dynamic constraint model is one of the key innovations of the present invention. This model effectively integrates equipment production capacity constraints, material availability constraints, personnel skill constraints, and order delivery date constraints, comprehensively reflecting various limiting conditions in the production process. Compared with traditional single-constraint or simple combined constraint models, it is closer to the actual production scenario, providing a solid foundation for formulating scheduling plans. When formulating production plans, enterprises can conduct scientific analysis based on this model, avoid unreasonable production plans caused by ignoring certain key constraints, and ensure the smooth progress of the production process.

[0020] Adopting a dual-driven optimization engine of mixed integer programming and improved genetic algorithm significantly improves the quality and generation efficiency of scheduling plans. Mixed integer programming can optimize objective functions such as production costs on the basis of meeting constraint conditions; the improved genetic algorithm enhances the search ability and convergence speed of the algorithm by designing a variable-length chromosome coding mechanism, adaptively adjusting parameters, and an elite retention strategy. The two cooperate with each other, and can quickly find a near-optimal initial scheduling plan in a complex production environment, saving a large amount of time and resources for enterprises. Compared with traditional single-algorithm optimization methods, the dual-driven optimization engine can better balance local search and global search, avoid the algorithm falling into local optimal solutions, and improve the overall quality of scheduling plans.

[0021] The introduction of a visual interaction module greatly improves the convenience and intuitiveness of production management. By displaying scheduling results through a drag-and-drop Gantt chart, production managers can clearly understand the progress arrangement and resource allocation of the entire production process at a glance. When making scheduling adjustments, the system can detect resource conflicts in real time and feedback them to users in a timely manner through color coding and warning prompts. This visual interaction method enables production managers to quickly discover problems and make adjustments, greatly improving the efficiency and accuracy of production decisions. Compared with traditional text or table-based scheduling displays, the visual interaction module reduces the difficulty of production management and reduces the occurrence of human errors.

[0022] The dynamic recalculation algorithm and real-time verification module ensure the feasibility and real-time nature of the scheduling plan. After the scheduling plan is adjusted, the dynamic recalculation algorithm can quickly verify whether the new plan meets the requirements of the four-dimensional dynamic constraint model. By capturing the changed process parameters, extracting the local recalculation area, and using the heuristic backtracking algorithm to traverse the feasible solutions, it not only ensures the accuracy of verification but also improves the calculation efficiency. At the same time, the four-dimensional dynamic constraint model is updated in real time, enabling the model to promptly reflect the changes in the production process and providing a more accurate basis for subsequent scheduling optimization. This real-time verification and update mechanism ensures that enterprises can quickly adjust the scheduling plan in the face of various changes in the production process, guaranteeing the continuity and stability of production. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is the working principle diagram of the production scheduling visualization method based on big data according to the present invention; Figure 2 is the flowchart of the generation process of the four-dimensional dynamic constraint model; Figure 3 is the flowchart of the real-time conflict detection process of the drag-and-drop Gantt chart; Figure 4 is the flowchart of the verification process of the dynamic recalculation algorithm. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] Please refer to Figures 1 - 4 , the present invention provides a technical solution: The present invention relates to a production scheduling visualization method based on big data, and the specific implementation solution is as follows: During the production process, production data is collected from multiple data sources. The real-time status data of the equipment is obtained by installing sensors on the equipment, and these sensors can monitor the operating conditions of the equipment in real time, thereby obtaining the available time window of the equipment and the fault warning information; the material inventory data can be extracted from the enterprise's inventory management system, from which the material completeness index and the replenishment cycle can be obtained; the personnel skill data is collected by sorting out the employee skill training records and work experience; and the order delivery date data is read from the order management system. Through this multi-channel data collection method, comprehensive and accurate multi-dimensional production data is obtained, providing a solid data foundation for subsequent scheduling analysis.

[0026] Based on the collected multi-dimensional production data, the dynamic constraint modeling algorithm is used to integrate equipment production capacity constraints, material completeness constraints, personnel skill constraints, and order delivery date constraints. This algorithm comprehensively considers the correlations between various data, incorporates constraint factors of different dimensions into a model, and generates a four-dimensional dynamic constraint model that can reflect the actual production situation, providing constraint conditions for subsequent scheduling optimization.

[0027] Using a dual-driven optimization engine of mixed integer programming and improved genetic algorithm, an initial scheduling plan is generated according to the four-dimensional dynamic constraint model. This optimization engine combines the advantages of the two algorithms, and on the premise of meeting various constraint conditions, searches for a better production scheduling plan, improves the utilization efficiency of production resources, and reduces production costs.

[0028] The generated initial scheduling plan is input into the visual interactive module, and the scheduling results are displayed through a drag-and-drop Gantt chart. When the user adjusts the scheduling, this module real-time detects resource conflict situations. The Gantt chart presents the time arrangement and resource allocation of production tasks in an intuitive graphical way, facilitating the user to quickly understand the production progress and resource usage status. When the user adjusts tasks, the system can promptly detect and prompt potential resource conflicts, facilitating the user to make reasonable adjustments.

[0029] Based on the adjusted scheduling plan, the dynamic recalculation algorithm is used to verify its feasibility. If the plan is feasible, the four-dimensional dynamic constraint model is updated according to the new scheduling results, enabling the model to promptly reflect changes in the actual production situation and providing a more accurate basis for subsequent scheduling optimization and adjustment.

[0030] Example 1: During the process of generating the four-dimensional dynamic constraint model, the real-time status data of the equipment is processed by availability segmentation to generate the equipment time occupancy rate curve. The specific steps are as follows: Predict the maintenance time window based on the equipment fault warning information, and divide the available time period and unavailable time period of the equipment. Assume that the equipment fault warning information is obtained through the built-in monitoring system of the equipment. When the operating parameters of a certain component of the equipment are detected to exceed the normal range, the system will issue a fault warning. Based on past maintenance experience and the operating conditions of the equipment, the maintenance time window is predicted. For example, if the temperature of a key component of a certain equipment continues to rise and approaches the warning value, through historical data analysis, it is predicted that maintenance is required within the next 24 hours. Then these 24 hours are the predicted maintenance time window, and the equipment is unavailable during this period, and other times are available time periods.

[0031] The time slice rotation algorithm is used to calculate the time occupancy weight of each process on the equipment. The formula of the time slice rotation algorithm is: , where represents the time occupancy weight of the th process on the equipment, represents the The time expected to occupy the equipment for a process represents the total number of processes that need to be executed on the equipment represents the total time expected to occupy the equipment for all processes. Through this formula, the relative weight of each process in terms of equipment usage time can be calculated, thus more reasonably allocating equipment resources.

[0032] Fit the equipment load balancing curve by combining historical production data and optimize the segmentation accuracy of the time occupancy rate. Collect the production data of the equipment over a past period, including the workload and operating efficiency of the equipment in different time periods. Use data analysis tools and adopt curve fitting methods, such as the least squares method, to fit the equipment load balancing curve. Through this curve, the load situation of the equipment in different time periods can be understood more accurately, and then the segmentation of the time occupancy rate can be optimized, making the generated equipment time occupancy rate curve better reflect the actual production situation.

[0033] Calculate the kit shortage based on the material inventory data and generate a dynamic material supply constraint chain in combination with the replenishment cycle. The material kit completeness index reflects the matching situation of the materials required for production. Assume that the material kit completeness index is represented by When it means that the materials are completely kitted; when it means that there is a material shortage. The formula for calculating the kit shortage is: , where represents the kit shortage. According to the replenishment cycle of the materials and the time arrangement of the production plan, a dynamic material supply constraint chain is generated. For example, if the replenishment cycle of a certain material is 7 days and in the production plan, this material needs to be put into use on the 10th day, then the supply of this material needs to be ensured starting from the 3rd day, otherwise the production progress will be affected, which forms a dynamic material supply constraint chain.

[0034] Match the process complexity of the order according to the personnel skill data and construct a skill-task association matrix. Classify and organize the personnel skill data according to skill types, skill levels, etc., and at the same time evaluate the process complexity of the order, which can be divided into different levels such as simple, medium, and complex. Let the number of personnel be , and the number of order process tasks be , and construct a skill-task association matrix , where represents the skill matching degree of the th person to the th task. When the th person fully possesses the skills required to complete the th task, ; when partially possessing, assign according to the degree of skill mastery a value between ; when it is completely absent, . Through this correlation matrix, the matching situation between personnel skills and order tasks can be clearly understood, providing a basis for personnel scheduling.

[0035] Decompose the order delivery date data into process-level time nodes, and fuse them with the equipment time occupancy rate curve, the dynamic constraint chain of material supply, and the skill-task correlation matrix to generate a four-dimensional dynamic constraint model. Decompose the delivery date of an order according to the production process to determine the latest start time and the latest completion time of each process. For example, if the delivery date of an order is the 30th day and this order contains 5 processes, according to the process sequence and production time estimation, decompose the delivery date into time nodes for each process. Then fuse these process-level time nodes with the previously generated equipment time occupancy rate curve, the dynamic constraint chain of material supply, and the skill-task correlation matrix, comprehensively considering various constraint factors such as equipment, materials, personnel, and time, to generate a four-dimensional dynamic constraint model that can accurately reflect the actual production situation.

[0036] Embodiment 2: The construction process of the dual-driven optimization engine of mixed integer programming and improved genetic algorithm is as follows: Initialize the objective function of the mixed integer programming model to minimize the total production cost. Let the total production cost be , which consists of equipment cost , material cost , personnel cost , etc., that is . In actual production, the equipment cost includes the purchase cost and maintenance cost of the equipment; the material cost is the expense for purchasing materials; the personnel cost covers the salary, welfare, etc. of employees. Load the four-dimensional dynamic constraint model as a constraint condition to ensure that the total production cost is optimized under the conditions of meeting constraints such as equipment production capacity, material completeness, personnel skills, and order delivery date.

[0037] Design a variable-length chromosome coding mechanism using an improved genetic algorithm, and encode equipment allocation, material consumption, and personnel scheduling as gene fragments. Assume that there are pieces of equipment, types of materials, and employees in the production process. For equipment allocation, it is represented by a sequence with a length equal to the number of production tasks, and each element in the sequence represents the equipment number assigned to that task; material consumption is represented by a matrix related to the production tasks, and the matrix elements represent the types and quantities of materials consumed by each task; personnel scheduling is also represented by a sequence, and each element corresponds to the employee number performing that task. Combine and encode this information into a chromosome, so that each chromosome represents a possible production scheduling plan.

[0038] Inject the local optimal solution of the mixed-integer programming model into the population mutation operation of the genetic algorithm through an alternating iteration mechanism to generate a globally optimized solution set. During the iteration process, first use the mixed-integer programming model to solve the current problem and obtain the local optimal solution. Then, take this local optimal solution as the mutation source and inject it into the population of the genetic algorithm. The mutation operation in the genetic algorithm randomly changes some gene segments of the chromosome, and injecting the local optimal solution of the mixed-integer programming can guide the genetic algorithm to search for the global optimal solution faster. Suppose in the th iteration, the local optimal solution obtained by the mixed-integer programming is , replace some gene segments in with the corresponding segments of some chromosomes in the genetic algorithm population, and then perform operations such as crossover and mutation of the genetic algorithm to generate a new population, continue the iteration, and gradually generate a globally optimized solution set.

[0039] Screen the initial scheduling plan according to the preset convergence threshold. The convergence threshold is a preset parameter used to judge whether the optimization process converges. Let the convergence threshold be , after each iteration, calculate the change rate of the objective function value of the current population. If in consecutive multiple iterations, the change rate of the objective function value is less than , it is considered that the optimization process converges. At this time, select the plan with higher fitness from the current population as the initial scheduling plan. The fitness can be comprehensively calculated according to indicators such as production cost and delivery date achievement rate. For example, the fitness function , where and are weight coefficients, is the total production cost of the current plan, is the maximum allowable production cost, is the delivery date achievement rate of the current plan, is the ideal maximum delivery date achievement rate. By adjusting the weight coefficients and , different optimization objectives can be emphasized according to the actual production requirements.

[0040] Example 3: The specific implementation process of real-time conflict detection through a drag-and-drop Gantt chart is as follows: Parse the operation time series in the scheduling plan to generate a task dependency tree with a hierarchical relationship. First, extract the start time, end time of each operation, and the precedence relationship between operations from the scheduling plan. Taking a simple production process as an example, assume there are operations A, B, and C, where operation B can only start after operation A is completed, and operation C can only start after operation B is completed. Based on these relationships, a directed acyclic graph structure is generated. Then, map the operation nodes to the equipment time axis based on the equipment allocation result to form a spatio-temporal dependency link. Assume operation A is assigned to equipment 1, operation B is assigned to equipment 2, and operation C is assigned to equipment 1. Map these operations to the corresponding equipment time axis in chronological order to clarify the dependency relationship of operations in terms of time and space. Finally, traverse the dependency link using the depth-first search algorithm to construct a task dependency tree with weight relationships. During the traversal process, different weights are assigned to the nodes according to the degree of dependency between operations. For example, if the degree of dependency of operation B on operation A is high, the edge weight between them can be set larger, thus constructing a task dependency tree with weight relationships.

[0041] When listening to the user's drag operation, extract the spatio-temporal coordinates and associated resource identifiers of the adjusted operation. When the user drags an operation on the Gantt chart, the system will capture this operation in real time. By obtaining the position information of the mouse on the Gantt chart and combining it with the coordinate system of the Gantt chart, determine the spatio-temporal coordinates of the adjusted operation. For example, the abscissa of the Gantt chart represents time, and the ordinate represents resources such as equipment or personnel. The specific position of the operation in the time and resource dimensions can be determined through the mouse position. At the same time, extract the resource identifiers associated with this operation from the data structure of the scheduling plan, such as equipment numbers, personnel numbers, etc., for subsequent resource conflict detection.

[0042] Traverse the resource occupancy status of adjacent nodes in the task dependency tree and calculate the time overlap degree and resource conflict coefficient. Let the adjusted operation be , and its adjacent node be . The calculation formula for the time overlap degree is: , where and are the start time and end time of operation respectively, and and are the start time and end time of operation respectively. When , it means that operation and operation overlap in time. The resource conflict coefficient is calculated based on the resource occupancy situations of operation and operation . Assume operation And the process both require the use of equipment . If the resource volume of the equipment is not sufficient to meet the requirements of both processes during the time overlap period, then the resource conflict coefficient will increase accordingly. For example, the maximum available resource volume of equipment within a certain time period is . The resource requirements of process and process for the equipment within this time period are and respectively. The resource conflict coefficient (when ), when , .

[0043] Dynamically label the conflict area through color coding and trigger a conflict warning prompt. According to the calculated time overlap degree and resource conflict coefficient, color code the conflict area in the Gantt chart. For example, when the time overlap degree is high and the resource conflict coefficient is greater than a certain threshold, label the conflict area as red; when the time overlap degree is low or the resource conflict coefficient is small but there is still a possibility of conflict, label it as yellow. At the same time, the system will trigger a conflict warning prompt, reminding the user of resource conflicts through pop-up windows or sounds, etc., so that the user can adjust the scheduling plan in a timely manner.

[0044] Example 4: The verification process of the dynamic recalculation algorithm is as follows: Capture the changed process parameters in the adjusted scheduling plan, extract the local recalculation area, construct a sub-problem solution space based on the four-dimensional dynamic constraint model, and retain the scheduling results of the unadjusted area. When the user adjusts the scheduling plan, the system will automatically detect which process parameters have changed, such as the start time, end time, resource allocation, etc. of the process. Taking a simple adjustment as an example, assume that the user advances the start time of a certain process, then the upstream and downstream processes related to this process and the resources involved may all be affected. Based on these changed parameters, determine the local recalculation area. Within this area, construct a sub-problem solution space based on the four-dimensional dynamic constraint model, that is, determine the feasible scheduling plan of the processes within this area under the constraint conditions of equipment production capacity, material availability, personnel skills, and order delivery date. At the same time, retain the scheduling results of the unadjusted area to avoid unnecessary recalculation of the entire scheduling plan and improve the calculation efficiency.

[0045] The heuristic backtracking algorithm is used to traverse the feasible solutions of sub-problems and perform consistency verification with the global scheduling scheme. The feasible solutions that pass the verification are merged into the global scheduling scheme, and the display data of the visual interaction module is updated. The heuristic backtracking algorithm first defines the resource conflict priority in the local recomputation area. The priority of resource conflicts is determined according to factors such as the importance and scarcity of resources. For example, the resource conflict priority of critical equipment is relatively high because the shutdown of critical equipment may have a greater impact on the entire production process. A backtracking search path is generated, and the invalid solution space is pruned through the constraint propagation algorithm to narrow the backtracking range. The constraint propagation algorithm excludes those solutions that obviously do not meet the constraints during the search process according to the constraint conditions in the four-dimensional dynamic constraint model. For example, if a certain process requires a certain material but the material cannot be supplied within the required time, then the solution containing the arrangement of this process is an invalid solution and can be excluded through the constraint propagation algorithm.

[0046] The memoization search technique is used to cache the states of the traversed solutions, select the optimal local solution according to the verification results, and trigger the update of the global scheduling scheme. The memoization search technique records the relevant information of the solutions that have been traversed. When the same sub-problem is encountered again, the previously cached results can be directly used to avoid repeated calculations. After traversing the feasible solutions of the sub-problems, each feasible solution is subjected to consistency verification with the global scheduling scheme. The consistency verification mainly checks whether there are conflicts between the local solution and the scheduling results of the unadjusted area and whether the constraints of the entire production system are met. If the verification passes, the feasible solution is merged into the global scheduling scheme. Finally, the display data of the visual interaction module is updated so that the Gantt chart can reflect the latest scheduling results in real time, facilitating users to view and further adjust.

[0047] Example 5: The parameter optimization process of the improved genetic algorithm is as follows: The crossover probability and mutation probability are adaptively adjusted according to the complexity of the four-dimensional dynamic constraint model. The complexity of the four-dimensional dynamic constraint model can be measured by the number of constraint conditions, the number of variables, and the complexity of the association between constraints in the model. Let the model complexity index be , when is relatively high, it indicates that the model is relatively complex and the search space is large. At this time, the mutation probability is appropriately increased to increase the diversity of the genetic algorithm during the search process and avoid falling into local optimal solutions. The adjustment formula for the mutation probability can be set as , where is the initial mutation probability, is the initial complexity index, is the maximum complexity index, is the maximum mutation probability. At the same time, the crossover probability is appropriately reduced to prevent the premature loss of excellent genes due to excessive crossover. The adjustment formula for the crossover probability can be set as , where is the initial crossover probability, is the minimum crossover probability. Through this adaptive adjustment method, the genetic algorithm can better adapt to problems with different complexities.

[0048] Design an elitist retention strategy to force the local optimal solutions output by the mixed-integer programming to be retained in the next-generation population. In each iteration process, the mixed-integer programming model will obtain a local optimal solution. This local optimal solution is directly copied into the population of the next-generation genetic algorithm to ensure that these excellent solutions are not lost during the evolution of the genetic algorithm. This can accelerate the convergence speed of the genetic algorithm and improve the optimization efficiency of the algorithm. For example, in a certain iteration, the local optimal solution obtained by the mixed-integer programming corresponds to a lower production cost and a higher on-time delivery rate. Retaining this solution in the next-generation population helps guide the genetic algorithm to find a better global solution faster.

[0049] Fuse the production cost, on-time delivery rate, and resource utilization rate indicators through a dynamic fitness function to optimize the chromosome screening mechanism. Let the production cost be , the on-time delivery rate be , and the resource utilization rate be . The dynamic fitness function can be expressed as: . Among them, , , are weight coefficients, and . These weight coefficients are not fixed, but are dynamically adjusted according to the actual production needs and the focus of different stages.

[0050] During the peak season of market demand, the timeliness of order delivery is crucial for the enterprise's reputation and market share. At this time, the weight of the on-time delivery rate can be appropriately increased. For example, adjust to , and correspondingly reduce the weight of the production cost and the weight of the resource utilization rate, so that the chromosome individuals that pay more attention to on-time order delivery have a greater advantage in screening. On the contrary, during the period when the enterprise has greater cost pressure, the weight of the production cost can be increased, such as setting it to

[0051] to guide the genetic algorithm to preferentially find a scheduling plan with a lower production cost. When calculating the on-time delivery rate , if the actual order completion time is , and the order-specified delivery time is . The resource utilization rate The calculation of needs to comprehensively consider various resources such as equipment, personnel, and materials. Taking equipment resources as an example, assume that the planned usage duration of a certain equipment is , and the actual effective usage duration is . Then the utilization rate of this equipment . By calculating the weighted average of the utilization rates of all equipment, the overall equipment resource utilization rate can be obtained. Similarly, the utilization rates of personnel and material resources are calculated, and finally

[0052] Example 6: This example details a specific implementation method of a production scheduling visualization system based on big data. The system consists of a data collection module, a dynamic modeling module, an optimization engine module, a visualization interaction module, and a real-time verification module.

[0053] The data collection module is responsible for collecting multi-dimensional production data sets, including real-time equipment status data, material inventory data, personnel skill data, and order delivery date data. In terms of collecting real-time equipment status data, sensors such as temperature sensors, pressure sensors, and rotational speed sensors are deployed on various production equipment to monitor the operating parameters of the equipment in real time. The data collected by these sensors is transmitted to the data collection terminal by wired or wireless means, and then the terminal sorts and stores the data in the database. For material inventory data, it is docked with the enterprise's existing inventory management system to obtain the latest material inventory quantity, material completeness index, replenishment cycle, and other information from this system at regular intervals. Personnel skill data is collected by establishing a dedicated employee skill management system. After employees complete various skill trainings and obtain relevant qualification certifications, they enter the information into this system, and the data collection module extracts data from it regularly. Order delivery date data is read from the enterprise's order management system to ensure that the latest and accurate order delivery time requirements are obtained.

[0054] The dynamic modeling module integrates equipment production capacity constraints, material availability constraints, personnel skill constraints, and order delivery date constraints through a dynamic constraint modeling algorithm to generate a four-dimensional dynamic constraint model. When dealing with equipment production capacity constraints, based on real-time equipment status data, the available time window and fault warning information of the equipment are analyzed. For example, if a piece of equipment is expected to require regular maintenance within the next week for two days, then during model construction, the production capacity of this equipment will be set as unavailable for these two days. For material availability constraints, the availability gap is calculated based on material inventory data. If a product requires three materials, A, B, and C, for production, and there is a shortage of material A in the current inventory, then an availability gap will occur. Combining the replenishment cycle to determine the time node for material supply, a dynamic material supply constraint chain is formed. Regarding personnel skill constraints, the process complexity of the order is matched according to personnel skill data. For example, if the production process of an order requires advanced welding skills, search for personnel with this skill in the model and establish the association between skills and tasks. Finally, decompose the order delivery date data into process-level time nodes, fuse them with other constraint factors, and generate a four-dimensional dynamic constraint model that can accurately reflect the actual production constraint conditions.

[0055] The optimization engine module generates an initial scheduling plan using a hybrid integer programming and improved genetic algorithm dual-driven optimization engine based on the four-dimensional dynamic constraint model. First, initialize the objective function of the hybrid integer programming model to minimize the total production cost, including equipment costs, material costs, personnel costs, etc. For example, equipment costs include equipment depreciation expenses, maintenance expenses, etc., material costs cover procurement costs, transportation costs, etc., and personnel costs include employee salaries, benefits, etc. Load the four-dimensional dynamic constraint model as a constraint condition to ensure that the scheduling plan is optimized while meeting various actual production constraints. Design a variable-length chromosome coding mechanism using an improved genetic algorithm, encoding equipment allocation, material consumption, and personnel scheduling as gene segments. In actual operation, for equipment allocation, numbers such as 1, 2, 3, etc. may be used to represent different equipment, and the arrangement of assigning tasks to the corresponding equipment is encoded as part of the chromosome; material consumption is encoded according to the type and quantity of materials; personnel scheduling is also represented by a specific coding to indicate the correspondence between personnel and tasks. Inject the local optimal solution of the hybrid integer programming model into the population mutation operation of the genetic algorithm through an alternating iteration mechanism, continuously optimize the population, and select the initial scheduling plan according to the preset convergence threshold.

[0056] The visualization interaction module displays the scheduling results through a drag-and-drop Gantt chart and real-time detects resource conflicts during adjustment operations. When displaying the scheduling results, the Gantt chart presents the start time, end time, and allocated resources of each production task in an intuitive graphical manner. The abscissa represents time, and the ordinate can be equipment, personnel, or other resource categories. When the user drags and adjusts a task in the Gantt chart, the system immediately activates the real-time conflict detection function. Parse the process time series in the scheduling plan to generate a task dependency tree with a hierarchical relationship. For example, a production process contains multiple processes with a sequential relationship, and a task dependency tree is generated by parsing these relationships. When listening to the user's drag operation, extract the spatio-temporal coordinates of the adjusted process and the associated resource identifier, traverse the resource occupancy status of adjacent nodes in the task dependency tree, and calculate the time overlap and resource conflict coefficient. If two processes occupy the same resource within the same time period and the resource quantity is insufficient to meet the requirements of both, it is determined as a resource conflict. Dynamically label the conflict area through color coding, such as red for serious conflicts and yellow for potential conflicts, and trigger a conflict warning prompt to remind the user to make timely adjustments.

[0057] Based on the adjusted scheduling plan, the real-time verification module verifies the feasibility through a dynamic recalculation algorithm and updates the four-dimensional dynamic constraint model. When the user completes the adjustment of the scheduling plan, the real-time verification module captures the changed process parameters in the adjusted scheduling plan and extracts the local recalculation area. For example, if the user adjusts the start time of a certain process, then the upstream and downstream processes related to this process and the involved resources may be affected, and these affected areas are the local recalculation areas. Construct a sub-problem solution space based on the four-dimensional dynamic constraint model, retain the scheduling results of the unadjusted areas to improve the calculation efficiency. Use a heuristic backtracking algorithm to traverse the feasible solutions of the sub-problems and perform a consistency check with the global scheduling plan. If the check passes, merge the feasible solutions into the global scheduling plan and update the four-dimensional dynamic constraint model at the same time to make it reflect the latest production actual situation and provide a more accurate basis for subsequent scheduling optimization and adjustment.

[0058] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0059] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A production scheduling visualization method based on big data, characterized in that, Including: Collecting a multi-dimensional production data set; the multi-dimensional production data includes equipment real-time status data, material inventory data, personnel skill data, and order delivery date data; the equipment real-time status data includes the equipment available time window and fault warning information, and the material inventory data includes the material completeness index and replenishment cycle; Based on the multi-dimensional production data, integrating equipment production capacity constraints, material completeness constraints, personnel skill constraints, and order delivery date constraints through a dynamic constraint modeling algorithm to generate a four-dimensional dynamic constraint model; According to the four-dimensional dynamic constraint model, adopting a hybrid integer programming and improved genetic algorithm dual-driven optimization engine to generate an initial scheduling plan; Inputting the initial scheduling plan into the visual interaction module, displaying the scheduling result through a drag-and-drop Gantt chart, and real-time detecting resource conflicts during adjustment operations; Based on the adjusted scheduling plan, verifying the feasibility through a dynamic recalculation algorithm and updating the four-dimensional dynamic constraint model.

2. The visualization method of production scheduling based on big data according to claim 1, wherein The process of integrating equipment production capacity constraints, material completeness constraints, personnel skill constraints, and order delivery date constraints through a dynamic constraint modeling algorithm to generate a four-dimensional dynamic constraint model includes: Performing availability segmentation processing on the equipment real-time status data to generate an equipment time occupancy rate curve; Calculating the completeness gap based on the material inventory data, and generating a dynamic material supply constraint chain in combination with the replenishment cycle; Matching the order process complexity according to the personnel skill data to construct a skill-task association matrix; Decomposing the order delivery date data into process-level time nodes, and fusing them with the equipment time occupancy rate curve, the dynamic material supply constraint chain, and the skill-task association matrix to generate a four-dimensional dynamic constraint model.

3. A method for visualizing production scheduling based on big data according to claim 1, characterized in that The construction method of the hybrid integer programming and improved genetic algorithm dual-driven optimization engine includes: Initializing the objective function of the hybrid integer programming model to minimize the total production cost, and loading the four-dimensional dynamic constraint model as a constraint condition; Adopting an improved genetic algorithm to design a variable-length chromosome coding mechanism, and coding equipment allocation, material consumption, and personnel scheduling as gene segments; Injecting the local optimal solution of the hybrid integer programming model into the population mutation operation of the genetic algorithm through an alternating iteration mechanism to generate a globally optimized solution set; Screening the initial scheduling plan according to a preset convergence threshold.

4. A visualization method for production scheduling based on big data according to claim 1, characterized in that, The method for real-time conflict detection through a drag-and-drop Gantt chart includes: Parsing the process time series in the scheduling plan to generate a task dependency tree with a hierarchical relationship; When listening to the user's drag operation, extracting the spatio-temporal coordinates of the adjusted process and the associated resource identifier; Traversing the resource occupancy status of adjacent nodes in the task dependency tree, and calculating the time overlap degree and resource conflict coefficient; Dynamically marking the conflict area through color coding and triggering a conflict warning prompt.

5. A visualization method for production scheduling based on big data according to claim 1, characterized in that, The verification method of the dynamic recalculation algorithm includes: Capturing the changed process parameters in the adjusted scheduling plan, extracting the local recalculation area, constructing a sub-problem solution space based on the four-dimensional dynamic constraint model, and retaining the scheduling results of the unadjusted area; Adopting a heuristic backtracking algorithm to traverse the feasible solutions of the sub-problems, performing consistency verification with the global scheduling plan, merging the feasible solutions that pass the verification into the global scheduling plan, and updating the display data of the visual interaction module.

6. A visualization method for production scheduling based on big data according to claim 2, characterized in that The method for generating the device time occupancy rate curve includes: Predicting the maintenance time window based on the device fault warning information, and dividing the available time period and unavailable time period of the device; Calculating the time occupancy weight of each process for the device by using the time slice rotation algorithm; Fitting the device load balancing curve in combination with historical production data to optimize the time occupancy rate segmentation accuracy.

7. A visualization method for production scheduling based on big data according to claim 3, characterized in that, The parameter optimization method of the improved genetic algorithm includes: Adapting the crossover probability and mutation probability according to the complexity of the four-dimensional dynamic constraint model; Designing an elite retention strategy to force the locally optimal solution output by the mixed integer programming to be retained in the next generation population; Optimizing the chromosome screening mechanism by fusing the production cost, delivery date achievement rate, and resource utilization rate indicators through a dynamic fitness function.

8. A visualization method for production scheduling based on big data according to claim 4, characterized in that The method for constructing the task dependency tree includes: Analyzing the precedence relationship of processes in the order data to generate a directed acyclic graph structure; Mapping the process nodes to the device time axis based on the device allocation result to form a spatio-temporal dependency link; Traversing the dependency link by using the depth-first search algorithm to construct a task dependency tree with weight relationships.

9. A visualization method for production scheduling based on big data according to claim 5, characterized in that The execution steps of the heuristic backtracking algorithm include: Defining the resource conflict priority of the local recomputation area, generating a backtracking search path, pruning the invalid solution space through the constraint propagation algorithm, and narrowing the backtracking range; Using the memoization search technique to cache the states of the traversed solutions, selecting the optimal local solution according to the verification result, and triggering the update of the global scheduling scheme.

10. A production scheduling visualization system based on big data, characterized in that, Including: Data acquisition module: used to acquire a multi-dimensional production data set, and the multi-dimensional production data includes device real-time status data, material inventory data, personnel skill data, and order delivery date data; Dynamic modeling module: configured to integrate device production capacity constraints, material availability constraints, personnel skill constraints, and order delivery date constraints through a dynamic constraint modeling algorithm to generate a four-dimensional dynamic constraint model; Optimization engine module: used to generate an initial scheduling scheme by using a dual-driven optimization engine of mixed integer programming and an improved genetic algorithm according to the four-dimensional dynamic constraint model; Visualization interaction module: displaying the scheduling result through a drag-and-drop Gantt chart and detecting resource conflicts in the adjustment operation in real time; Real-time verification module: verifying the feasibility based on the adjusted scheduling scheme through a dynamic recomputation algorithm and updating the four-dimensional dynamic constraint model.

Images