Task scheduling method based on MES system collaboration

Abstract

The invention discloses a task scheduling method based on MES system collaboration, and belongs to the technical field of task scheduling. The invention relates to a task scheduling method based on MES (Manufacturing Execution System) system collaboration, which is characterized by comprising the following steps of: S1, carrying out production data acquisition and standardization processing; s2, sensing and predicting a production resource state; s3, performing task scheduling constraint modeling; s4, cooperative scheduling calculation; s5, carrying out visualization and explanation generation on the scheduling scheme; s6, task issuing and execution feedback; and S7, historical data accumulation and scheduling strategy optimization. According to the method, the production resource state is predicted through a time sequence prediction model, a Bayesian model and a skill matching model based on the equipment sensor data, the material state and the personnel scheduling information which are acquired in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of task scheduling, in particular to a task scheduling method based on MES system cooperation. BACKGROUND

[0002] In modern manufacturing enterprises, the production execution system (Manufacturing Execution System, MES) is widely used in production planning, scheduling and workshop management. The traditional MES system mainly relies on static scheduling rules and manual experience to allocate production tasks, and usually has the following characteristics: the existing MES system relies on the experience of schedulers and fixed rules when allocating tasks, and lacks dynamic perception and prediction ability for real-time state of the production site. Task priority, process sequence and equipment load usually use static parameters, which cannot be adjusted flexibly according to real-time information on the site. The existing system cannot obtain timely information on the availability of equipment, personnel and materials, and sensor data and inventory information often have delays, resulting in scheduling decisions lagging behind actual production conditions. Key factors such as equipment failure prediction, material shortage prediction and personnel attendance prediction lack quantitative analysis and mathematical modeling. The current MES system mainly considers the constraints of process sequence and equipment capacity in task scheduling, and lacks comprehensive modeling of multi-dimensional constraints such as personnel skill matching, material supply timing, energy consumption limit, etc. Constraint information has not formed a mathematical model that can be directly used for scheduling optimization, and the feasibility and optimality of the scheduling result are difficult to guarantee. The existing method usually uses experience rules or simple heuristic algorithms, and lacks predictive analysis of equipment status, material availability, personnel availability and energy consumption limit. There is a lack of collaborative modeling and constraint fusion among different resources, which is prone to production scheduling problems such as equipment idling, personnel idling, material delay or energy consumption exceeding the limit.

[0003] Therefore, the existing technology still has the following shortcomings: it is not possible to fully utilize equipment sensor data, material status and personnel information for dynamic scheduling; task scheduling constraints do not comprehensively cover factors such as process sequence, equipment capacity, personnel skills, material supply and energy consumption limit; the scheduling process lacks predictability and multi-resource collaboration capability, and cannot achieve optimization of production efficiency and resource utilization. SUMMARY

[0004] The present application aims to provide a task scheduling method based on MES system cooperation to solve the problems raised in the background art.

[0005] A task scheduling method based on MES system cooperation, comprising the following steps:

[0006] S1, production data collection and standardization: collect production data from MES system, ERP system, WMS system and production field sensors, the data including order information, equipment status, personnel status, material inventory and process constraints; convert the data into a standardized format and store it in the production scheduling data buffer;

[0007] S2, production resource state perception and prediction: real-time state analysis and short-term prediction of the standardized data obtained in step S1, the prediction content including device available time, capacity utilization, material arrival time and personnel availability; the prediction result is stored in JSON format and transmitted to the task scheduling engine;

[0008] S3, task scheduling constraint modeling: according to the data of step S1 and step S2, a task scheduling constraint model is established, the constraints including process sequence constraints, equipment capacity constraints, personnel skill matching constraints, material supply constraints and energy consumption limitation constraints; the constraint model is stored in the form of structured file for scheduling calculation;

[0009] S4, collaborative scheduling calculation: based on the constraint model and prediction data of step S3, a dynamic multi-objective optimization algorithm is used to calculate the scheduling scheme, the optimization objectives including shortest production cycle, highest equipment utilization, lowest energy consumption and production load balancing; the calculation process supports bidirectional data interaction between MES system and ERP system, realizing task coordination and conflict resolution across systems; the scheduling result is output in the form of scheduling scheme data set;

[0010] S5, scheduling scheme visualization and explanation generation: the scheduling scheme of step S4 is converted into a visual Gantt chart, resource occupation chart and critical path analysis chart; at the same time, a scheduling scheme explanation text is generated, the explanation content including the calculation basis of scheduling priority, the scheduling reason of key tasks and the solution of resource conflict;

[0011] S6, task issuing and execution feedback: the scheduling scheme of step S4 is issued to the production execution layer through the MES system, including device task sheet, personnel task sheet and material release instruction; collect execution feedback information, including task completion time, abnormal events and resource change, and update it to the scheduling engine in real time, realizing closed-loop scheduling optimization;

[0012] S7, historical data accumulation and scheduling strategy optimization: store the execution feedback information collected in step S6 into the historical scheduling database, analyze the historical scheduling data, optimize the scheduling algorithm parameters and constraint weights, and realize long-term adaptive optimization of scheduling strategy.

[0013] Preferably, the production resource state prediction in step S2 includes the following substeps:

[0014] S21, predicting the availability of the equipment in the next period based on the real-time collected equipment sensor data and using a time series prediction model;

[0015] S22, predicting the material arrival time and possible material shortage risk based on the production order and the in-process state;

[0016] S23, predicting the available personnel and skill matching degree based on the personnel scheduling table and the skill database;

[0017] S24, packaging the above prediction results into a resource prediction data packet and transmitting the resource prediction data packet to the scheduling engine.

[0018] Preferably, the task scheduling constraint model construction in step S3 comprises the following sub-steps:

[0019] S31, establishing a process sequence constraint matrix to specify the preceding and following relationship between processes;

[0020] S32, establishing an equipment capacity constraint table to record the processing capacity, processing precision and maximum load of each equipment;

[0021] S33, establishing a personnel skill matching table to map the task demand skills and the personnel skill levels;

[0022] S34, establishing a material supply constraint table to record the material inventory, arrival time and batch information;

[0023] S35, establishing an energy consumption limit constraint to specify the maximum energy consumption threshold in different time periods.

[0024] Preferably, the collaborative scheduling calculation in step S4 adopts a dynamic multi-objective optimization algorithm, which is a hybrid algorithm combining genetic algorithm and reinforcement learning, and can quickly adjust the scheduling scheme when the real-time production data changes and ensure the stability of the global optimal solution.

[0025] Preferably, the scheduling scheme visualization in step S5 comprises synchronously displaying the task plan on the MES system client and the mobile terminal, so that the workshop operators can real-time view the task execution sequence, resource occupation and estimated completion time.

[0026] Preferably, the execution feedback collection in step S6 is realized through a production data collection interface (PDC interface) of the MES system, and the interface supports second-level data collection and event-triggered reporting.

[0027] Preferably, the scheduling strategy optimization in step S7 adopts a historical data playback and simulation verification method to compare the effects of different scheduling algorithms and select the scheduling strategy with the best performance as the default scheme in the next period.

[0028] Compared with the prior art, the advantages of the present application are that:

[0029] (1) The present application is based on real-time collected equipment sensor data, material state and personnel scheduling information, and predicts the production resource state through time series prediction model, Bayesian model and skill matching model. Compared with the traditional static rule scheduling, the present application can dynamically adjust task allocation, realize real-time response to equipment availability, material arrival and personnel skill matching, and significantly reduce the risk of production delay.

[0030] (2) The present application constructs a constraint model including process sequence, equipment capacity, personnel skill, material supply and energy consumption limit, to ensure that the scheduling scheme is generated under the condition of meeting all constraints. The comprehensive modeling of multi-dimensional constraints helps to avoid the problems of equipment idle, material delay, personnel idle and energy consumption exceeding in traditional scheduling, and improves the resource utilization.

[0031] (3) The present application predicts the availability of equipment, material arrival time, personnel attendance and skill matching degree in the future period, so that the scheduling engine can adjust the task sequence and resource allocation in advance to realize predictive scheduling. The coordination constraints among tasks, equipment, personnel and materials are fully reflected, and the overall coordination and stability of the production system are improved.

[0032] (4) The present application uses a formulaic constraint model and clear resource prediction index, so that the scheduling scheme can be directly used for optimization algorithm calculation to ensure the executability of the generated scheme. The scheduling result can be quantitatively evaluated, including equipment utilization, task completion rate, production cycle time, material shortage risk and energy consumption level, to realize the optimization of production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The flowchart of the method of the present application. DETAILED DESCRIPTION

[0034] Embodiment: A task scheduling method based on MES system coordination, comprising the following steps:

[0035] S1, production data acquisition and standardized processing: acquiring production data from MES system, ERP system, WMS system and production field sensor, the data including order information, equipment state, personnel state, material inventory and process constraints; converting the data into standardized format and storing to production scheduling data buffer area;

[0036] S2, production resource state sensing and prediction: performing real-time state analysis and short-term prediction on the standardized data obtained in step S1, the prediction content including equipment available time, capacity utilization, material arrival time and personnel availability; the prediction result is stored in JSON format and transmitted to the task scheduling engine;

[0037] S3, task scheduling constraint modeling: according to the data of step S1 and step S2, a task scheduling constraint model is established, and the constraints include process sequence constraints, equipment capacity constraints, personnel skill matching constraints, material supply constraints and energy consumption limitation constraints; the constraint model is stored in the form of a structured file, and is used for scheduling calculation;

[0038] S4, collaborative scheduling calculation: based on the constraint model and the prediction data of step S3, a dynamic multi-objective optimization algorithm is used to calculate a scheduling scheme, and the optimization objectives include the shortest production cycle, the highest equipment utilization, the lowest energy consumption and the balanced production load; during the calculation process, bidirectional data interaction between the MES system and the ERP system is supported, task coordination and conflict resolution across systems are realized; the scheduling result is output in the form of a scheduling scheme data set;

[0039] 1. Data preparation: from the constraint model and the prediction data of S3, production order information (product type, quantity, delivery time), equipment state information (availability, load, maintenance plan), energy consumption data (unit production task energy consumption), raw material inventory and logistics state are obtained; standardization processing is performed on the data, and time unit, production batch number, etc. are unified.

[0040] Initialization of scheduling optimization model: define scheduling decision variables: task allocation of each device in each time period.

[0041] 2. Define multi-objective function: minimize total production cycle T total = max (task completion time);

[0042] Maximize equipment utilization

[0043] Minimize energy consumption E total = ∑ i E i ;

[0044] Balance load

[0045] 3. Set constraint conditions: equipment capacity constraints (single time period task number, parallel ability); task priority constraints (order delivery period, process flow sequence); material availability constraints; equipment maintenance and resource conflict constraints fed back by the MES / ERP system.

[0046] Scheduling algorithm running: a dynamic multi-objective optimization algorithm (such as NSGA-II, MOEA / D, particle swarm optimization + Pareto trade-off) is used to iteratively search for an optimal scheduling scheme.

[0047] 4. In each iteration: Calculate objective function values based on current solution; Update non-dominated solution set according to Pareto dominance relationship; Punish or correct solutions that violate constraints; Support two-way interaction between MES and ERP system: MES provides real-time equipment status and execution feedback; ERP provides material plan, inventory and order priority update; Algorithm real-time corrects scheduling scheme to ensure cross-system coordination.

[0048] 5. Scheduling scheme output: Generate scheduling scheme dataset from optimization results, including: task list and start / end time for each device, expected energy consumption and device load, production cycle, completion rate index, optimization objective weight and Pareto frontier solution information, write dataset to MES system for execution control; feedback to ERP for order update and resource adjustment if necessary.

[0049] 6. Result verification and iteration: Perform feasibility verification on the generated scheduling scheme: whether it meets all constraint conditions, if conflicts or infeasible solutions are found, make local adjustments and re-iterate optimization, support periodic or event-triggered (such as equipment failure, urgent orders) dynamic rescheduling.

[0050] S5, Scheduling scheme visualization and explanation generation: Convert the scheduling scheme of step S4 into visual Gantt chart, resource occupation chart and critical path analysis chart; At the same time, generate scheduling scheme explanation text, explanation content includes calculation basis of scheduling priority, scheduling reason of key task and solution method of resource conflict;

[0051] S6, Task assignment and execution feedback: Assign the scheduling scheme of step S4 to the production execution layer through the MES system, including device task list, personnel task list and material issue instruction; Collect execution feedback information, including task completion time, abnormal events and resource change, and update it to the scheduling engine in real time to realize closed-loop scheduling optimization;

[0052] S7, Historical data accumulation and scheduling strategy optimization: Store the execution feedback information collected in step S6 into the historical scheduling database, analyze the historical scheduling data, optimize the scheduling algorithm parameters and constraint weights, and realize long-term adaptive optimization of scheduling strategy.

[0053] The production resource state prediction in step S2 includes the following sub-steps:

[0054] S21, Based on real-time collected device sensor data, predict availability in the next cycle:

[0055] First, denoise and smooth the data:

[0056]

[0057] Wherein, Smoothed equipment health indicator value at time t; z t : Sensor raw measurement (e.g. vibration RMS, current, temperature); K t : Kalman gain, automatically updated according to measurement noise and process noise;

[0058] Then predict the remaining life:

[0059]

[0060] Where, y t+1 : Predicted future one-period equipment availability (0~1); f LSTM : Long short-term memory network prediction function; n: input historical time steps

[0061] S22, based on production order and work-in-process status, predict material arrival time and possible material shortage risk:

[0062] First, estimate the arrival time calculation:

[0063]

[0064] Where, T arrival : Estimated arrival time; T now : Current time; D remain : Remaining transportation distance of material (obtained by GPS); V avg : Average transportation speed (historical statistical value); Material shortage risk index (regression prediction + normalization):

[0065] R shortage = σ (a · ΔT + β · S stock ) R shortage : Material shortage risk (0~1); σ (·): Sigmoid function, compressing the value to the interval 0~1; ΔT = T arrival - T need : Arrival delay time; S stock : Current inventory status (negative value indicates gap); a, b: regression model coefficients (fitted by historical data)

[0066] S23, personnel availability and skill matching prediction:

[0067] Staff on-site probability (Bayesian model):

[0068]

[0069] Where, P present (i): The probability of the i-th staff on-site; P(present | shift i ): On-site probability under the condition of this shift (historical statistics); P(shifti P(present): The probability of assignment for this shift; P(present): The overall probability of attendance.

[0070] Skill matching degree calculation:

[0071]

[0072] Where M(i,j): the skill matching degree between employee i and task j (0~1); s ik : Employee i's score on skill dimension k; r jk Task j requires skill dimension k; K: Total number of skill dimensions (e.g., welding, programming, inspection, etc.);

[0073] S24 resource prediction data packet generation:

[0074] The above prediction results are packaged together:

[0075] Packet={EquipAvail,MatETA,ShortRisk,StaffAvail,SkillMatch}

[0076] Wherein, EquipAvail: an array of device future lifecycle availability [y t+1 ]; MatETA: Array of estimated material arrival times [T arrival ShortRisk: Shortage Risk Array [R] shortage StaffAvail: Array of employee attendance probabilities [P] present [(i)]; SkillMatch: Skill matching matrix [M(i,j)]; The data packet is transmitted to the task scheduling engine in JSON format to provide input for the constraints of the subsequent scheduling model.

[0077] Step S3, the construction of the task scheduling constraint model, includes the following sub-steps:

[0078] S31, Process sequence constraint matrix:

[0079] Establish constraints on the sequential relationships between processes to ensure that production tasks are executed in accordance with the process requirements:

[0080]

[0081] Among them, C proc (i,j): Elements of the process sequence constraint matrix; i,j: Process numbers; when C proc When (i,j)=1, the scheduling engine must ensure that StartTime(j)≥EndTime(i);

[0082] S32, Equipment Capacity Constraint Table

[0083] Record the equipment's processing capacity, accuracy, and maximum load to constrain the feasibility of the task on the equipment:

[0084] Processing capacity constraints

[0085]

[0086] Where i: task number; d: device number; device load constraint;

[0087]

[0088] Among them, L d (t): Load of device d at time t; Tasks d (t): The set of tasks processed on device d at time t; w i : The load required for task i; Maximum allowable load of device d;

[0089] Machining accuracy constraints

[0090]

[0091] Among them, P d (i): The actual machining accuracy of equipment d for task i; The machining accuracy requirements for task i;

[0092] S33, Personnel Skills Matching Table:

[0093] Constraints on task and personnel skill matching:

[0094]

[0095] Where i: task number; j: personnel number; in scheduling optimization, if M ij =0, the scheduling engine prohibits assigning task i to person j;

[0096] Further quantifiable skill matching degree (0~1):

[0097]

[0098] Among them, S ij : Skill matching degree between task i and personnel j; s jk Personnel j's ability score in skill dimension k; r ik Task i requires skill dimension k; K: Total number of skill dimensions;

[0099] S34, Material Supply Constraints Table:

[0100] Ensure that the materials required for the task are in place before processing:

[0101] T start (i)≥T arrival (m)

[0102] Among them, T start (i): Estimated start time of task i; T arrival (m): The estimated arrival time of material m required for task i;

[0103] Simultaneously calculate the material shortage risk index:

[0104]

[0105] Among them, R shortage (i): Material shortage risk for task i (0-1); R shortage (m): Material shortage risk of material m;

[0106] S35. Energy Constraints:

[0107] Ensure that total production energy consumption does not exceed the threshold during different time periods:

[0108]

[0109] Where E(t): total energy consumption during time period t; Tasks(t): set of tasks executed during time period t; e i E: Estimated energy consumption for task i max (t): The maximum allowable energy consumption during time period t.

[0110] The collaborative scheduling calculation in step S4 adopts a dynamic multi-objective optimization algorithm. The algorithm is a hybrid algorithm that combines genetic algorithm and reinforcement learning. It can quickly adjust the scheduling scheme when real-time production data changes and ensure the stability of the global optimal solution.

[0111] The visualization of the scheduling scheme in step S5 includes synchronously displaying the task plan on the MES system client and mobile terminal, enabling workshop operators to view the task execution sequence, resource usage, and estimated completion time in real time.

[0112] The execution feedback collection in step S6 is achieved through the production data collection interface (PDC interface) of the MES system, which supports second-level data collection and event-triggered reporting.

[0113] Step S7 optimizes the scheduling strategy by using historical data playback and simulation verification to compare the effects of different scheduling algorithms and select the best-performing scheduling strategy as the default scheme for the next cycle.

[0114] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A task scheduling method based on MES system collaboration, characterized in that, Includes the following steps: S1. Production Data Acquisition and Standardization Processing: Collect production data from the MES system, ERP system, WMS system, and sensors on the production site. The data includes order information, equipment status, personnel status, material inventory, and process constraints. Convert the data into a standardized format and store it in the production scheduling data cache. S2. Production Resource Status Perception and Prediction: Perform real-time status analysis and short-term prediction on the standardized data obtained in step S1. The prediction content includes equipment availability time, capacity utilization rate, material arrival time, and personnel availability. The prediction results are stored in JSON format and transmitted to the task scheduling engine. S3. Task scheduling constraint modeling: Based on the data from steps S1 and S2, a task scheduling constraint model is established. The constraints include process sequence constraints, equipment capacity constraints, personnel skill matching constraints, material supply constraints, and energy consumption limit constraints. This constraint model is stored in the form of a structured file for scheduling calculation. S4. Cooperative scheduling calculation: Based on the constraint model and prediction data in step S3, a dynamic multi-objective optimization algorithm is used to calculate the scheduling scheme. The optimization objectives include the shortest production cycle, the highest equipment utilization rate, the lowest energy consumption, and production load balance. During the calculation process, bidirectional data interaction between the MES system and the ERP system is supported to realize cross-system task coordination and conflict resolution; the scheduling results are output in the form of a scheduling scheme dataset. S5. Visualization and Explanation Generation of Scheduling Scheme: The scheduling scheme from step S4 is transformed into a visual Gantt chart, resource occupancy chart, and critical path analysis chart; at the same time, an explanation text of the scheduling scheme is generated, which includes the basis for calculating the scheduling priority, the reasons for scheduling critical tasks, and the methods for resolving resource conflicts. S6. Task Issuance and Execution Feedback: The scheduling plan from step S4 is issued to the production execution layer through the MES system, including equipment task orders, personnel task orders, and material issuance instructions; execution feedback information is collected, including task completion time, abnormal events, and resource changes, and updated to the scheduling engine in real time to achieve closed-loop scheduling optimization. S7. Historical Data Accumulation and Scheduling Strategy Optimization: Store the execution feedback information collected in step S6 into the historical scheduling database, analyze the historical scheduling data, optimize the scheduling algorithm parameters and constraint weights, and achieve long-term adaptive optimization of the scheduling strategy.

2. The task scheduling method based on MES system collaboration according to claim 1, characterized in that: Step S2, which involves predicting the state of production resources, includes the following sub-steps: S21. Based on real-time collected device sensor data, use a time series prediction model to predict the availability of the device in the next cycle; S22. Based on production orders and work-in-process status, predict material arrival time and potential material shortage risks; S23. Based on the personnel schedule and skills database, predict the availability of personnel and the degree of skill matching; S24. The above prediction results are packaged into a resource prediction data packet and transmitted to the scheduling engine.

3. The task scheduling method based on MES system collaboration according to claim 1, characterized in that: Step S3, the construction of the task scheduling constraint model, includes the following sub-steps: S31. Establish a process sequence constraint matrix to define the sequential relationships between processes; S32. Establish an equipment capacity constraint table to record the processing capacity, processing accuracy, and maximum load of each piece of equipment; S33. Establish a personnel skill matching table to map the skills required for the task to the personnel skill levels; S34. Establish a material supply constraint table to record material inventory, delivery time and batch information; S35. Establish energy consumption limits and specify the maximum energy consumption threshold for different time periods.

4. The task scheduling method based on MES system collaboration according to claim 1, characterized in that: The collaborative scheduling calculation in step S4 adopts a dynamic multi-objective optimization algorithm. The algorithm is a hybrid algorithm that combines genetic algorithm and reinforcement learning. It can quickly adjust the scheduling scheme when real-time production data changes and ensure the stability of the global optimal solution.

5. A task scheduling method based on MES system collaboration according to claim 1, characterized in that: The visualization of the scheduling scheme in step S5 includes synchronously displaying the task plan on the MES system client and mobile terminal, enabling workshop operators to view the task execution sequence, resource usage, and estimated completion time in real time.

6. The task scheduling method based on MES system collaboration according to claim 1, characterized in that: The execution feedback collection in step S6 is achieved through the production data collection interface (PDC interface) of the MES system, which supports second-level data collection and event-triggered reporting.

7. A task scheduling method based on MES system collaboration according to claim 1, characterized in that: Step S7 optimizes the scheduling strategy by using historical data playback and simulation verification to compare the effects of different scheduling algorithms and select the best-performing scheduling strategy as the default scheme for the next cycle.

Images