Intelligent production dynamic scheduling method, device, electronic equipment and storage medium

Abstract

The present application relates to the field of intelligent manufacturing technology, and in particular to an intelligent production dynamic scheduling method, device, electronic device and storage medium. By acquiring production data and determining that there are no anomalies, feature extraction is performed to obtain a production data feature vector, obtain optimization targets and their corresponding weights, and form a dynamic weight matrix; a multi-objective scheduling plan is determined based on the production data feature vector, the optimization target and the dynamic weight matrix, and the multi-objective scheduling plan is optimized to output multiple feasible scheduling plans, and the feasibility of multiple feasible scheduling plans is verified to output the optimal scheduling plan. The present application adapts to dynamic production scenarios through a dynamic weight matrix, balances multi-objective conflicts, and has a local optimization mechanism, which can quickly respond to production changes and improve production scheduling efficiency and accuracy.

Description

Technical Field

[0001] The present application relates to the field of intelligent manufacturing technology, and in particular to an intelligent production dynamic scheduling method, device, electronic equipment and storage medium. Background Art

[0002] In today's intelligent manufacturing landscape, intelligent production scheduling has become a key component in improving production efficiency and optimizing resource allocation. With the increasing diversification of market demand, the dynamic nature of production scenarios, and the rise of flexible, high-mix, small-batch production models, traditional production scheduling methods employ static weighting (e.g., 0.6 for delivery time and 0.3 for line changeover cost) for multi-objective optimization. This approach is unable to adapt to dynamic production scenarios (e.g., urgent order insertions and equipment failures). Furthermore, with fluctuating order priorities and resource constraints, traditional production scheduling methods struggle to balance delivery deadline achievement (typically below 85%) with line changeover costs (which can vary by ±15%). Furthermore, when key production metrics, such as equipment effectiveness (OEE), fluctuate significantly—for example, when OEE fluctuates by more than ±15%—traditional production scheduling methods require a full recalculation, which is time-consuming and increases the risk of plan failure.

[0003] Therefore, for multi-objective optimization problems, traditional production scheduling methods, using fixed weight strategies, are unable to adapt to dynamic production scenarios. When order priorities fluctuate and resource constraints change, it is difficult to effectively balance conflicting objectives such as delivery deadline achievement and line change costs. Traditional heuristic algorithms also lack global optimization capabilities and are prone to falling into local optimal solutions, resulting in low resource utilization. Regarding real-time response issues, traditional systems require long-term full recalculation and lack local optimization mechanisms. They are unable to quickly respond to production changes due to equipment OEE fluctuations or inserted orders. Their lack of real-time data processing capabilities makes it difficult to meet the needs of multimodal data fusion, resulting in a high risk of plan failure and delayed execution on the shop floor. Summary of the Invention

[0004] In order to overcome the deficiencies of the prior art, the present invention provides an intelligent production dynamic scheduling method, device, electronic device and storage medium to improve production scheduling efficiency and accuracy.

[0005] A first aspect of the present application provides a method for intelligent production dynamic scheduling, the method comprising:

[0006] Obtain production data;

[0007] When it is determined that there is no abnormality in the production data, feature extraction is performed on the production data to obtain a production data feature vector;

[0008] Obtaining an optimization target and determining a weight corresponding to the optimization target to obtain a dynamic weight matrix;

[0009] Optimizing a multi-objective production scheduling plan and outputting a plurality of feasible production scheduling plans, wherein the multi-objective production scheduling plan is determined according to the production data feature vector, the optimization target, and the dynamic weight matrix;

[0010] Conduct feasibility verification on the multiple feasible production scheduling plans and output the optimal production scheduling plan.

[0011] In an optional embodiment, the optimizing the multi-objective production scheduling plan and outputting multiple feasible production scheduling plans includes:

[0012] Initializing a population, wherein the population includes a plurality of the multi-objective production scheduling plans;

[0013] Performing non-dominated sorting on each of the multi-objective production scheduling plans in the population to obtain an optimal non-dominated solution set;

[0014] Calculating the congestion degree of the multi-objective production scheduling scheme in the same non-dominated layer to obtain the congestion degree of each solution in the optimal non-dominated solution set;

[0015] Performing population iterative update based on the optimal non-dominated solution set and the congestion degree of each solution;

[0016] When it is determined that the preset iteration termination conditions are met, the Pareto frontier solution set in the last generation population is output and multiple feasible production scheduling plans are obtained.

[0017] In an optional embodiment, the production data includes an order urgency coefficient, and obtaining the production data includes:

[0018] Get the remaining delivery days and remaining working days;

[0019] The order urgency coefficient is determined according to the remaining delivery days and the remaining working days.

[0020] In an optional embodiment, the production data includes resource tension, and obtaining the production data includes:

[0021] Obtain equipment OEE comprehensive efficiency and team load rate;

[0022] The resource tension is determined based on the equipment OEE comprehensive efficiency and the team load rate.

[0023] In an optional embodiment, the optimization objectives include delivery time achievement rate, line change cost, and equipment utilization rate, and determining the weights corresponding to the optimization objectives to obtain a dynamic weight matrix includes:

[0024] Determine a first weight corresponding to the delivery date achievement rate, a second weight corresponding to the line change cost, and a third weight corresponding to the equipment utilization rate;

[0025] Adjusting the first weight according to the urgency of the order, adjusting the second weight according to the actual line change cost, and adjusting the third weight according to production demand;

[0026] A dynamic weight matrix Weights=[ , , ],in is the first weight, is the second weight, is the third weight.

[0027] In an optional embodiment, the method further comprises:

[0028] Acquire workshop capacity data, and perform capacity analysis based on the workshop capacity data to generate a workshop capacity heat map;

[0029] Acquire production task information, and generate a production scheduling Gantt chart based on the production task information;

[0030] When a click instruction from the user on the workshop capacity heat map is received, the production task information of the corresponding time period in the production scheduling Gantt chart is automatically located.

[0031] In an optional embodiment, the method further comprises:

[0032] When receiving an order request submitted by a user, determining order information corresponding to the order request;

[0033] Check whether the corresponding inventory is sufficient according to the order information;

[0034] When the inventory is sufficient, confirm the order and enter the payment process;

[0035] When the payment is confirmed to be successful, a shipping request is generated according to the order information; when the payment is confirmed to be unsuccessful, a payment failure prompt is displayed;

[0036] When the inventory is insufficient, out-of-stock information is generated and returned to the user.

[0037] A second aspect of the present application provides an intelligent production dynamic scheduling device, the device comprising:

[0038] Acquisition module, used to obtain production data;

[0039] a feature extraction module, configured to extract features from the production data to obtain a production data feature vector when it is determined that the production data does not have anomalies;

[0040] A weight allocation module is used to obtain an optimization target and determine the weight corresponding to the optimization target to obtain a dynamic weight matrix;

[0041] an optimization module, configured to optimize a multi-objective production scheduling plan and output a plurality of feasible production scheduling plans, wherein the multi-objective production scheduling plan is determined based on the production data feature vector, the optimization target, and the dynamic weight matrix;

[0042] The verification module is used to verify the feasibility of the multiple feasible production scheduling plans and output the optimal production scheduling plan.

[0043] The third aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the intelligent production dynamic scheduling method when executing the computer program.

[0044] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned intelligent production dynamic scheduling method.

[0045] In summary, the intelligent production dynamic scheduling method, device, electronic device, and storage medium provided in this application have at least one of the following beneficial effects:

[0046] 1. Obtain the optimization objectives and determine their corresponding weights to obtain a dynamic weight matrix. That is, the weights are no longer fixed, but can be dynamically adjusted according to changes in the production scenario. The dynamic weight matrix can better adapt to dynamic production scenarios and balance multi-objective conflicts. 2. Determine a multi-objective production scheduling plan based on the production data feature vector, optimization objectives, and dynamic weight matrix, optimize the multi-objective production scheduling plan, and output multiple feasible production scheduling plans. By comprehensively considering the characteristics of production data, optimization objectives, and dynamic weight matrices, various factors in the production process can be more comprehensively considered, thereby generating multiple feasible production scheduling plans, which helps avoid falling into local optimal solutions and improve resource utilization.

[0047] 3. After acquiring production data, the system first determines whether the data contains anomalies. If not, it then performs feature extraction and subsequently generates and optimizes the production schedule. This data-driven scheduling approach enables the system to make local adjustments and optimizations based on real-time production data, rather than recalculating the entire production plan. For example, when equipment OEE fluctuates, the system can quickly adjust the production schedule based on the new equipment status data without recalculating the entire production plan, thereby improving real-time responsiveness.

[0048] 4. Acquire production data and perform feature extraction. Production data contains information in multiple modes. By extracting features from these multimodal data, we can have a more comprehensive understanding of the production status, thereby generating a more reasonable production scheduling plan, which helps to improve the system's ability to perceive production changes, reduce the risk of plan failure, and reduce workshop execution delays. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a method for intelligent production dynamic scheduling shown in an embodiment of the present application;

[0050] Figure 2 This is a second flow chart of an intelligent production dynamic scheduling method shown in an embodiment of the present application;

[0051] Figure 3 This is a third flow chart of an intelligent production dynamic scheduling method shown in an embodiment of the present application;

[0052] Figure 4 This is a flow chart of an intelligent production scheduling optimization method based on the NSGA-II algorithm shown in an embodiment of the present application;

[0053] Figure 5 This is a flowchart of a visual monitoring and analysis system for production management shown in an embodiment of the present application;

[0054] Figure 6 This is a flowchart of an order processing process shown in an embodiment of the present application;

[0055] Figure 7 This is a functional module diagram of an intelligent production dynamic scheduling device shown in an embodiment of the present application;

[0056] Figure 8 It is a structural diagram of an electronic device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0057] The present invention will be further described below with reference to the accompanying drawings and examples.

[0058] The following will clearly and completely describe the concept, specific structure and technical effects of the present invention in combination with the embodiments and drawings, so as to fully understand the purpose, characteristics and effects of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention. In addition, all the connection / connection relationships involved in the patent do not refer to the direct connection of components, but refer to the fact that a better connection structure can be formed by adding or reducing connection accessories according to the specific implementation situation. The various technical features in the invention can be combined interactively without conflicting with each other.

[0059] Reference Figure 1 1 is a flow chart of an intelligent production dynamic scheduling method according to an embodiment of the present application, and the intelligent production dynamic scheduling method includes the following steps.

[0060] S11, obtaining production data.

[0061] Among them, production data may include order urgency coefficient, resource tension index, and production capacity fluctuation threshold (±15% dynamic range).

[0062] Refer to Figure 2 In some embodiments, for the injection molding scheduling scheme, the operation data (such as spindle speed, temperature, and vibration frequency) is collected in real time through the CNC machine tool, and the CNC machine tool is connected to the electronic device (the edge computing gateway in this application) through the Ethernet / TCP-IP protocol. In addition, the RFID tag is deployed in the material flow link, and the material location and batch information are captured through the reader / writer. The data is uploaded to the acquisition layer after protocol conversion. When the operation data is obtained, the electronic device performs a preliminary screening of the operation data, filters out invalid values ​​(such as sensor transient noise, null values ​​caused by communication packet loss), and only passes valid data to the stream processing engine, where the stream processing engine is a Flink stream processing engine, which processes data streams with a delay of <50ms, supports high-throughput concurrency, and completes data format unification (such as JSON to Protobuf) and timestamp synchronization. Next, a Min-Max scaling algorithm is used to map features such as machine tool temperature (range 50-150°C) and vibration frequency (range 0-10kHz) to the [0,1] interval, eliminating the impact of dimensional differences on subsequent analysis. Equipment fault data is then separated based on clustering algorithms (such as DBSCAN). For example, vibration spectra that deviate from the normal operating cluster are marked as abnormal and stored in a separate log for engineers to review.

[0063] In an optional embodiment, the production data includes an order urgency coefficient, and obtaining the production data includes:

[0064] Get the remaining delivery days and remaining working days;

[0065] The order urgency coefficient is determined according to the remaining delivery days and the remaining working days.

[0066] In some embodiments, the electronic device can obtain the delivery deadline countdown (i.e., the number of days remaining in the delivery period) and the remaining working days, and determine the order urgency coefficient CR based on the dynamic ratio of the remaining delivery days to the remaining working days. Specifically, the order urgency coefficient CR is determined using the following formula:

[0067] CR=remaining days of delivery / remaining working days;

[0068] Among them, the remaining working days = remaining working hours / average working hours per person in the work center, and the remaining working hours = planned working hours - completed working hours. The completed working hours include preparation, processing, movement and queuing time.

[0069] When the order urgency coefficient CR is obtained, the electronic device can determine that when the order urgency coefficient CR is greater than 1, it means there is plenty of time, and the priority adjustment strategy can be set to normal scheduling; when the order urgency coefficient CR is equal to 1, it means a critical state, and the priority adjustment strategy can be set to progress monitoring; when the order urgency coefficient CR is less than 1, it means insufficient time / urgent, and the priority adjustment strategy can be set to priority scheduling resources.

[0070] For example, assuming that the current time is 2025-06-05, the delivery date is 2025-06-10, the remaining working hours are 20 hours, and the average daily working hours of the work center are 8 hours, the electronic device can calculate the remaining working days = 20 / 8 = 2.5 days, and further calculate the urgency coefficient CR = (10-5) / 2.5 = 2.0, indicating that there is plenty of time.

[0071] In an optional embodiment, the production data includes resource tension, and obtaining the production data includes:

[0072] Obtain equipment OEE comprehensive efficiency and team load rate;

[0073] The resource tension is determined based on the equipment OEE comprehensive efficiency and the team load rate.

[0074] At the same time, electronic equipment can also obtain real-time equipment OEE comprehensive efficiency and team load rate. The equipment OEE comprehensive efficiency includes availability, performance rate, and yield rate. The equipment OEE comprehensive efficiency = availability × performance rate × yield rate. The equipment OEE comprehensive efficiency is determined by integrating equipment OEE data through an industrial Internet of Things platform (such as a SCADA system). The team load rate includes actual working hours and theoretical available working hours. The team load rate = actual working hours / theoretical available working hours. The team load rate is obtained by recording team working hours through RFID / barcode scanning and uploaded to the MES system. Next, the resource tension index is determined based on the equipment OEE comprehensive efficiency and team load rate. Specifically, the resource tension is determined using the weighted geometric mean algorithm using the following formula:

[0075] ;

[0076] in, is the weight corresponding to the comprehensive efficiency of equipment OEE, is the weight corresponding to the team load rate, + = 1. For example, the weight of the equipment OEE comprehensive efficiency can be Set to 0.6 (equipment-intensive production line), and set the weight of the team load rate Set to 0.4 (labor-intensive production line).

[0077] Electronic devices can also pre-set threshold ranges to determine whether resource tension is resource-intensive. For example, when the resource tension is greater than 0.8, it can be determined that the current state is resource-intensive and a red alert is issued; when 0.6 ≤ resource tension ≤ 0.8, it can be determined that the current state is normal; and when the resource tension is greater than 0.6, it can be determined that the current state is resource-limited.

[0078] At the same time, electronic equipment can also obtain steady-state production capacity benchmarks , the dynamic range is As the center, the allowed fluctuation range is ±15%, that is, ×0.85≤Δt≤ ×1.15, for example, = 10 seconds, the allowable range is 8.5 seconds to 11.5 seconds. Refers to the average time between outputs of a single piece when the system is in continuous production under ideal conditions (for example: = 10 seconds / piece), based on the steady-state capacity benchmark ±15% determines the capacity fluctuation threshold. Then, through the cache capacity ( and ) is affected by fluctuations in occupancy rate assessment and setting of exceeding threshold classification standards, as shown in Table 1 below.

[0079] Table 1:

[0080]

[0081] In some embodiments, the electronic device can also count the Z-score value of Δt in a continuous production cycle, and trigger a fluctuation warning if it exceeds ±1.5 standard deviations. The steady-state production capacity benchmark can also be recalculated based on the actual production capacity stability according to a preset time period (for example, every 24 hours). Among them, the capacity fluctuation threshold (±15%) is a key indicator for evaluating the stability of the production system. By combining the grading standard with the cache occupancy rate, fluctuation risks can be quickly identified and measures can be taken to ensure production continuity and efficiency.

[0082] S12: When it is determined that there is no abnormality in the production data, feature extraction is performed on the production data to obtain a production data feature vector.

[0083] Refer to Figure 2 In some embodiments, when acquiring production data, the electronic device needs to perform an abnormal data assessment on the production data. Regarding order urgency, if the data is normal, it conforms to historical distribution patterns (e.g., a normal distribution) and fluctuates within a ±15% threshold. Regarding resource intensity, if the data is normal, the equipment's overall efficiency (OEE) is stable between 85% and 95%, and the team load factor is between 60% and 80%. Regarding the capacity fluctuation threshold, if the data is normal, the equipment efficiency fluctuation is within a ±15% dynamic range.

[0084] Furthermore, electronic devices can fuse and process the collected multi-source production data. For this fused data, relevant features are standardized by eliminating dimensional differences to achieve a unified data scale. Furthermore, a 72-hour moving average is calculated for time series data within the production data, such as indicators related to order urgency, to capture cyclical fluctuations in the production data. Based on these processing results, the electronic devices construct a production data feature vector. This feature vector contains at least a standardized feature for order urgency, a feature for resource strain derived from a weighted combination of overall equipment effectiveness (OEE) and load factor, and a feature representing the percentage deviation of current production capacity from a baseline. This feature-calculated production data is then pushed to the MES / ERP system, triggering production scheduling or quality traceability processes.

[0085] To facilitate understanding of the inventive concept of this embodiment, a code example is provided to represent the output production data feature vector:

[0086] # Feature vector example

[0087] feature_vector = {

[0088] "order_urgency": 0.72, # Normalized urgency

[0089] "resource_tension": 0.85, # Weighted composite value of OEE and load rate

[0090] "capacity_deviation": -0.08# The percentage by which the current capacity deviates from the baseline

[0091] In other embodiments, the electronic device can perform outlier detection, that is, use the Z-score method to calculate the standardized deviation of the data point from the mean. If |Z|>3, it is determined to be an outlier and abnormal data exists; at the same time, trend mutation detection is performed, that is, if the data of three consecutive cycles exceeds the ±15% threshold, it is determined to be a trend mutation and abnormal data exists; at the same time, logical conflict detection is performed, that is, if the order urgency is inconsistent with the delivery deadline countdown (such as 1 day left but the urgency is lower than the threshold), it is determined to be a logical anomaly and abnormal data exists; at the same time, equipment status anomaly detection is performed, that is, if the OEE drops sharply by more than 20% or the team load rate is continuously >95%, it is determined to be an equipment overload or failure risk and abnormal data exists.

[0092] Refer to Figure 2When abnormal data is identified, electronic devices can clean the data. Specifically, to clean outliers, electronic devices can examine the data collection process and device operating status to determine whether the outliers are caused by accidental factors such as data collection errors or sudden equipment failures, or whether the data inherently exhibits these extreme conditions. If the outliers are caused by data collection errors or accidental factors and are relatively few in number, they can be directly deleted. For example, if a data point in the production data of an electronic device has an abnormally high value due to a sensor failure, it can be deleted. Alternatively, statistical measures such as the mean and median can be used for replacement. For example, the mean of the data set excluding the outliers can be calculated and used to replace the outliers. If the outliers represent real and meaningful information, such as high performance data achieved by an electronic device under specific conditions, these data should be retained but should be separately processed or annotated in subsequent analysis. To clean out sudden trend changes, electronic devices can review the data collection equipment and recording methods to confirm whether there are any data recording errors or equipment failures. If a sudden trend change is caused by data fluctuations or short-term interference, data can be smoothed using methods such as moving averages and exponential smoothing to reduce the impact of the sudden change. For example, a three-period moving average method can be used to replace the data in the middle period with the average of the data from three adjacent periods. If a sudden trend change represents a real business change, such as a production strategy adjustment or a shift in market demand, the data can be segmented according to the point of change, and each segment can be analyzed and processed separately. If a data recording error is confirmed, the data can be corrected based on the actual situation to align with the normal trend. To clean up logical conflicts, electronic equipment can examine the source and calculation method of data such as order urgency and delivery deadline countdown to determine whether there are data entry or calculation errors. If a data entry error is the cause, the order urgency or delivery deadline countdown data can be corrected based on the correct order information and business rules. For some complex logical conflicts, manual intervention may be required to determine and resolve them based on the business context and actual situation. For example, communication with relevant business personnel can be used to understand the specific circumstances of the order and then make appropriate adjustments to the data. To clean up equipment status anomalies, electronic equipment can review equipment operation and maintenance records to determine whether the abnormal status is caused by equipment failure, improper operation, or other reasons. If data recording errors cause inaccurate equipment status data, correct the relevant data. For example, check the equipment sensor data and correct any incorrect OEE calculations. If data is missing or abnormal due to a serious equipment failure, exclude the data from that period and supplement it based on the data patterns during normal equipment operation or relevant business rules. If the equipment status anomaly is real, perform timely maintenance and repairs. Once the equipment returns to normal operation, update the relevant data to ensure data accuracy and completeness.

[0093] After cleaning abnormal data, the cleaned data also needs to be verified to check whether the cleaning process has introduced new errors and ensure that the data quality meets the analysis requirements.

[0094] Through multiple detection methods, anomalies in production data are identified and cleaned, effectively improving data quality, reducing the interference of abnormal data on subsequent analysis, and providing more accurate and reliable data support for electronic equipment-related business decisions.

[0095] S13, obtaining an optimization target, and determining a weight corresponding to the optimization target to obtain a dynamic weight matrix.

[0096] Refer to Figure 2 In some embodiments, electronic equipment can set phased improvement goals by analyzing the current gaps in equipment availability (reducing downtime), performance rate (actual speed / theoretical speed), and yield rate (percentage of qualified products); and set optimization goals based on historical failure frequency (such as the average number of downtimes per month), downtime cost (loss per hour), and industry benchmarks (such as the automobile manufacturing requirement of a downtime rate ≤ 2%). The optimization goals include at least delivery time achievement rate, line change cost, and equipment utilization rate, and are assigned corresponding weights. Specifically, the first weight is , second weight and the third weight Based on this, a dynamic weight allocation mechanism is constructed to dynamically adjust the weight according to the actual production situation. The specific rules are as follows: Adjust the delivery achievement rate weight according to the urgency of the order When there is an urgent order and the remaining time of the delivery period is within the preset urgent threshold range, the Assignment of the value; determine the line change cost weight according to the actual line change cost Assignment of value; Set equipment utilization weight according to production demand The value of the equipment utilization rate is assigned, and the equipment utilization rate weight is used as a secondary optimization target in a specific production scenario; at the same time, a resource tension threshold is set. When the resource tension exceeds the threshold, the delivery time achievement rate weight is automatically increased. The value of . Among them, the weight distribution satisfies Constraints, construct the dynamic weight matrix Weights=[ , , ].

[0097] For example, in an actual production scenario, when an urgent order occurs and the delivery period is within a preset urgent threshold time (for example, 2 hours), you can set =0.7; when the line change cost is at a medium level, set =0.2; when equipment utilization is the secondary optimization target, set =0.1.

[0098] It should be noted that when the resource tension is greater than the preset resource tension threshold (for example, 0.9), Automatically increase by 0.2.

[0099] S14, optimize the multi-objective production scheduling plan and output multiple feasible production scheduling plans.

[0100] The multi-objective production scheduling scheme is determined based on the production data feature vector, the optimization target and the dynamic weight matrix. Figure 2 and Figure 3 When obtaining the production data feature vector, the electronic device can use a combination of random production and historical injection to generate an initial population based on the production data feature vector. Random generation is to randomly combine equipment, orders, and timing genes within the parameter boundary. Historical injection is to inject the elite solutions in the chromosome reservation area into the initial population at a ratio of approximately 10%, and output an initial population containing N groups of parameter combinations. Each group of parameter combinations corresponds to a production scheduling draft, that is, a multi-objective production scheduling plan (such as minimizing completion time, maximizing equipment utilization, and minimizing energy consumption), so as to further optimize the multi-objective production scheduling plan through NSGA-II.

[0101] In an optional embodiment, the optimizing the multi-objective production scheduling plan and outputting multiple feasible production scheduling plans includes:

[0102] Initializing a population, wherein the population includes a plurality of the multi-objective production scheduling plans;

[0103] Performing non-dominated sorting on each of the multi-objective production scheduling plans in the population to obtain an optimal non-dominated solution set;

[0104] Calculating the congestion degree of the multi-objective production scheduling scheme in the same non-dominated layer to obtain the congestion degree of each solution in the optimal non-dominated solution set;

[0105] Performing population iterative update based on the optimal non-dominated solution set and the congestion degree of each solution;

[0106] When it is determined that the preset iteration termination conditions are met, the Pareto frontier solution set in the last generation population is output and multiple feasible production scheduling plans are obtained.

[0107] Refer to Figure 3 and Figure 4 In some embodiments, when an initial population consisting of multiple multi-objective production scheduling plans is obtained, each multi-objective production scheduling plan is an individual, which is composed of relevant parameters such as equipment, orders, and timing.

[0108] (1) Initialize the population.

[0109] In some embodiments, an electronic device can obtain historically optimal production scheduling solutions (equipment allocation sequence and order priority combination) and construct a chromosome retention zone based on these historically optimal production scheduling solutions. This chromosome retention zone corresponds to the elitism strategy in genetic algorithms. Specifically, during each generation of evolution, the production scheduling solutions in the current population are non-dominated and sorted, the Pareto rank of each individual is determined, and the crowding degree of individuals within the same Pareto rank is calculated. The production scheduling solution with the highest Pareto rank and the highest crowding degree is marked as an elite individual. These elite individuals are excluded from subsequent crossover and mutation operations and directly replace the lowest-quality individuals in the offspring population. The equipment allocation sequence and order priority combination included in these elite individuals are then stored in the chromosome retention zone to serve as the starting point for subsequent initial population optimization. In other words, the solution with the highest Pareto rank and the highest crowding degree in the current population is marked as elite. These elite individuals directly enter the next generation, replacing the lowest-quality individuals in the offspring population. The updated chromosome retention zone and elite solution are then output.

[0110] After the chromosome reserved area is constructed and the production data feature vector (including order urgency, resource tension, and production capacity fluctuation threshold) is obtained, the electronic equipment can randomly generate multiple sets of parameter combinations (such as equipment allocation, process timing, and order priority) based on the production data feature vector, and inject the elite solution (accounting for 10%) in the chromosome reserved area to obtain the initial population of N groups of scheduling drafts, that is, the multi-objective scheduling plan.

[0111] Next, the electronic device determines the parameter ranges involved in each multi-objective scheduling plan in the population, such as the equipment availability range, order delivery time range, and process sequence constraints. Within these parameter bounds, it randomly generates multiple parameter combinations, each of which constitutes a multi-objective scheduling plan. These multi-objective scheduling plans collectively form the initial population. For example, if there are parameters such as equipment selection, order priority, and process start time, the device is randomly assigned to each order, and the priority and process start time are set, forming individual scheduling plans.

[0112] (2) Non-dominated sorting.

[0113] In some embodiments, the electronic device can perform non-dominated sorting on each multi-objective production scheduling scheme in the population to determine their superiority and inferiority relationships in the multi-objective space and obtain an optimal non-dominated solution set. Specifically, an objective function is constructed based on the optimization objective, and the following operations are performed on each multi-objective production scheduling scheme individual in the population: the production scheduling scheme individuals are non-dominated sorted according to the Pareto dominance relationship and a preset objective function; wherein the Pareto dominance relationship determination rule is: if scheme A is not inferior to scheme B in all objectives and is better than at least one objective, then scheme A dominates scheme B; the objective function includes the delivery date achievement rate (with maximization as the goal), the line change cost (with minimization as the goal), and the equipment utilization rate (with maximization as the goal); after sorting, the schemes are layered to obtain the Front 1 optimal non-dominated solution set containing the optimal individuals.

[0114] (3) Calculation of congestion.

[0115] For multi-objective scheduling plans within the same non-dominated layer, the congestion degree is calculated based on the distance between adjacent solutions in the target space. The distance between adjacent solutions in the target space is calculated using the Euclidean distance method in the two-dimensional space of delivery time and cost. After the calculation, a numerical result is output, which represents the sparsity of the solution. The larger the value, the fewer surrounding solutions. Based on the congestion degree calculation result, sparsely distributed solutions are preferentially selected within the same layer to maintain the population diversity of the optimal non-dominated solution set.

[0116] (4) Population iterative update.

[0117] Furthermore, electronic devices can select individuals to enter the next generation population based on the number of non-dominated sorting layers and the crowding degree. Specifically, individuals with a low number of non-dominated layers are usually preferred. If the number of non-dominated layers is the same, individuals with a high crowding degree are selected to maintain the diversity of the population. The selected individuals are subjected to crossover mutation. For example, some parameters of two individuals are exchanged by single-point crossover, multi-point crossover, or uniform crossover to generate new individuals. The newly generated individuals are mutated to increase the diversity of the population. For example, a parameter value of an individual is randomly changed, but it is necessary to ensure that the mutated parameter is still within a reasonable range. The parent population and the child population are merged, and non-dominated sorting and crowding degree calculations are performed. Excellent individuals are selected to form a new population to ensure that excellent genes are retained, that is, steps (2) and (3) are iteratively executed.

[0118] The preset iteration termination condition is at least one of the following: meeting a convergence condition (K consecutive generations, e.g., the improvement in the optimal solution after 50 generations is less than a threshold, such as 0.1%), resource constraints (reaching a maximum number of iterations, such as 1000 generations or a computational time limit), or additional production scenario conditions (providing feasibility verification, such as a resource conflict rate less than 5%). The generated data feature vectors are converted into objective function weights (e.g., urgency influences handover weights), which participate in the fitness evaluation of each individual generation. The final output is the Pareto front (Front 1) solution set for the final generation of the population, known as the Pareto front solution set. The chromosomes are then decoded, and the gene sequence is converted into an equipment allocation table and a process schedule, resulting in multiple feasible production scheduling plans. In other words, multiple feasible production scheduling plans are generated by chromosome decoding: gene sequence → equipment allocation table + process schedule.

[0119] By initializing the population, non-dominated sorting and congestion calculation, three sets of alternative plans are generated in each round of iteration. It should be noted that the output of three sets of feasible production scheduling plans in the embodiment of the present application is based on engineering experience values. The three sets of feasible production scheduling plans can cover the "aggressive-balanced-conservative" strategy to meet different decision-making preferences. For example, Plan 1 prioritizes delivery time (high equipment utilization and low cost); Plan 2 balances multiple objectives (compromises among various indicators); Plan 3 prioritizes cost (low delivery time and high equipment utilization). At the same time, considering the optimization of computational efficiency, the number of plans is limited to avoid the Pareto solution set being too large, thereby reducing the complexity of decision-making; and the three sets of feasible production scheduling plans are usually located in different areas of the Pareto frontier to ensure the coverage of the target space and guarantee diversity.

[0120] Through the above optional implementations, optimized using NSGA-II, an elitist approach is introduced during population initialization to improve the quality of the starting point. Non-dominated sorting and crowding calculations ensure the selection of high-quality and diverse individuals. Iterative population updates preserve the best genes. Ultimately, three feasible scheduling plans, spanning "aggressive," "balanced," and "conservative" strategies, are output. These strategies address diverse decision preferences, ensure target space coverage, and ensure solution diversity. These plans also avoid overly large solution sets, reduce decision complexity, and improve the efficiency and quality of production scheduling.

[0121] Refer to Figure 3 In the chromosome retention area, the dotted box marks the region where 80% of the gene fragments are retained. A built-in conflict detection algorithm is used for anomaly detection. Specifically, electronic devices can set different exception codes (CP01 / MT02 / TL03) for different anomalies. For example, a CP01 exception triggers resource reallocation, automatically migrating conflicting tasks to idle equipment or teams; an MT02 exception dynamically delays the conflicting process and inserts a buffer time; and a TL03 exception forcibly increases the priority of the chromosome gene for the order (moving the gene position forward).

[0122] Differentiating exception types by exception code allows precise identification of issues such as resource conflicts, process conflicts, and order priority issues, enabling targeted measures to improve the accuracy of production scheduling adjustments. Furthermore, conflict detection and corresponding handling mechanisms enable timely resolution of exceptions, preventing them from invalidating production scheduling plans, ensuring smooth production scheduling and enhancing the stability and reliability of the plan.

[0123] S15, performing feasibility verification on the multiple feasible production scheduling plans and outputting the optimal production scheduling plan.

[0124] Refer to Figure 2 and Figure 3 In some embodiments, after obtaining multiple feasible production scheduling plans, the electronic device can perform feasibility verification on the multiple feasible production scheduling plans, including resource conflict verification, delivery feasibility verification, and rule compliance verification. Specifically, the electronic device can use digital twin technology to build a production system simulation model to monitor the equipment load peak in real time. If the load rate of any equipment exceeds a preset threshold (such as 95%) during the scheduling cycle or there is task time overlap, it is determined that there is a resource conflict; at the same time, an LSTM neural network is used to predict the delay probability of each process. If the delay probability of the critical path process exceeds 10% or the overall order delivery on-time rate is less than 90%, the scheduling plan is determined to be infeasible; at the same time, more than 200 process constraints (such as equipment specificity, process priority, and personnel skill matching) are loaded through the rule engine. If the scheduling plan violates any constraint, a rule conflict alarm is triggered. Only when multiple feasible production scheduling plans pass the three verifications, the multi-objective optimization results based on the NSGA-II algorithm, that is, the multi-objective production scheduling plan (such as minimizing completion time, maximizing equipment utilization, and minimizing energy consumption) are used to screen the Pareto optimal solution, that is, the optimal production scheduling plan, and output the optimal production scheduling plan to the production execution system.

[0125] Through feasibility verification operations, we can solve emergencies such as temporary order insertions and equipment failures that are not covered by the algorithm in response to dynamic disturbances; at the same time, we can consider implicit constraints and take into account unquantified rules such as employee skill preferences and VIP order privileges; at the same time, we can support multi-objective trade-offs, allowing decision makers to adjust target priorities according to real-time needs; after feasibility verification, we can finally output the optimal production scheduling plan. Through multi-dimensional feasibility verification, we can effectively solve emergencies such as temporary order insertions and equipment failures under dynamic disturbances, and improve the scientificity and rationality of production scheduling.

[0126] In an optional implementation, the production scheduling plan can be visualized through the manual interaction interface provided by the electronic device, which supports the feasibility verification of the plan after the user drags and drops the adjustment, and provides real-time feedback on resource conflict warnings (red flashing prompts). Figure 3Specifically, through the visual production scheduling interface, users can intuitively adjust the production schedule by dragging and dropping. For example, they can drag and drop a production task for an order from one piece of equipment to another, or adjust the production sequence of an order. After each drag and drop operation, the system automatically records the plan changes before and after the adjustment, preparing for subsequent feasibility verification. When the production schedule is adjusted after a drag and drop operation, the electronic device can check whether the adjusted schedule meets various resource constraints, such as the production capacity of the equipment, the supply of raw materials, and the working hours of the personnel. For example, it can calculate whether the total production time of each piece of equipment within the scheduling cycle exceeds its maximum available time. It also verifies whether all orders can be completed within the specified delivery period. This can be done by calculating the estimated completion time for each order and comparing it with the delivery date. This ensures that production tasks are scheduled according to the correct process sequence. For example, some processes may have sequential requirements, and it is necessary to check whether the adjusted plan violates these process constraints. When the production scheduling plan passes the feasibility verification, the system will mark the plan as "locked". The locked plan will be issued to each production link as the formal production plan. At the same time, unauthorized further modifications are prohibited. The locked plan can be stored in the database, and the relevant verification information and operation logs will be recorded for subsequent traceability and auditing.

[0127] Furthermore, if a resource conflict is detected during feasibility verification, the system immediately flashes red on the production schedule interface to indicate the conflicting resources and related orders. Detailed warning information can also be displayed, detailing the specific conflict type (e.g., equipment overload, raw material shortage), the resources involved, and order information. Warnings can be communicated to relevant personnel through various means, such as sound and pop-up windows, for prompt action. Throughout the production schedule, multiple manual intervention points are set up, allowing authorized personnel to adjust the schedule. These intervention points can be triggered by time (e.g., daily schedule review), events (e.g., receiving an urgent order), or conditions (e.g., resource utilization falling below a certain threshold). During manual intervention, the system provides a wealth of data analysis and visualization tools to help personnel make informed decisions. Once the production schedule is obtained, the electronic device uses graph theory algorithms (e.g., the critical path method) to calculate the critical path within the schedule—the longest path that determines the overall project completion time. Tasks and processes along the critical path are highlighted on the schedule interface using different colors or line styles. Bottleneck processes can be identified by analyzing indicators such as equipment load and process wait times. Bottleneck processes are marked with special icons or colors, and detailed bottleneck analysis reports are provided, including the causes of the bottlenecks, the scope of impact, and possible solutions.

[0128] This optional implementation allows users to drag and drop adjustments to production schedules through a visual interface. The system automatically records and verifies feasibility, including resource constraints, delivery dates, and process sequences. Once verified, the plan is locked and released. Resource conflicts are highlighted with a flashing red warning and detailed information. Multiple manual intervention points can be configured, and analysis tools are provided. Critical paths can also be calculated and bottleneck processes identified and annotated, making scheduling adjustments more intuitive and efficient, enabling timely handling of exceptions and ensuring the rationality and feasibility of production plans.

[0129] In an optional embodiment, the method further comprises:

[0130] Acquire workshop capacity data, and perform capacity analysis based on the workshop capacity data to generate a workshop capacity heat map;

[0131] Acquire production task information, and generate a production scheduling Gantt chart based on the production task information;

[0132] When a click instruction from the user on the workshop capacity heat map is received, the production task information of the corresponding time period in the production scheduling Gantt chart is automatically located.

[0133] Reference Figure 5 In some embodiments, the electronic device provides a scheduling cockpit interactive interface system through a manual interactive interface, including a dual-view linkage mechanism module, an early warning system module, and an equipment topology network model module. For the dual-view linkage mechanism module, the workshop capacity heat map on the left is colored according to the workshop dimension, and the production schedule Gantt chart on the right. Specifically, based on the workshop capacity data, capacity analysis is performed. When the capacity is overloaded by 100%, the corresponding workshop area is colored dark red; when the idle rate is greater than 30%, it is colored green; other states are gradient transition colors, and support drilling down to production line granularity analysis; at the same time, based on the production task information, a production schedule Gantt chart is generated and displayed on the right side of the interface. The timeline of the production schedule Gantt chart supports 72-hour zooming, and the task bars are color-coded, with blue representing normal tasks and purple representing insert tasks. Based on the constructed workshop capacity heat map and production scheduling Gantt chart, a linkage mechanism between the heat map and the Gantt chart is established. When the user clicks on any heat block in the workshop capacity heat map, the production task information of the corresponding time period in the production scheduling Gantt chart is automatically located; when the user double-clicks the task bar in the production scheduling Gantt chart, the mold sharing constraint details pop up.

[0134] The early warning system module displays a real-time scrolling notification in the status bar at the top of the interface regarding conflict types, such as CP01 (resource capacity conflict), MT02 (process timing conflict), TL03 (logic priority conflict), TR04 (tool conflict), MA05 (material conflict), and QC06 (quality compliance conflict). Insufficient CP01 capacity triggers a full-screen red flash. Additionally, a floating Gantt chart task bar displays information on changeover time savings, such as a 30% reduction in changeover time for continuous production of the same product type. Specifically, electronic equipment can trigger dynamic data collection based on different conflict codes, including internal production data and real-time operating condition data. Internal production data may include equipment operation data (OEE (availability × performance × yield rate), failure frequency, maintenance records (real-time collection by sensors), order execution data (process progress, labor consumption, delivery countdown (provided by MES system)), resource status data (team load rate, mold occupancy status, energy consumption (collected through PLC / SCADA)), external environment data (for risk prediction), supply chain data (raw material inventory level, supplier delivery punctuality rate (ERP system integration)), market dynamics data (urgent order requirements, VIP customer priority (crawler crawling external order system)) and policies and regulations (environmental protection production restriction requirements, safety compliance standards (policy) Real-time operating data can include process parameters (temperature, pressure, flow rate (high-precision sensors sample every second)), quality indicators (product rejection rate, defect type (real-time feedback from visual inspection systems)), and evidence of timing conflicts (process intervals, equipment switching records (DAG process path verification)). Similarly, when collecting internal production data and real-time operating data, the collected data can be cleaned by interpolating missing values ​​and removing outliers (3σ principle) to eliminate noise interference; feature extraction can be performed to extract time series features (moving average, lag terms) to capture production capacity fluctuations; and data fusion can be performed to correlate multi-source data based on timestamps to construct a complete production profile.

[0135] For the equipment topology network model module, information such as equipment type, MTBF value and logistics frequency can be obtained to build an equipment topology network model. Equipment is classified by type icon, and the diameter size reflects the MTBF value. Equipment is classified by type icon, including CNC machine tools, AGV, and stereoscopic warehouses. The diameter size reflects the MTBF value, and ≥95% is displayed as a green solid circle; the edge connection rule is that the line width represents the logistics frequency, the thick line represents >50 pieces per hour, the thin line represents <10 pieces, and the dotted line represents the backup path. In addition, when a device fails, the faulty device is automatically marked in red and the topology reconstruction algorithm is started. The backup path is generated within 5 seconds and presented with a green dotted arrow and the shortest hop count prompt; click on any device node to view the real-time OEE curve and maintenance record, and support one-click startup of the backup synchronization program.

[0136] In addition, electronic equipment can also display the workshop layout in three dimensions and support thermal projection of equipment load rate. Red spots on the ground indicate overloaded areas. The right panel of the topology map integrates the equipment health score (0-100 points). If the score is below 60, the automatic repair process will be triggered.

[0137] By presenting a scheduling cockpit interactive interface with dual-view linkage, early warning prompts, equipment topology network model interaction and visualization enhancement functions, it provides users with an intuitive and efficient workshop production scheduling and equipment management view, assisting users in making production decisions and equipment maintenance.

[0138] In an optional embodiment, the method further comprises:

[0139] When receiving an order request submitted by a user, determining order information corresponding to the order request;

[0140] Check whether the corresponding inventory is sufficient according to the order information;

[0141] When the inventory is sufficient, confirm the order and enter the payment process;

[0142] When the payment is confirmed to be successful, a shipping request is generated according to the order information; when the payment is confirmed to be unsuccessful, a payment failure prompt is displayed;

[0143] When the inventory is insufficient, out-of-stock information is generated and returned to the user.

[0144] Reference Figure 6In some embodiments, the intelligent dynamic production scheduling method is applied to the order processing and production scheduling system of an e-commerce platform. When a user submits an order request through the e-commerce platform interface, the system immediately captures the request. The system then parses the order request and extracts key order information, including but not limited to the product name, specifications, quantity, delivery address, and user contact information. For example, a user places an order for 10 electronic products of model A, with the delivery address being a street in a certain district of a certain city and the contact number being 138xxxxxxxx. Based on the extracted order information, the system queries the inventory database to check whether the corresponding product is in sufficient stock. For example, in the aforementioned case where the user purchases 10 electronic products of model A, the system searches the inventory records for the current inventory quantity of model A. If the inventory quantity is greater than or equal to 10, the inventory is determined to be sufficient; if the inventory quantity is less than 10, the inventory is determined to be insufficient. When the inventory is sufficient, the system automatically confirms the order. After confirming the order, the system guides the user into the payment process. The payment process typically supports multiple payment methods, such as online payment (including Alipay, WeChat Pay, etc.) and bank card payment. The system interacts with the corresponding payment platform and waits for the user to complete the payment. When the payment platform reports a successful payment, the system generates a shipping request based on the order information. This shipping request includes detailed product information and the delivery address. This shipping request is then sent to the logistics department or warehouse management system, which arranges the picking, packaging, and shipment of the goods based on the shipping request. If the payment platform reports a payment failure, a prompt will appear on the user's interface, informing the user of the payment failure and prompting them to check their payment information or try another payment method. When inventory is low, the system automatically generates an out-of-stock message. This out-of-stock message clearly informs the user that the purchased item is currently out of stock and provides an estimated delivery time (if available). The system then sends this out-of-stock message back to the user via SMS or in-site messaging, updating them on the current status of their order.

[0145] Compared with traditional production scheduling methods, this application is verified through multiple indicators, such as the processing time of 500 work orders, the approval rate of the insertion plan, and the manual adjustment conflict rate, as shown in Table 2 below.

[0146] Table 2:

[0147]

[0148] This application can be applied to automotive manufacturing (e.g., collaborative scheduling of multiple processes such as stamping, welding, painting, and assembly), electronics and semiconductor manufacturing (e.g., scheduling optimization of hundreds of processes such as wafer lithography, etching, packaging, and testing, and / or resource allocation for multi-variety, small-batch orders (e.g., custom chips) and large-volume standard product production), and mechanical equipment manufacturing (e.g., customized production of large equipment (e.g., motors and turbine blades), and / or optimizing the processing sequence of multi-model, small-batch tasks). To facilitate understanding of the inventive concept of this application, a specific factory example is provided as follows: Assume that there are order information for order A203 and order B815. Order A203 has a quantity of 500, a delivery date of 2025-05-28, and a priority of 1 (highest). Order B815 has a quantity of 200, a delivery date of 2025-05-29, and a priority of 2. Constraints also exist, including equipment reliability (MTBF) and material preparation. Among them, the reliability of the equipment includes CNC03 and CNC07. CNC03 represents 92% (high reliability, suitable for high-priority tasks), while CNC07 represents 88% (lower reliability, requiring careful allocation). The material preparation for order A203 is expected to arrive at 2:00 PM on May 27, 2025. Next, the order is sorted according to priority, resulting in order A203 (priority 1, due on May 28, 2025) and order B815 (priority 2, due on May 29, 2025). According to the equipment allocation principle, high-priority orders are assigned to the highly reliable equipment (CNC03), while low-priority orders are assigned to the less reliable equipment (CNC07). However, risk assessment is required. It is possible to assign order A203 to CNC03 (MTBF = 92%), and order B815 to CNC03 (due to CNC07's lower reliability and B815's closer delivery date, which warrants priority). Then, the orders are split and scheduled. For order A203 (quantity 500), assuming CNC03's single-batch processing capacity is 250 pieces, the order needs to be split into two batches: A203-1: 250 pieces and A203-2: 250 pieces. For order B815 (quantity 200), assuming CNC03's single-batch processing capacity is 200 pieces, no split is required, i.e., B815-1: 200 pieces. The scheduling order is determined as: CNC03: A203-1 → A203-2 → B815-1. Further, time planning is performed, including production time estimation and delivery time verification. Specifically, the production time estimation assumes a 4-hour processing time per batch and a 1-hour mold change time.The start time of A203-1 is 2025-05-27 15:00 (1 hour after the materials arrive) and the end time is 2025-05-27 19:00; the start time of A203-2 is 2025-05-27 20:00 (after the mold change) and the end time is 2025-05-28 00:00; the start time of B815-1 is 2025-05-28 01:00 (after the mold change) and the end time is 2025-05-28 05:00. Through verification at delivery, it can be determined that order A203 is completed at 2025-05-28 00:00, meeting the delivery date; order B815 is completed at 2025-05-28 05:00, meeting the delivery date. Then, risk assessment and adjustment are carried out, including equipment reliability risk and material delay risk. The equipment reliability risk assessment can be specifically expressed as follows: if the MTBF of CNC03 is 92%, the failure probability is low, but it still needs to be monitored; if CNC03 fails, it is necessary to immediately switch to CNC07 and notify relevant personnel. The material delay risk assessment can be specifically expressed as follows: if the material does not arrive on time, the schedule needs to be adjusted, that is, the start time of A203-1 needs to be delayed, and the delivery date of B815 needs to be prioritized (because B815 is closer). The final output production plan is as follows: (1) Task A203-1: The first batch of processing tasks for order A203, with a quantity of 250 pieces; the start time is 15:00 on May 27, 2025 (starting 1 hour after the material arrives), and the end time is 19:00 on May 27, 2025. (2) Task A203-2: The second batch of processing tasks for order A203, with a quantity of 250 pieces; the start time is 20:00 on May 27, 2025 (after the completion of task A203-1, one hour after the mold change), and the end time is 00:00 on May 28, 2025. (3) Task B815-1: The processing task for order B815, with a quantity of 200 pieces; the start time is 01:00 on May 28, 2025 (after the completion of task A203-2, one hour after the mold change), and the end time is 05:00 on May 28, 2025. Ensure that both orders A203 and B815 can be completed before their delivery date and fully utilize the production capacity of equipment CNC03. At the same time, the scheduling also takes into account equipment reliability and material preparation to reduce production risks and avoid the 8-hour downtime loss caused by the failure of CNC07, thereby improving the delivery time achievement rate to 98.7%.

[0149] The following is a code example of a production scheduling plan:

[0150] / / Input data

[0151] {

[0152] "orders": [

[0153] {"id": "A203", "qty": 500, "due_date": "2025-05-28", "priority": 1},

[0154] {"id": "B815", "qty": 200, "due_date": "2025-05-29", "priority": 2}

[0155] ],

[0156] "constraints": {

[0157] "machine_MTBF": {"CNC03": 0.92, "CNC07": 0.88},

[0158] "material_ready": ["A203": "2025-05-27 14:00"]

[0159] }

[0160] }

[0161] / / Output solution

[0162] {

[0163] "schedule": {

[0164] "CNC03": ["A203-1", "A203-2", "B815-1"],

[0165] "CNC07": ["A203-3", "B815-2"]

[0166] },

[0167] "metrics": {

[0168] "due_date_achievement": 98.7%,

[0169] "changeover_cost": 2350

[0170] }

[0171] }

[0172] Reference Figure 7 , which is a functional module diagram of an intelligent production dynamic scheduling device shown in an embodiment of the present application.

[0173] In some embodiments, the intelligent production dynamic scheduling device 70 may include multiple functional modules composed of computer program segments. The computer program of each program segment of the intelligent production dynamic scheduling device 70 may be stored in the memory of an electronic device and executed by at least one processor to perform (see Figure 1 (Description) Intelligent production dynamic scheduling function. Based on the functions performed, it can be divided into multiple functional modules. These functional modules may include: an acquisition module 701, a feature extraction module 702, a weight assignment module 703, an optimization module 704, a verification module 705, and a processing module 706. A module as referred to in this application refers to a series of computer program segments that can be executed by at least one processor and can perform fixed functions, and is stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0174] The acquisition module 701 is used to acquire production data.

[0175] The feature extraction module 702 is configured to extract features from the production data to obtain a production data feature vector when it is determined that the production data does not have anomalies.

[0176] The weight allocation module 703 is used to obtain an optimization target and determine the weight corresponding to the optimization target to obtain a dynamic weight matrix.

[0177] The optimization module 704 is used to optimize the multi-objective production scheduling plan and output multiple feasible production scheduling plans. The multi-objective production scheduling plan is determined based on the production data feature vector, the optimization target and the dynamic weight matrix.

[0178] The verification module 705 is used to verify the feasibility of the multiple feasible production scheduling plans and output the optimal production scheduling plan.

[0179] The optimization module 704 is specifically used to: initialize a population, which includes a plurality of the multi-objective production scheduling plans; perform non-dominated sorting on each of the multi-objective production scheduling plans in the population to obtain an optimal non-dominated solution set; perform congestion calculation on the multi-objective production scheduling plans in the same non-dominated layer to obtain the congestion of each solution in the optimal non-dominated solution set; perform iterative update of the population based on the optimal non-dominated solution set and the congestion of each solution; and when it is determined that the preset iteration termination condition is met, output the Pareto frontier solution set in the last generation population to obtain a plurality of feasible production scheduling plans.

[0180] The acquisition module 701 is specifically configured to: acquire the remaining days of the delivery period and the remaining working days; and determine the order urgency coefficient according to the remaining days of the delivery period and the remaining working days.

[0181] The acquisition module 701 is further specifically configured to: acquire the equipment OEE comprehensive efficiency and the team load rate; and determine the resource tension according to the equipment OEE comprehensive efficiency and the team load rate.

[0182] The weight allocation module 703 is further specifically configured to: determine a first weight corresponding to the delivery achievement rate, a second weight corresponding to the line change cost, and a third weight corresponding to the equipment utilization rate; adjust the first weight according to the urgency of the order, adjust the second weight according to the actual line change cost, and adjust the third weight according to production demand; construct a dynamic weight matrix Weights=[ , , ],in is the first weight, is the second weight, is the third weight.

[0183] The processing module 706 is used to: obtain workshop capacity data, perform capacity analysis based on the workshop capacity data, and generate a workshop capacity heat map; obtain production task information, and generate a production scheduling Gantt chart based on the production task information; and automatically locate the production task information of the corresponding time period in the production scheduling Gantt chart when receiving a user's click instruction on the workshop capacity heat map.

[0184] The processing module 706 is also used to: when receiving an order request submitted by a user, determine the order information corresponding to the order request; check whether the corresponding inventory is sufficient based on the order information; when the inventory is sufficient, confirm the order and enter the payment process; when it is determined that the payment is successful, generate a shipping request based on the order information; when it is determined that the payment fails, prompt the payment failure; when the inventory is insufficient, generate out-of-stock information, and return the out-of-stock information to the user.

[0185] It should be understood that the various variations and specific embodiments of the intelligent production dynamic scheduling method provided in the above embodiments are also applicable to the intelligent production dynamic scheduling device of this embodiment. Through the above detailed description of the intelligent production dynamic scheduling method, those skilled in the art can clearly know the implementation method of the intelligent production dynamic scheduling device in this embodiment. For the sake of brevity of the specification, it will not be described in detail here.

[0186] See Figure 8 FIG. 1 is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. In a preferred embodiment of the present application, the electronic device 8 includes a memory 81 , at least one processor 82 and at least one communication bus 83 .

[0187] Those skilled in the art should understand that Figure 8 The structure of the electronic device shown does not constitute a limitation of the embodiments of the present application, and can be either a bus structure or a star structure. The electronic device 8 can also include more or less other hardware or software than shown in the figure, or a different component arrangement.

[0188] In some embodiments, the electronic device 8 is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits, programmable gate arrays, digital processors, and embedded devices. The electronic device 8 may also include user devices, which include but are not limited to any electronic product capable of human-computer interaction with a user via a keyboard, mouse, remote control, touchpad, or voice-controlled device, such as a personal computer, tablet computer, smartphone, digital camera, etc.

[0189] In the above embodiments provided in the present application, it should be understood that the disclosed methods, devices, computer-readable storage media, and electronic devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple components or modules can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or components or modules, which can be electrical, mechanical or other forms.

[0190] The components described as separate parts may or may not be physically separate, and the components shown as components may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of these components may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0191] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing module, or each component may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or software functional modules.

[0192] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or 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 for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0193] It should be noted that for the aforementioned method embodiments, for ease of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0194] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0195] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. An intelligent production dynamic scheduling method, characterized in that: The method comprises: Obtain production data; When it is determined that there is no abnormality in the production data, feature extraction is performed on the production data to obtain a production data feature vector; Obtain the optimization target and determine the weight corresponding to the optimization target to obtain a dynamic weight matrix; the optimization target includes delivery achievement rate, line change cost and equipment utilization rate; including: adjusting the delivery achievement rate weight according to the urgency of the order When there is an urgent order and the remaining time of the delivery period is within the preset urgent threshold range, the Assignment of the value; determine the line change cost weight according to the actual line change cost Assignment of value; Set equipment utilization weight according to production demand The value of the equipment utilization rate is assigned, and the equipment utilization rate weight is used as a secondary optimization target in a specific production scenario; at the same time, a resource tension threshold is set. When the resource tension exceeds the threshold, the delivery time achievement rate weight is automatically increased. The value of The multi-objective production scheduling scheme is optimized to output multiple feasible production scheduling schemes, wherein the multi-objective production scheduling scheme is determined according to the production data feature vector, the optimization target and the dynamic weight matrix; the method comprises: initializing a population, wherein the population comprises multiple multi-objective production scheduling schemes; performing non-dominated sorting on the production scheduling scheme individuals of the current population according to the Pareto dominance relationship and the preset objective function; wherein the Pareto dominance relationship determination rule is: if scheme A is not inferior to scheme B in all objectives and is better in at least one objective, then scheme A dominates scheme B; the objective function comprises the delivery date achievement rate, the line change cost and the equipment utilization rate, wherein the delivery date achievement rate is aimed at maximizing the value, the line change cost is aimed at minimizing the value, and the equipment utilization rate is aimed at maximizing the value; after sorting, the schemes are layered to obtain a Front containing the optimal individuals.

1. Optimal non-dominated solution set; performing congestion calculation on the multi-objective production scheduling schemes in the same non-dominated layer to obtain the congestion of each solution in the optimal non-dominated solution set; performing population iterative update based on the optimal non-dominated solution set and the congestion of each solution; when it is determined that the preset iteration termination condition is met, outputting the Pareto front solution set in the last generation population to obtain multiple feasible production scheduling schemes; wherein, performing congestion calculation on the multi-objective production scheduling schemes in the same non-dominated layer to obtain the congestion of each solution in the optimal non-dominated solution set includes: for the multi-objective production scheduling schemes in the same non-dominated layer, performing congestion calculation based on the distance between adjacent solutions in the target space, the distance between adjacent solutions in the target space is calculated using the Euclidean distance in the delivery-cost two-dimensional space; outputting a numerical result after calculation, the numerical value represents the sparsity of the solution, and the larger the value, the fewer the surrounding solutions; using the congestion calculation result, preferentially selecting sparsely distributed solutions in the same layer to maintain the population diversity of the optimal non-dominated solution set; The feasibility of the multiple feasible production scheduling plans is verified and the optimal production scheduling plan is output.

2. The intelligent production dynamic scheduling method according to claim 1, characterized in that: The production data includes an order urgency coefficient, and obtaining the production data includes: Get the remaining days of delivery and remaining working days; The order urgency coefficient is determined according to the remaining delivery days and the remaining working days.

3. The intelligent production dynamic scheduling method according to claim 1, characterized in that: The production data includes resource tension, and obtaining the production data includes: Obtain equipment OEE comprehensive efficiency and team load rate; The resource tension is determined based on the equipment OEE comprehensive efficiency and the team load rate.

4. The intelligent production dynamic scheduling method according to any one of claims 1 to 3, characterized in that: The method further comprises: Acquire workshop capacity data, and perform capacity analysis based on the workshop capacity data to generate a workshop capacity heat map; Acquire production task information, and generate a production scheduling Gantt chart based on the production task information; When a click instruction from the user on the workshop capacity heat map is received, the production task information of the corresponding time period in the production scheduling Gantt chart is automatically located.

5. The intelligent production dynamic scheduling method according to any one of claims 1 to 3, characterized in that: The method further comprises: When receiving an order request submitted by a user, determining order information corresponding to the order request; Check whether the corresponding inventory is sufficient according to the order information; When the inventory is sufficient, confirm the order and enter the payment process; When the payment is confirmed to be successful, a shipping request is generated according to the order information; when the payment is confirmed to be unsuccessful, a payment failure prompt is displayed; When the inventory is insufficient, out-of-stock information is generated and returned to the user.

6. An intelligent production dynamic scheduling device, characterized in that: The device comprises: Acquisition module, used to obtain production data; a feature extraction module, configured to extract features from the production data to obtain a feature vector of the production data when it is determined that the production data does not have anomalies; The weight allocation module is used to obtain the optimization target and determine the weight corresponding to the optimization target to obtain a dynamic weight matrix; the optimization target includes the delivery achievement rate, line change cost and equipment utilization rate; including: adjusting the delivery achievement rate weight according to the urgency of the order When there is an urgent order and the remaining time of the delivery period is within the preset urgent threshold range, the Assignment of the value; determine the line change cost weight according to the actual line change cost Assignment of value; Set equipment utilization weight according to production demand The value of the equipment utilization rate is assigned, and the equipment utilization rate weight is used as a secondary optimization target in a specific production scenario; at the same time, a resource tension threshold is set. When the resource tension exceeds the threshold, the delivery time achievement rate weight is automatically increased. The value of The optimization module is used to optimize the multi-objective production scheduling scheme and output multiple feasible production scheduling schemes, wherein the multi-objective production scheduling scheme is determined according to the production data feature vector, the optimization target and the dynamic weight matrix; the module includes: initializing a population, wherein the population includes multiple multi-objective production scheduling schemes; performing non-dominated sorting on the production scheduling scheme individuals of the current population according to the Pareto dominance relationship and the preset objective function; wherein the Pareto dominance relationship judgment rule is: if scheme A is not inferior to scheme B in all objectives and at least one objective is better, then scheme A dominates scheme B; the objective function includes delivery date achievement rate, line change cost and equipment utilization rate, wherein the delivery date achievement rate is aimed at maximizing the value, the line change cost is aimed at minimizing the value, and the equipment utilization rate is aimed at maximizing the value; after sorting, the schemes are layered to obtain the Front containing the optimal individuals.

1. Optimal non-dominated solution set; performing congestion calculation on the multi-objective production scheduling schemes in the same non-dominated layer to obtain the congestion of each solution in the optimal non-dominated solution set; performing population iterative update based on the optimal non-dominated solution set and the congestion of each solution; when it is determined that the preset iteration termination condition is met, outputting the Pareto front solution set in the last generation population to obtain multiple feasible production scheduling schemes; wherein, performing congestion calculation on the multi-objective production scheduling schemes in the same non-dominated layer to obtain the congestion of each solution in the optimal non-dominated solution set includes: for the multi-objective production scheduling schemes in the same non-dominated layer, performing congestion calculation based on the distance between adjacent solutions in the target space, the distance between adjacent solutions in the target space is calculated using the Euclidean distance in the delivery-cost two-dimensional space; outputting a numerical result after calculation, the numerical value represents the sparsity of the solution, and the larger the value, the fewer the surrounding solutions; using the congestion calculation result, preferentially selecting sparsely distributed solutions in the same layer to maintain the population diversity of the optimal non-dominated solution set; The verification module is used to verify the feasibility of the multiple feasible production scheduling plans and output the optimal production scheduling plan.

7. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the intelligent production dynamic scheduling method according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent production dynamic scheduling method according to any one of claims 1 to 5 are implemented.

Images