An MES intelligent scheduling method and system for batch production

Abstract

The present application provides an MES intelligent scheduling method and system for batch production, which relates to the field of production management technology, including: storing a first equipment topology relationship for batch production; detecting the operating status of each device according to the equipment topology relationship to obtain a production anomaly index; when the production anomaly index is greater than a preset threshold, updating the first edge weight of the first equipment topology relationship by gradient descent to output a second edge weight, and obtaining a second equipment topology relationship; performing a scheduling update according to the second equipment topology relationship to obtain an updated scheduling plan. The present application solves the technical problem in the prior art that the faulty equipment cannot be effectively removed in the event of a production anomaly due to the solidification of the equipment collaborative relationship, thereby affecting production stability. It achieves the technical effect of reducing the collaborative connection of the faulty equipment by dynamically adjusting the equipment topology relationship, avoiding the spread of anomalies, and thereby improving production stability and production scheduling flexibility.

Description

Technical Field

[0001] The present application relates to the technical field of production management, and in particular to an MES intelligent scheduling method and system for batch production. Background Art

[0002] In modern manufacturing, MES scheduling management for batch production is a critical link in ensuring production efficiency and product quality. Traditional MES scheduling management methods primarily rely on fixed equipment collaboration relationships, optimizing production processes through preset production plans and equipment scheduling strategies. When a device fails or experiences an anomaly, these methods adjust the schedule by reallocating tasks or manually intervening. However, due to the rigidity of equipment collaboration relationships, when an anomaly occurs, the faulty device may still maintain a strong collaborative connection with other devices and cannot be promptly removed from the production process. This not only impacts the execution of the production plan but also causes the anomaly to spread throughout the production line, affecting the stability of the entire production system, leading to reduced production capacity or production interruptions. Summary of the Invention

[0003] The present application provides an MES intelligent scheduling method and system for batch production, which solves the technical problem in the prior art that the faulty equipment cannot be effectively removed in the event of production anomalies due to the solidification of the equipment coordination relationship, thereby affecting production stability. It achieves the technical effect of reducing the coordinated connection of faulty equipment by dynamically adjusting the equipment topology relationship, avoiding the spread of anomalies, and thereby improving production stability and production scheduling flexibility.

[0004] In view of the above problems, on the one hand, the present application provides an MES intelligent scheduling method for batch production, which includes: storing a first equipment topology relationship for batch production, the edge of the first equipment topology relationship includes a first edge weight that identifies the collaboration strength between the devices; detecting the operating status of each device according to the equipment topology relationship to obtain a production anomaly index; when the production anomaly index is greater than a preset threshold, updating the first edge weight of the first equipment topology relationship by gradient descent to output a second edge weight, and obtaining a second equipment topology relationship corresponding to the second edge weight, wherein the second edge weight is less than the first edge weight; connecting to the MES scheduling execution engine, updating the schedule according to the second equipment topology relationship, and obtaining an updated scheduling plan.

[0005] On the other hand, the present application also provides an MES intelligent scheduling system for batch production, the system including: an equipment topology module for storing a first equipment topology relationship for batch production, the edge of the first equipment topology relationship including a first edge weight that identifies the collaborative strength between the devices; an operation status detection module for detecting the operation status of each device according to the equipment topology relationship to obtain a production abnormality index; an edge weight adjustment module for updating the first edge weight of the first equipment topology relationship by gradient descent and outputting a second edge weight when the production abnormality index is greater than a preset threshold, and obtaining a second equipment topology relationship corresponding to the second edge weight, wherein the second edge weight is less than the first edge weight; a scheduling update module for connecting to the MES scheduling execution engine, performing scheduling updates according to the second equipment topology relationship, and obtaining an updated scheduling plan.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By storing a collaborative relationship network between equipment used in batch production, where edge weights represent the strength of collaboration between equipment, this provides a foundation for subsequent scheduling optimization. By quantifying the degree of collaboration between equipment, scheduling can be adjusted based on actual production needs. By monitoring equipment operating status in real time, obtaining production anomaly indicators, and determining whether any anomalies exist in the current production process, this provides data support for scheduling adjustments, enabling timely detection of production anomalies and the implementation of corresponding optimization strategies. A gradient descent algorithm is used to dynamically adjust equipment collaboration relationships. When a production anomaly exceeds a preset threshold, the collaboration weight between the faulty equipment and other equipment is reduced, thereby minimizing its impact on the overall schedule. This allows the faulty equipment to be separated, enabling the production system to adapt to the anomaly and prevent the spread of the fault's impact. The MES scheduling plan is adjusted using the updated topology to adapt it to the current production status, ensuring the execution effect of the optimized schedule. Through intelligent adaptive adjustments, production stability and efficiency are improved.

[0008] In summary, this application realizes the intelligent optimization of MES scheduling by constructing equipment topology relationships and introducing a dynamic adjustment mechanism for collaborative weights. During the production process, the equipment status is detected in real time. When an anomaly is found, the collaborative relationship between the equipment is dynamically adjusted through the gradient descent algorithm to generate a new equipment topology relationship, which is then applied to the MES scheduling execution engine to generate an updated scheduling plan. This can not only quickly remove faulty equipment and avoid the spread of anomalies, but also dynamically adapt to equipment changes, improve the flexibility of batch production scheduling, and the stability and fault tolerance of the production system, and significantly improve the stability and efficiency of the production process.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flowchart of an MES intelligent scheduling method for batch production provided in an embodiment of the present application.

[0011] Figure 2 A flowchart of updating the first edge weight of a first device topology relationship by gradient descent and outputting the second edge weight in an MES intelligent scheduling method for batch production provided in an embodiment of the present application.

[0012] Figure 3 A schematic diagram of a flow chart of scheduling updates according to a second device topology relationship in an MES intelligent scheduling method for batch production provided in an embodiment of the present application.

[0013] Figure 4 A schematic diagram of the structure of an MES intelligent scheduling system for batch production provided in an embodiment of the present application.

[0014] Description of reference numerals: device topology module 10 , operation status detection module 20 , edge weight adjustment module 30 , schedule update module 40 . DETAILED DESCRIPTION

[0015] The embodiments of the present application provide an MES intelligent scheduling method and system for batch production, which solves the technical problem in the prior art that, due to the solidification of equipment collaborative relationships, faulty equipment cannot be effectively removed in the event of production anomalies, thereby affecting production stability. The present application achieves the technical effect of reducing the collaborative connection of faulty equipment by dynamically adjusting the equipment topology relationship, avoiding the spread of anomalies, and thereby improving production stability and production scheduling flexibility.

[0016] Example 1, as Figure 1 As shown, an embodiment of the present application provides an MES intelligent scheduling method for batch production, the method comprising:

[0017] Step S1: storing a first device topology relationship for batch production, wherein the edges of the first device topology relationship include first edge weights that identify the collaboration strength between devices.

[0018] Specifically, a device topology describes the connections and interactions between devices, represented by a graph data structure. Devices are nodes, and the relationships between them are edges. Edge weights are numerical attributes of edges in the first device topology, representing the strength of collaboration between devices. Higher values ​​indicate stronger collaboration.

[0019] In mass production, by analyzing the layout and functions of an enterprise's production equipment, the connectivity and synergy strength between the equipment are determined, and the first equipment topology is established. For example, in an automobile manufacturing line, there is a certain degree of synergy between stamping equipment, welding equipment, painting equipment, and assembly equipment. By analyzing the process flows and synergy requirements of these equipment, a device topology diagram is constructed, with production equipment as nodes and the connections between equipment as edges. The edge weights represent the synergy strength between the equipment. For example, stamping equipment and welding equipment require close coordination during the production process, so the synergy strength between them is high, resulting in a high edge weight (e.g., 0.8). On the other hand, painting equipment and assembly equipment have relatively independent processes, so the synergy strength between them is low, resulting in a lower edge weight (e.g., 0.3).

[0020] The first equipment topology relationship clarifies the initial collaborative relationship and strength between equipment, providing a basic framework for subsequent production management, facilitating subsequent adjustments and optimizations based on actual production conditions.

[0021] Step S2: detecting the operating status of each device according to the device topology relationship to obtain production abnormality indicators.

[0022] Specifically, production anomaly indicators are used to quantify abnormalities in the production process, such as equipment failure rate, production errors, and downtime. Based on the topological relationships of each device, sensors (such as temperature sensors, current sensors, and vibration sensors) and a monitoring system (such as a data acquisition system or a monitoring module built into the MES system) monitor the operating status of each device in real time. For example, in an electronics manufacturing production line, sensors can monitor parameters such as temperature, vibration, and current on production equipment, while the monitoring system can capture equipment operation logs and fault information. Based on this equipment operating status data, production anomaly indicators such as equipment failure rate and production efficiency reduction are calculated. For example, if the failure rate of a particular device exceeds a preset threshold, this anomaly is recorded and the corresponding production anomaly indicator is calculated.

[0023] By detecting the operating status of the equipment in real time and obtaining production anomaly indicators, abnormal situations in the production process can be discovered in a timely manner, providing trigger conditions for subsequent abnormal processing, helping to quickly respond to abnormalities in the production process and avoid further deterioration of abnormal situations, thereby improving production stability.

[0024] Step S3: When the production anomaly index is greater than a preset threshold, the first edge weight of the first device topology relationship is updated by gradient descent to output a second edge weight, and a second device topology relationship corresponding to the second edge weight is obtained, wherein the second edge weight is less than the first edge weight.

[0025] Specifically, when a production anomaly indicator exceeds a preset threshold, a gradient descent algorithm is initiated to update the edge weights in the first device topology, reducing the synergy between the faulty device and other devices. A second edge weight is output, and a new device topology, i.e., the second device topology, is generated based on the second edge weight. The preset threshold can be determined based on previous production experience or data analysis. The second edge weight is smaller than the first edge weight to reduce the synergy between the faulty device and its upstream and downstream devices. For example, if device B fails, the original topology is: A→B (0.8), B→C (0.6). At this point, the gradient descent algorithm is applied to reduce B's synergy weight, for example: A→B (from 0.8 to 0.3), and B→C (from 0.6 to 0.2).

[0026] By dynamically adjusting edge weights, changing the collaborative relationship between devices, and adjusting the device topology relationship from a state based on the first edge weight to a state based on the second edge weight, a preliminary response to production anomalies is achieved, making the collaborative relationship between devices more adaptable to abnormal situations and developing in a direction that can reduce the impact of anomalies.

[0027] Step S4: connecting to the MES scheduling execution engine, updating the schedule according to the second device topology relationship, and obtaining an updated scheduling plan.

[0028] Specifically, the MES scheduling execution engine is the scheduling module within the MES (Manufacturing Execution System) and is responsible for generating and executing production schedules based on production plans and equipment status. The updated equipment topology (the second equipment topology) is passed to the MES scheduling execution engine. Based on the new equipment coordination relationships, the MES scheduling execution engine uses optimization algorithms (such as genetic algorithms and simulated annealing) to optimize the original schedule and generate a new one. For example, if a machining center in a machinery manufacturing production line experiences a malfunction, the updated equipment topology is passed to the MES scheduling execution engine. The engine adjusts the production plan based on the new equipment coordination relationships, reallocating the affected production tasks to other equipment and generating a new schedule. For example, after obtaining the second equipment topology, the MES scheduling execution engine recalculates the production plan to ensure production continuity. For example, the original task was: B was responsible for executing 100 pieces, and C was responsible for executing 80 pieces. Due to B's failure, the tasks are adjusted to: A is responsible for executing 50 pieces, and C is responsible for executing 130 pieces.

[0029] By applying the updated equipment topology relationship to the MES scheduling execution engine, an optimized scheduling plan can be generated, which improves the flexibility of the production plan, helps to quickly adjust the production plan during the production process, avoids production interruptions caused by equipment failure, and thus improves production stability.

[0030] Furthermore, obtaining the second device topology relationship corresponding to the second edge weight in step S3 further includes:

[0031] When the production anomaly index is greater than the preset threshold, determine whether the equipment in the first equipment topology relationship has changed; if the equipment in the first equipment topology relationship has not changed, configure the first equipment topology relationship based on the second edge weight to obtain a second equipment topology relationship.

[0032] Specifically, when the production anomaly indicator exceeds a preset threshold and the second equipment topology is determined, the system first determines whether any equipment in the first equipment topology has changed. This determines whether the update of the equipment topology should take into account the impact of the equipment change. The system queries the current equipment list in the MES database and compares it with the first equipment topology to determine whether any equipment has been added, removed, or replaced. For example, the equipment management system maintains a device registry that records basic information about each device (such as device number, model, and installation time). When the production anomaly indicator exceeds a preset threshold, the current equipment status (data such as performance and functionality obtained from the equipment status monitoring system) is compared with the initial information in the device registry. If the number, model, and key performance indicators of the equipment remain unchanged, the system can be considered to have not changed. Alternatively, manual intervention, such as an operator manually entering equipment change information into the system, can be used to determine whether any equipment has changed.

[0033] If the devices in the first device topology relationship remain unchanged, the first device topology relationship can be directly configured based on the second edge weight. The edge weights in the first device topology relationship are replaced with the second edge weights to obtain the second device topology relationship. For example, in an automobile manufacturing production line, if the synergy between stamping equipment and welding equipment needs to be reduced, the edge weight between the two can be reduced from 0.8 to 0.5, thereby obtaining a new device topology relationship, which is the second device topology relationship.

[0034] By adding a judgment on whether the equipment has changed in step S3, the equipment topology relationship can be updated more accurately, ensuring that the obtained second equipment topology relationship is more consistent with the actual production situation, and improving the accuracy and effectiveness of production exception processing.

[0035] Furthermore, determining whether a device in the first device topology relationship has changed further includes:

[0036] If the device in the first device topology relationship changes, obtain the change type, including adding a device, replacing a device, and deleting a device; replace the first device topology relationship according to the change type, obtain the changed first device topology relationship, and configure the changed first device topology relationship based on the second edge weight to obtain the second device topology relationship.

[0037] Specifically, if the devices in the first device topology relationship change, the first device topology relationship needs to be adjusted on a larger scale. First, obtain the change type, including adding equipment, replacing equipment, and deleting equipment. Adding equipment refers to adding new equipment to the original first device topology relationship, such as production line expansion; replacing equipment refers to replacing an original equipment with a new equipment, such as equipment upgrade; deleting equipment is to remove a faulty equipment from the first device topology relationship, or eliminate it due to production adjustments. Obtain the change type by querying the equipment operation record through the MES database. The equipment operation record contains operation information such as equipment procurement, installation, and disassembly. If there is a record of installation of a new device, it can be judged as adding equipment; if there is a record of disassembly of equipment and installation of a new device and the functions of the old and new equipment are similar, it can be judged as replacing equipment; if there is only a record of disassembly of the equipment, it is judged as deleting the equipment.

[0038] According to different change types, the first device topology relationship is replaced and changed accordingly to obtain the changed first device topology relationship. Specifically: for adding devices, it is necessary to add new nodes in the device topology relationship, and add corresponding edges and weights based on the collaborative relationship between the new device and other devices; for replacing devices, it is necessary to delete the nodes and related edges of the old device in the device topology relationship, add the nodes of the new device, and add corresponding edges and weights based on the collaborative relationship between the new device and other devices; for deleting devices, it is necessary to delete the nodes and related edges of the device in the device topology relationship. After obtaining the changed first device topology relationship, the changed device topology relationship is configured based on the second edge weight to obtain the second device topology relationship.

[0039] By determining whether a device has changed, further obtaining the change type and making corresponding replacement changes to the device topology relationship, it is possible to handle device changes more flexibly and accurately, ensuring the real-time and accuracy of the device topology relationship. In the event of production anomalies, scheduling updates can be made based on the latest device topology relationship, thereby improving the adaptability of production plans and the stability of the production process.

[0040] Further, such as Figure 2 As shown, in step S3, the first edge weight of the first device topology relationship is updated by gradient descent to output the second edge weight, including:

[0041] Step S31: setting constraints, which include device availability constraints and connectivity constraints.

[0042] Step S32: Perform forward propagation update on the first edge weight of the first device topology relationship under the conditions of the availability constraint and the connectivity constraint, calculate the connectivity loss in real time, take minimizing the connectivity loss as the gradient descent target, and obtain the second edge weight that reaches the descent target.

[0043] Specifically, constraints are restrictions that must be met during the adjustment of the first device topology, including device availability constraints and connectivity constraints. Device availability constraints ensure that devices can operate normally during scheduling and are not unavailable due to reasons such as device failure or maintenance. Connectivity constraints ensure that connections between devices remain connected and are not interrupted due to adjustments to edge weights. Device availability constraints, such as failure rate ranges and maintenance cycles, are determined by analyzing historical device operating data and technical specifications. For example, for a lathe with high machining precision requirements, its weekly downtime must not exceed eight hours based on historical failure records and maintenance manuals. This is the device availability constraint. In practice, these constraints can be stored as variables in the code or as configuration parameters in a database. Connectivity constraints are set by analyzing the interactions between production processes and devices. For example, in an electronics assembly line, a device connection diagram can be drawn to clearly define the required connections between devices in each assembly link, such as the material transfer channel between circuit board soldering equipment and component mounting equipment. Process modeling tools (such as BizagiModeler) can be used to assist in analyzing and determining connectivity constraints, and then store these constraints in the form of logical expressions or rules.

[0044] Under the constraints, the edge weights in the first device topology relationship are updated through forward propagation. Specifically, based on the device availability and connectivity constraints, the first edge weight is updated, and the connectivity loss is calculated in real time. Due to device failure or weight adjustment, the production link may be broken, resulting in the inability to complete the production task. Connectivity loss is used to measure the degree of connectivity destruction in the device topology relationship. The smaller the connectivity loss, the better the connectivity between the devices. The connectivity loss can be quantified based on the actual connection between the devices (for example, whether there is a disruption in material or information transmission). If there is no connection between two devices due to the adjustment of the edge weight, the connectivity loss is calculated according to a pre-set loss calculation method (such as assigning a certain value based on the degree of impact on production). For example, the impact of topology adjustment can be measured by defining a loss function L: ,in, is the edge weight of the current topology, represents ideal connectivity. Using a gradient descent algorithm, aiming to minimize the connectivity loss, edge weights are continuously adjusted until the descent target is reached, thereby obtaining the second edge weight. The gradient descent algorithm calculates the gradient of the connectivity loss function with respect to the edge weights and then adjusts the edge weights in the opposite direction of the gradient. In implementation, optimization libraries (such as the optimize module in the Python Scikit-learn library) can be used to execute the gradient descent algorithm until the preset descent target is reached, thereby obtaining the second edge weight.

[0045] By setting device availability and connectivity constraints, we ensure that devices are adjusted under normal production conditions when updating edge weights. This prevents devices from malfunctioning or losing effective connectivity due to irrational edge weight updates, ensuring production feasibility. Gradient descent, with the goal of minimizing connectivity loss, to obtain the second edge weight optimizes the collaborative relationship between devices, making connections between them more reasonable. This improves the efficiency of collaborative work between devices while meeting production needs and reduces production delays or resource waste caused by poor connectivity.

[0046] Furthermore, step S3 further includes:

[0047] Step S34: hierarchically classify the first device topology relationship according to the subordinate relationship, and output a multi-layer device topology relationship.

[0048] Step S35: Identify a plurality of first edge weights corresponding to the multi-layer device topology relationship.

[0049] Step S36: performing gradient descent to update the plurality of first edge weights respectively to output a plurality of second edge weights, obtaining a multi-layer device topology relationship corresponding to the plurality of second edge weights, and generating a second device topology relationship.

[0050] Specifically, production equipment typically has different levels of control and collaboration. For example, at the top level (scheduling level), the MES system is responsible for global scheduling; at the middle level (production unit level), equipment in each production unit collaborates with each other; and at the bottom level (equipment level), specific equipment performs tasks. When a production anomaly occurs, not only must the edge weights of the equipment topology be adjusted, but the equipment must also be hierarchically classified according to their subordinate relationships, allowing for hierarchical adjustments and optimization of the topological structures at different levels.

[0051] In device topology, a dependency relationship represents a hierarchical dependency between devices. For example, in an automobile manufacturing plant, engine assembly equipment is subordinate to the vehicle assembly line, and certain subcomponent assembly equipment within the engine assembly equipment is in turn subordinate to the engine assembly equipment. This dependency relationship reflects the different hierarchical levels and functional positioning of equipment within the overall production system. Dependent relationships are determined based on factors such as the function of the equipment, the sequence in the production process, and the management ownership of the equipment. The first device topology is hierarchically graded according to the dependency relationship. This hierarchical hierarchy of device topology relationships is represented using a tree structure or layered diagram, and multi-layered device topology relationships are output to ensure that production coordination at different levels is not disrupted during device topology adjustments.

[0052] Next, each connection relationship in the multi-layer device topology is traversed to identify multiple first edge weights at each level. Each identified first edge weight is updated separately using the previously mentioned gradient descent method (the gradient descent method under constraints). For example, in an electronics manufacturing production line, the device topology is divided into three levels based on subordination: the raw material processing equipment level, the assembly equipment level, and the test and packaging equipment level. The edge weights at each level represent the strength of collaboration between the devices. Gradient descent is performed on the edge weights at each level to minimize the connectivity loss at each level, ultimately obtaining multiple second edge weights in the multi-layer device topology. Based on these second edge weights, the connection relationships between devices are reconstructed, and the data structure storing the device topology relationship (such as a relevant table in a database or the node and edge attributes in a graph database) is updated to generate a second device topology relationship.

[0053] By stratifying the layers according to their dependencies and updating the edge weights at each level separately, we can more precisely manage the production relationships between devices. Devices at different levels play different roles and are more important in the production process. This hierarchical approach better reflects the complex collaborative relationships between devices, enabling the production system to adapt more quickly and effectively to production anomalies, thereby improving the stability and efficiency of the entire production system.

[0054] Furthermore, step S2 includes:

[0055] Step S21: obtaining a historical healthy operation data set under multiple historical working conditions of each device in the device topology relationship, wherein the historical healthy operation data set at least includes a vibration spectrum, a temperature gradient, and a current waveform.

[0056] Step S22: Calculate the standard deviation of the historical healthy operation data set using the isolation forest algorithm to obtain a standard deviation index, compare the standard deviation index with a preset standard deviation threshold, and when it is greater than the preset standard deviation threshold, generate a production abnormality index based on the difference between the preset standard deviation threshold and the standard deviation index.

[0057] Specifically, to identify production anomaly indicators, we first need to collect a historical health data set for each device in the device topology. This data is a collection of operational data collected over a period of time while the device was operating normally. This data includes the device's vibration spectrum, temperature gradient, and current waveform. The vibration spectrum reflects the operating status of the device's internal components, the temperature gradient reflects the temperature distribution across the device, and the current waveform reflects the device's load and operating status.

[0058] Next, the isolation forest algorithm is used to calculate the standard deviation of these historical healthy operation data sets and determine the standard deviation index. This standard deviation index is a statistical indicator that measures the degree of data dispersion. A larger standard deviation indicates greater fluctuations in the equipment operating parameters and a greater degree of deviation from normal operation. The isolation forest algorithm constructs multiple decision trees by randomly selecting features and split values, thus forming a forest. The degree of isolation of each data point in the forest can be measured by the number of steps required to reach the isolated point. For the healthy operation data set of the equipment, the more isolated the data point, the more likely it is to be an anomaly. For example, in the historical healthy operation data set of the motor equipment, if a vibration spectrum data point has a high degree of isolation, this may mean that the motor is experiencing abnormal vibration at that moment, possibly due to bearing wear or other faults.

[0059] After calculating the standard deviation index using the isolation forest algorithm, it is compared with a preset standard deviation threshold. If the standard deviation index exceeds the preset threshold, it indicates that the equipment's operating parameters are fluctuating significantly, indicating an anomaly. A production anomaly index is then generated based on the difference between the preset standard deviation threshold and the standard deviation index. For example, if the preset standard deviation threshold is 0.5 and the calculated standard deviation index is 0.7, the production anomaly index could be 0.2 (0.7 - 0.5). The larger this difference, the greater the degree of equipment anomaly. In actual implementation, the isolation forest algorithm can be implemented using programming languages ​​such as MATLAB or Python. For example, the isolationForest function in MATLAB can be used to quickly construct an isolation forest model and analyze equipment health data sets. Database technologies such as MySQL or MongoDB can also be used to store historical equipment health data sets, as well as the calculated standard deviation index and production anomaly indicators.

[0060] By analyzing the historical healthy operation data set under multiple historical operating conditions of the equipment, using the isolation forest algorithm to calculate the standard deviation index and generate the production anomaly index, potential anomalies can be detected before the equipment has obvious failures, and the degree to which the equipment operating data deviates from the normal range can be quantified, so as to more intuitively understand the severity of the equipment's anomaly, thereby adjusting the production plan more targetedly, reducing equipment downtime, and improving production stability.

[0061] Furthermore, after generating the production abnormality indicator, it also includes:

[0062] The fault level is determined based on the production abnormality indicator, and the fault level includes a first-level fault and a second-level fault; the first-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a first attenuation gradient, and the second-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a second attenuation gradient, and output the second edge weight, wherein the attenuation gradient is the percentage of the weight decaying to the original value.

[0063] Specifically, after the production anomaly index is generated, the fault level is further determined based on the numerical value of the index. Fault levels are generally divided into level one and level two. For example, a level one fault is when the production anomaly index exceeds a higher preset threshold, indicating that the equipment abnormality is more serious and may have a greater impact on the production process; a level two fault is when the production anomaly index exceeds another lower threshold, but has not yet reached the level of a level one fault, indicating that the equipment abnormality is relatively mild. For example, for a certain type of processing equipment, if the production anomaly index is between 0.1-0.3, it is determined to be a level two fault, and when the production anomaly index is greater than 0.3, it is determined to be a level one fault.

[0064] According to different fault levels, different attenuation gradients are used to update the first edge weight by gradient descent. Among them, the attenuation gradient is the percentage of the weight decaying to the original value. When it is determined to be a level one fault, the first edge weight of the topological relationship of the first device is updated by gradient descent according to the first attenuation gradient. For a level two fault, it is updated according to the second attenuation gradient. For example, according to the level one and level two faults in the above examples, the first attenuation gradient used for the level one fault is larger, which means that when adjusting the collaborative strength between devices, the first edge weight will be reduced to a greater extent, thereby removing the faulty device more quickly and avoiding further spread of the abnormal situation; while the second attenuation gradient used for the level two fault is relatively small, which will relatively gently adjust the collaborative relationship between devices to adapt to minor abnormalities of the devices.

[0065] For example, the first attenuation gradient is a (0<a<1), and the first edge weight of a device is w1, then the updated edge weight w2=w1×a. In actual operation, programming can be used. For example, in Python, by traversing the edge weights in the device topology relationship, the update is performed according to the above formula. For secondary faults, the update is performed according to the second attenuation gradient. Suppose the second attenuation gradient is b (0<b<1). Similarly, for the edge weight w1, the updated edge weight w2=w1×b. During the update process, it is necessary to ensure that after the edge weight is updated, the device topology relationship still meets the previously mentioned conditions such as the device availability constraints and connectivity constraints. It can be verified after the update. If it is not satisfied, the update strategy needs to be adjusted or the attenuation gradient needs to be recalculated.

[0066] By determining the fault level based on production anomaly indicators and updating edge weights according to different attenuation gradients, different response strategies can be adopted for faults of different severities. While responding to faults, the rationality of the equipment coordination relationship can be maintained, so that the collaboration between equipment can still be as efficient as possible in the event of a fault, reducing the significant decline in production efficiency caused by faults.

[0067] Further, such as Figure 3 As shown, step S4 includes:

[0068] Step S41: constructing a dual-objective optimization scheduling function, wherein the dual-objective optimization scheduling function includes an objective function of minimizing delay time and an objective function of maximizing equipment utilization.

[0069] Step S42: Connect to the MES scheduling execution engine and create a scheduling task template.

[0070] Step S43: Receive the second equipment topology relationship and the execution work order information from the MES data interface layer through the API interface of the MES scheduling execution engine, call the dual-objective optimization scheduling function to update the schedule, and obtain an updated scheduling plan.

[0071] Specifically, the dual-objective optimization scheduling function is a function that comprehensively considers two objectives and is used to coordinate the two objectives simultaneously during the scheduling process to find a balance point. Among them, the optimization objectives include minimizing delay time and maximizing equipment utilization. The objective function of minimizing delay time aims to reduce the delay of tasks in the production process, such as the delay of product delivery time. The objective function of minimizing delay time can be constructed based on factors such as the planned start time, planned end time, actual start time and actual end time of the production task. For example, let t p−start is the planned start time, t p−end is the planned end time, t a−start is the actual start time, t a−endis the actual end time, then the delay time D=max(0,t a−end −t p−end ). The goal is to minimize D by adjusting the order of production tasks, equipment allocation, etc.

[0072] The objective function of maximizing equipment utilization focuses on how to make full use of equipment resources, reduce equipment idle time, and improve the efficiency of equipment in the production process. Equipment utilization can be calculated by the ratio of the actual working time of the equipment to the equipment's available working time. Let T a−work is the actual working time of the equipment, T a−total is the working time of the equipment, and the equipment utilization rate U=T a−work / T a−total The goal is to maximize U by adjusting the execution order and timing of tasks on devices to reduce device idle time. A weighted approach is used to treat delay time as a negative effect (e.g., delay penalty) and device utilization as a positive effect (e.g., device utilization reward). By assigning different weights to the two objective functions, a dual-objective optimization scheduling function is constructed.

[0073] The MES scheduling execution engine communicates with the engine through its API. The API receives execution order information and secondary equipment topology data from the MES data interface layer. For example, within the MES system, a RESTful API is used to transmit JSON-formatted data via HTTP, delivering secondary equipment topology and execution order information to the MES scheduling execution engine. The structure and content of the scheduling task template are also determined based on factors such as the production task type and equipment characteristics. A scheduling task template is a predefined task framework that specifies the basic structure and requirements of a scheduled task. Templates can be defined using data formats such as XML (Extensible Markup Language) or JSON (JavaScript Object Notation). For example, a template might include basic elements such as the task's start and end time, required equipment resources, and task priority. Creating scheduling task templates improves the efficiency and standardization of scheduled task creation.

[0074] The MES scheduling execution engine receives the second equipment topology relationship and execution work order information through the API interface. The execution work order information contains specific information about the production task, such as the product, quantity, priority, deadline, etc. After receiving the data, the MES scheduling execution engine calls the dual-objective optimization scheduling function for calculation. Based on the execution work order information and the second equipment topology relationship, this function calculates a scheduling plan that minimizes delay time and maximizes equipment utilization, and obtains an updated scheduling plan. For example, in an electronics manufacturing production line, based on the delivery date of the product order and the availability of the equipment, the scheduling function will optimize the allocation of production tasks to minimize delay time while making the best possible use of the production equipment.

[0075] By building a dual-objective optimization scheduling function and calling the MES scheduling execution engine, the scheduling plan can be automatically updated while taking into account changes in equipment topology and production task requirements, achieving the optimal balance between production efficiency and equipment utilization. This avoids problems such as low production efficiency or waste of resources caused by focusing only on a single goal, making production plans more reasonable and efficient.

[0076] Furthermore, the method for outputting the second edge weight in the embodiment of the present application further includes:

[0077] When the production anomaly index is greater than a preset threshold, the first edge weight of the first device topology relationship is subjected to independence conversion to output a second edge weight, wherein the second edge weight includes an edge weight that identifies the independent relationship between devices.

[0078] Specifically, independence conversion refers to the process of converting the collaborative relationship between devices into an independent relationship. In the event of a production anomaly, independence conversion can reduce the collaborative strength between devices, thereby removing the faulty device and preventing the anomaly from spreading.

[0079] The production anomaly index being greater than a preset threshold is used as the triggering condition for independence conversion. When the production anomaly index is greater than the preset threshold, the first edge weight of the first device topological relationship is converted to independence, and a second edge weight is output. The second edge weight at this time is used to represent the independent relationship between devices. The preset threshold is determined based on factors such as the characteristics of the equipment, the production environment, and historical data. For example, in a high-precision machining workshop, the preset threshold of the production anomaly index is set to 0.5 by analyzing the impact of previous equipment failures on production. When the production anomaly index in actual production is greater than 0.5, it is considered that the relationship between the devices needs to be adjusted.

[0080] Independence transformation uses the isolation forest algorithm to analyze the operating data of devices in the device topology relationship and determine the independence relationship between devices. The devices in the device topology relationship are used as data points to construct an isolation forest model. The isolation forest algorithm divides data points into different isolation trees by randomly selecting features and randomly selecting split points. For each device, its anomaly score in the isolation forest is calculated. The higher the anomaly score, the greater the difference between the device and other devices, and the more likely it is to be independent. An independence threshold is determined based on the characteristics and production environment of the device. When the anomaly score of the device is greater than the threshold, the device is considered independent. Based on the independence judgment result of the device, a second edge weight is generated. For independent devices, the edge weight between them and other devices is set to 0 or a smaller value to indicate the independent relationship between them; for non-independent devices, their edge weights can be adjusted based on the original edge weights in the first device topology relationship, for example, multiplied by a coefficient less than 1 to reduce the degree of their connection.

[0081] When the production anomaly index exceeds the preset threshold, the second edge weight is output through independence conversion, which can timely adjust the equipment topology relationship to adapt to the abnormal production situation. When a serious anomaly occurs in the equipment, the production system can re-plan the collaborative relationship between the equipment to avoid the excessive impact of the faulty equipment on other equipment, thereby improving the stability and fault tolerance of the production system.

[0082] In summary, the MES intelligent scheduling method for batch production provided by the embodiments of the present application has the following technical effects:

[0083] The embodiments of the present application achieve effective response to equipment failures during the production process by dynamically adjusting device topology relationships. When a production anomaly occurs, the system first monitors the equipment's operating status and, based on historical data and standard deviation indicators, obtains real-time production anomaly indicators. The system then appropriately adjusts the coordination relationships between devices based on the anomaly level (e.g., level 1 or level 2 failure). When the anomaly indicator exceeds a preset threshold, the edge weights of the device topology relationships are updated using a gradient descent method to optimize the coordination strength between devices, thereby reducing the risk of fault propagation. Furthermore, the system flexibly adjusts the device topology relationships to address potential equipment changes during the production process (e.g., adding, replacing, or removing equipment), maintaining the stability of the production plan and the flexibility of scheduling. Furthermore, by introducing an independence conversion mechanism, the system reduces interdependence between devices when a failure occurs, improving the production system's fault tolerance and device independence, further minimizing the impact of the failure on production. Finally, by combining a dual-objective optimization scheduling function, the system maximizes equipment utilization while minimizing delays, achieving a more efficient and stable scheduling solution.

[0084] Overall, the embodiments of the present application can effectively remove faulty equipment when production anomalies occur and dynamically adapt to equipment changes by dynamically adjusting equipment topology relationships and optimizing scheduling, thereby improving the flexibility of batch production scheduling and the stability and fault tolerance of the production system, significantly improving the stability and efficiency of the production process, and ensuring the efficiency and continuity of the production process under uncertain conditions.

[0085] Example 2, as Figure 4 As shown, based on the same inventive concept as the aforementioned embodiment 1, the embodiment of the present application provides an MES intelligent scheduling system for batch production, the system comprising:

[0086] The device topology module 10 is configured to store a first device topology relationship for batch production, wherein the edges of the first device topology relationship include first edge weights that identify the collaboration strength between devices.

[0087] The operation status detection module 20 is used to detect the operation status of each device according to the device topology relationship and obtain production abnormality indicators.

[0088] The edge weight adjustment module 30 is used to update the first edge weight of the first device topology relationship by gradient descent and output a second edge weight when the production abnormality index is greater than a preset threshold, so as to obtain a second device topology relationship corresponding to the second edge weight, wherein the second edge weight is less than the first edge weight.

[0089] The scheduling update module 40 is used to connect to the MES scheduling execution engine, perform scheduling updates according to the second device topology relationship, and obtain an updated scheduling plan.

[0090] Furthermore, the edge weight adjustment module 30 in the embodiment of the present application is further configured to perform the following steps:

[0091] When the production anomaly index is greater than the preset threshold, determine whether the equipment in the first equipment topology relationship has changed; if the equipment in the first equipment topology relationship has not changed, configure the first equipment topology relationship based on the second edge weight to obtain a second equipment topology relationship.

[0092] Furthermore, the edge weight adjustment module 30 in the embodiment of the present application is further configured to perform the following steps:

[0093] If the device in the first device topology relationship changes, obtain the change type, including adding a device, replacing a device, and deleting a device; replace the first device topology relationship according to the change type, obtain the changed first device topology relationship, and configure the changed first device topology relationship based on the second edge weight to obtain the second device topology relationship.

[0094] Furthermore, the edge weight adjustment module 30 in the embodiment of the present application is further configured to perform the following steps:

[0095] Setting constraints, which include device availability constraints and connectivity constraints; performing forward propagation updates on the first edge weight of the first device topology relationship under the conditions of the availability constraints and the connectivity constraints, calculating the connectivity loss in real time, taking minimizing the connectivity loss as the gradient descent target, and obtaining the second edge weight that achieves the descent target.

[0096] Furthermore, the edge weight adjustment module 30 in the embodiment of the present application is further configured to perform the following steps:

[0097] The first device topology relationship is hierarchically graded according to the subordinate relationship, and a multi-layer device topology relationship is output; multiple first edge weights corresponding to the multi-layer device topology relationship are identified; the multiple first edge weights are updated by gradient descent to output multiple second edge weights, and the multi-layer device topology relationship corresponding to the multiple second edge weights is obtained to generate a second device topology relationship.

[0098] Furthermore, the operating status detection module 20 of the embodiment of the present application is further configured to perform the following steps:

[0099] Obtain a historical healthy operation data set under multiple historical operating conditions of each device in the device topology relationship, wherein the historical healthy operation data set includes at least a vibration spectrum, a temperature gradient, and a current waveform; perform standard deviation calculation on the historical healthy operation data set using an isolation forest algorithm to obtain a standard deviation index, compare the standard deviation index with a preset standard deviation threshold, and when the standard deviation index is greater than the preset standard deviation threshold, generate a production abnormality index based on the difference between the preset standard deviation threshold and the standard deviation index.

[0100] Furthermore, the operating status detection module 20 of the embodiment of the present application is further configured to perform the following steps:

[0101] The fault level is determined based on the production abnormality indicator, and the fault level includes a first-level fault and a second-level fault; the first-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a first attenuation gradient, and the second-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a second attenuation gradient, and output the second edge weight, wherein the attenuation gradient is the percentage of the weight decaying to the original value.

[0102] Furthermore, the schedule updating module 40 in the embodiment of the present application is further configured to perform the following steps:

[0103] Construct a dual-objective optimization scheduling function, which includes an objective function of minimizing delay time and an objective function of maximizing equipment utilization; connect to the MES scheduling execution engine and create a scheduling task template; receive the second equipment topology relationship and execution work order information from the MES data interface layer through the API interface of the MES scheduling execution engine, call the dual-objective optimization scheduling function to update the schedule, and obtain an updated scheduling plan.

[0104] Furthermore, the edge weight adjustment module 30 in the embodiment of the present application is further configured to perform the following steps:

[0105] When the production anomaly index is greater than a preset threshold, the first edge weight of the first device topology relationship is subjected to independence conversion to output a second edge weight, wherein the second edge weight includes an edge weight that identifies the independent relationship between devices.

[0106] Through the detailed description of an MES intelligent scheduling method for batch production in the foregoing specification, those skilled in the art can clearly understand an MES intelligent scheduling system for batch production in this embodiment. For the system disclosed in Example 2, since it corresponds to the method disclosed in Example 1 and has corresponding functional modules and beneficial effects, the relevant details can be referred to the description of the method part.

[0107] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A MES intelligent scheduling method for batch production, characterized by: The method comprises: Storing a first device topology relationship for batch production, wherein an edge of the first device topology relationship includes a first edge weight that identifies a collaboration strength between the devices; Detect the operating status of each device according to the device topology relationship to obtain production abnormality indicators; When the production abnormality index is greater than a preset threshold, performing gradient descent on the first edge weight of the first equipment topology relationship to output a second edge weight, thereby obtaining a second equipment topology relationship corresponding to the second edge weight, wherein the second edge weight is less than the first edge weight; Connecting to the MES scheduling execution engine, updating the schedule according to the second device topology relationship, and obtaining an updated scheduling plan; Updating the first edge weight of the first device topology relationship by gradient descent to output a second edge weight, the method comprising: Setting constraints, including device availability constraints and connectivity constraints, where the connectivity constraints are used to ensure that connections between devices remain connected; The first edge weight of the first device topology relationship is forward propagated and updated under the conditions of the availability constraint and the connectivity constraint, and the connectivity loss is calculated in real time. The minimization of the connectivity loss is used as the gradient descent target to obtain the second edge weight that achieves the descent target. The connectivity loss is used to measure the degree of connectivity destruction in the device topology relationship.

2. The MES intelligent scheduling method for batch production according to claim 1, characterized in that: Obtaining a second device topology relationship corresponding to the second edge weight, the method further includes: When the production abnormality index is greater than the preset threshold, determining whether a device in the first device topology relationship has changed; If the devices in the first device topology relationship have not changed, the first device topology relationship is configured based on the second edge weight to obtain a second device topology relationship.

3. The MES intelligent scheduling method for batch production according to claim 2, characterized in that: Determining whether a device in the first device topology relationship has changed, the method further includes: If a device in the first device topology relationship changes, obtain the change type, including adding a device, replacing a device, and deleting a device; The first device topology relationship is replaced and changed according to the change type to obtain the changed first device topology relationship, and the changed first device topology relationship is configured based on the second edge weight to obtain the second device topology relationship.

4. The MES intelligent scheduling method for batch production according to claim 1, characterized in that: Outputting the second edge weight, the method further includes: hierarchically classifying the first device topology relationship according to the subordinate relationship, and outputting a multi-layer device topology relationship; Identifying a plurality of first edge weights corresponding to the multi-layer device topology relationship; The multiple first edge weights are updated respectively by gradient descent to output multiple second edge weights, and multi-layer device topology relationships corresponding to the multiple second edge weights are obtained to generate a second device topology relationship.

5. The MES intelligent scheduling method for batch production according to claim 1, characterized in that: The operating status of each device is detected according to the device topology relationship to obtain production abnormality indicators, and the method includes: Acquire a historical healthy operation data set under multiple historical working conditions of each device in the device topology relationship, wherein the historical healthy operation data set includes at least a vibration spectrum, a temperature gradient, and a current waveform; The standard deviation of the historical healthy operation data set is calculated using the isolation forest algorithm to obtain a standard deviation index, which is then compared with a preset standard deviation threshold. When the standard deviation index is greater than the preset standard deviation threshold, a production abnormality index is generated based on the difference between the preset standard deviation threshold and the standard deviation index.

6. The MES intelligent scheduling method for batch production according to claim 5, characterized in that: After generating the production anomaly indicator, the method further includes: Determining a fault level according to the production abnormality indicator, wherein the fault level includes a primary fault and a secondary fault; The first-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a first attenuation gradient, and the second-level fault is used to update the first edge weight of the first device topology relationship by gradient descent according to a second attenuation gradient, and output the second edge weight, wherein the attenuation gradient is the percentage of the weight decaying to the original value.

7. The MES intelligent scheduling method for batch production according to claim 1, characterized in that: Connecting to the MES scheduling execution engine and updating the schedule according to the second device topology relationship, the method includes: Constructing a dual-objective optimization scheduling function, wherein the dual-objective optimization scheduling function includes an objective function of minimizing delay time and an objective function of maximizing equipment utilization; Connect to the MES scheduling execution engine and create scheduling task templates; The second equipment topology relationship and the execution work order information from the MES data interface layer are received through the API interface of the MES scheduling execution engine, and the dual-objective optimization scheduling function is called to update the scheduling to obtain an updated scheduling plan.

8. The MES intelligent scheduling method for batch production according to claim 1, characterized in that: The method of outputting the second edge weight can also be done in the following steps: When the production anomaly index is greater than a preset threshold, the first edge weight of the first device topology relationship is subjected to independence conversion to output a second edge weight, wherein the second edge weight includes an edge weight that identifies the independent relationship between devices, and the independence conversion uses an isolation forest algorithm to determine the independent relationship between devices.

9. An MES intelligent scheduling system for batch production, characterized by: The system is used to execute the MES intelligent scheduling method for batch production according to any one of claims 1 to 8, comprising: A device topology module, configured to store a first device topology relationship for batch production, wherein an edge of the first device topology relationship includes a first edge weight that identifies a collaboration strength between devices; An operation status detection module is used to detect the operation status of each device according to the device topology relationship and obtain production abnormality indicators; an edge weight adjustment module, configured to, when the production anomaly index is greater than a preset threshold, update the first edge weight of the first device topology relationship by gradient descent to output a second edge weight, thereby obtaining a second device topology relationship corresponding to the second edge weight, wherein the second edge weight is less than the first edge weight; The scheduling update module is used to connect to the MES scheduling execution engine, perform scheduling updates according to the second device topology relationship, and obtain an updated scheduling plan.

Images