A distributed platform manufacturing service scheduling algorithm testing method and system

Abstract

The application relates to a distributed platform manufacturing service scheduling algorithm test method and system, and belongs to the technical field of industrial internet, and comprises the following steps: platform information sensing, which is used for describing platform dynamic characteristics, sensing manufacturing task attributes and real-time information of service nodes; platform state evaluation, which is used for evaluating the urgency of a manufacturing task and the load state and shortage degree of a service node; manufacturing service scheduling algorithm test, which is used for constructing constraint conditions and an objective function of platform manufacturing task allocation and manufacturing service scheduling problems, and further screening a suitable scheduling algorithm for test; algorithm evaluation, which is used for evaluating the running result of a manufacturing service scheduling algorithm and the performance of the algorithm itself; and algorithm recommendation, which is used for analyzing the current platform operation state and performing algorithm recommendation in combination with algorithm evaluation data. The application effectively improves the efficiency of platform manufacturing task allocation and service scheduling, and greatly reduces the occurrence of service queue congestion.

Description

Technical Field

[0001] The present invention belongs to the field of industrial Internet technology, and specifically relates to a distributed platform manufacturing service scheduling algorithm testing method and system. Background Art

[0002] With the continuous penetration of the digital economy ecosystem into the manufacturing sector, industrial internet platforms that enable the sharing of socialized manufacturing resources and capabilities are crucial vehicles for promoting the digital transformation and intelligent upgrade of the manufacturing industry. Manufacturing demands submitted to industrial internet platforms are characterized as manufacturing tasks with clear demand attributes; manufacturing resources and capabilities are digitally encapsulated into manufacturing service access platforms, where different manufacturing services can aggregate and collaborate on demand to complete a complex manufacturing task. Manufacturing tasks delivered by industrial internet platforms feature multi-functional, complex requirements, large-scale operations, heterogeneity, and randomness. Platform manufacturing services are heterogeneous and the number of collaborative services is limited. Different manufacturing services with similar functions often have varying service capabilities. This leads to conflicting long-term service throughput maximization and service utilization fairness for similar services, under the dual uncertainties of supply and demand during the platform's actual operations. This inevitably leads to congestion in some service queues, impacting the platform's continued robust operation.

[0003] In response to the above problems, there is an urgent need for a distributed platform manufacturing service scheduling algorithm testing method and system. The system includes a platform information perception module, a platform status evaluation module, a manufacturing service scheduling algorithm testing module, an algorithm evaluation module, and an algorithm recommendation module. Through the collaboration of various modules, real-time perception of manufacturing platform information can be achieved, and a recommendation algorithm can be given according to the platform operation status to realize platform manufacturing task allocation and manufacturing service scheduling algorithm testing and evaluation. Summary of the Invention

[0004] To address the above technical issues, the present invention proposes a distributed platform manufacturing service scheduling algorithm testing method and system. The method includes five steps: platform information perception, platform status assessment, manufacturing service scheduling algorithm testing, algorithm evaluation, and algorithm recommendation. By perceiving the manufacturing platform status in real time, the present invention can evaluate the adaptability of multiple algorithms in specific task scenarios, conduct multi-algorithm testing and comparison, and select the optimal algorithm to regulate task allocation and service scheduling among service nodes. This method achieves a balance between maximizing the platform service community's time-averaged throughput and utilization fairness, while reducing service queue congestion. This provides support for research on the sustained and robust operation of industrial internet platforms.

[0005] The present invention provides a distributed platform manufacturing service scheduling algorithm testing method, comprising the following steps:

[0006] Step 1: Platform information perception, including: describing the dynamic characteristics of manufacturing service node clustering and combined collaboration relationships within and between different manufacturing service communities on the platform based on the established manufacturing service aggregation network topology; perceiving manufacturing task attributes after receiving manufacturing task arrival signals; and monitoring the status of the platform's manufacturing service nodes in real time to obtain real-time information about the service nodes.

[0007] Step 2: Platform status assessment, including: assessing the urgency of manufacturing tasks based on perceived manufacturing task attributes; assessing the load status and shortage of manufacturing service nodes based on perceived manufacturing service node status information;

[0008] Step 3: Testing the manufacturing service scheduling algorithm, including: constructing constraints for platform manufacturing task allocation and manufacturing service scheduling based on the status of the platform's manufacturing service nodes; constructing an optimization objective function, which is further decomposed into two sub-problems for task allocation and service scheduling; based on the objective function and constraints, selecting appropriate scheduling algorithms from a pre-built algorithm library for testing according to the real-time perception and evaluation of the platform's operating status and manufacturing task requirements;

[0009] Step 4: Algorithm evaluation, including: Based on the operational results of the manufacturing service scheduling algorithm, evaluating the scheduling solutions generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing; and evaluating the performance of different algorithms themselves in terms of time complexity, memory consumption, and scalability;

[0010] Step 5: Algorithm recommendation, including: in-depth analysis of the current platform operation status, evaluating the current platform operation status by perceiving the manufacturing task attributes and manufacturing service status; and making algorithm recommendations based on the platform operation status and the algorithm evaluation data obtained by the algorithm evaluation.

[0011] The present invention also provides a distributed platform manufacturing service scheduling algorithm testing system, comprising:

[0012] Platform information perception module: Based on the established manufacturing service aggregation network topology, the platform information perception module describes the dynamic characteristics of the clustering and combined collaborative relationships of manufacturing service nodes within and between different manufacturing service communities on the platform. After receiving the manufacturing task arrival signal, it perceives the attributes of the manufacturing task, monitors the status of the platform's manufacturing service nodes in real time, and obtains real-time information about the service nodes.

[0013] Platform status assessment module: The platform status assessment module assesses the urgency of manufacturing tasks based on the perceived attributes of manufacturing tasks; and assesses the load status and shortage of manufacturing service nodes based on the perceived status information of manufacturing service nodes;

[0014] Manufacturing service scheduling algorithm testing module: This module constructs constraints for platform manufacturing task allocation and manufacturing service scheduling based on the status of the platform's manufacturing service nodes. It also constructs an optimization objective function, which is further decomposed into two sub-problems for task allocation and service scheduling. Based on the objective function and constraints, it selects appropriate scheduling algorithms from a pre-built algorithm library for testing, according to the real-time perception and evaluation of the platform's operating status and manufacturing task requirements.

[0015] Algorithm evaluation module: Based on the operating results of the manufacturing service scheduling algorithm, the algorithm evaluation module evaluates the scheduling solutions generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing; and evaluates the performance of different algorithms themselves in terms of time complexity, memory consumption, and scalability;

[0016] Algorithm recommendation module: The algorithm recommendation module conducts an in-depth analysis of the current platform operation status, and evaluates the current platform operation status by perceiving the manufacturing task attributes and manufacturing service status; based on the platform operation status, the algorithm evaluation data obtained by the algorithm evaluation is combined to make algorithm recommendations.

[0017] The advantages of the present invention compared with the prior art are:

[0018] (1) The platform status assessment step proposed in the present invention can effectively assess the actual dynamic operation status of the platform, the urgency of manufacturing tasks, and the shortage of service nodes based on the perceived manufacturing task information and service node information.

[0019] (2) The manufacturing service scheduling algorithm test steps proposed in this invention can effectively regulate the task allocation and service scheduling among service nodes during the platform operation process, achieving a balance between maximizing the time-averaged throughput of the platform service community and utilization fairness.

[0020] (3) The algorithm recommendation step proposed in the present invention can comprehensively evaluate the performance indicators of different algorithms. By analyzing the current platform operation status, it automatically recommends the most appropriate scheduling algorithm, effectively improving the efficiency of platform manufacturing task allocation and service scheduling, and greatly reducing the occurrence of service queue congestion. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. To achieve the above-mentioned objectives, the present invention adopts the following technical solutions.

[0023] The present invention will be described in further detail below with reference to the accompanying drawings.

[0024] This paper proposes a distributed platform manufacturing service scheduling algorithm testing method and system. The method includes five steps: platform information perception, platform status assessment, manufacturing service scheduling algorithm testing, algorithm evaluation, and algorithm recommendation. The system includes five modules: platform information perception module, platform status assessment module, manufacturing service scheduling algorithm testing module, algorithm evaluation module, and algorithm recommendation module.

[0025] The flow chart of the present invention is as follows Figure 1 As shown, the specific implementation is as follows:

[0026] Step 1: Platform information perception, including: based on the established manufacturing service aggregation network topology, describing the dynamic characteristics of manufacturing service node clustering and combined collaboration relationships within and between different manufacturing service communities on the platform; perceiving manufacturing task attributes after receiving manufacturing task arrival signals; and monitoring the status of the platform's manufacturing service nodes in real time to obtain real-time information about the service nodes. Specific implementation methods are as follows:

[0027] Step 1.1: To improve the efficiency of platform information perception and task processing, the platform manufacturing resources and manufacturing capabilities are abstracted and encapsulated into manufacturing service nodes, while considering the functional and non-functional attributes of the nodes (Quality of Service, QoS for short). Based on graph theory, the topology of the manufacturing service aggregation network is described as a graph. ,satisfy ,in Represents the set of all service nodes in the network, To serve the community The nodes, To serve the community The number of nodes in is the number of all service communities in the network, ; It is the set of all edges in the network, representing the clustering / combination collaboration relationship between nodes, that is, clustering / combination collaboration edges. The number of nodes in different service communities is the same. Each service node is the smallest service unit and has one or more functional attributes. After the node function is decomposed and replicated, it belongs to different service communities: within the same service community, they are all clustering collaboration edges and are undirected; between different service communities, they are all combination collaboration edges and are directed; several nodes in the same service community can cluster and collaborate to complete a large-scale task, and nodes in different service communities can combine and collaborate to complete a task with multiple functional requirements;

[0028] Step 1.2: When the system receives the manufacturing task arrival signal, it perceives the information of the manufacturing task. A certain period of time during the platform operation is divided into several moments , T represents the length of a specific time period, and every two adjacent moments and The unit time length between .exist At this moment, the manufacturing tasks arriving at the platform are decomposed according to the functional requirements and divided into different subtask sets, representing the current moment with the first The subtask set of class functional requirements is , the number of subtasks that arrive is , the number of completed (i.e., departed) subtasks is .exist At this moment, Represents a subtask set The mth subtask in the task, the specific task attributes can be described as .in, Indicates the functional requirement type, Represents the load of each subtask, and the subset of candidate service nodes that meet the QoS requirements is ; According to the carrying capacity limit of the platform during operation, the number of tasks that reach the subset The maximum number of tasks allowed should not be exceeded . Represents a manufacturing subtask Deadline; express Moment subtask set Manufacturing subtasks in Total execution time; express Moment subtask set Manufacturing subtasks in Length of time it has been executed; express Moment subtask set Manufacturing subtasks in The remaining execution time of ; Indicates the completion of the manufacturing subtask The number of service nodes required in each service cluster, To complete the manufacturing subtask The number of service nodes required in the jth service cluster. The number of nodes that need to be called in each service cluster should not exceed the total number of nodes in the cluster. . express Moment subtask set Manufacturing subtasks in Arrival, =1 means the task has arrived, =0 means not arrived; express Moment subtask set Manufacturing subtasks in Will it be completed on time? =1 means the task will be completed on time, =0 means the task cannot be completed on time; express Moment subtask set Manufacturing subtasks in Is it being executed? =1 means the task is being executed. =0 means the task has not been executed. Task types include regular manufacturing tasks, emergency manufacturing tasks, and special manufacturing tasks. Different types of manufacturing tasks have different requirements for manufacturing resources and time. By sensing task information, service nodes can be allocated and scheduled according to the specific situation of the manufacturing task to ensure efficient execution of the task;

[0029] Step 1.3: Monitor the status of the platform manufacturing service nodes in real time and obtain real-time information of the service nodes. At this moment, the service node The available status is , Indicates that the service node is available. Indicates that the node is unavailable. If and only if the service node is available, the service node can be matched with subtasks with the same functional requirements; service node The matching is , Indicates that the service node matches a subtask. Indicates that the service node does not match any subtask; represent Assigned to the service node at this moment The number of subtasks, There must be By sensing the status information of service nodes, it provides strong data support for the rational allocation of manufacturing resources and ensures the efficient operation of the manufacturing platform.

[0030] Step 2: Platform status assessment, including: assessing the urgency of manufacturing tasks based on perceived manufacturing task attributes; assessing the load status and shortage of manufacturing service nodes based on perceived manufacturing service node status information; the specific implementation method is as follows:

[0031] Step 2.1: Based on the perceived manufacturing task information, evaluate the urgency of the manufacturing task. To quantify the urgency of the manufacturing task and determine the priority of task allocation and service scheduling based on the urgency of the manufacturing task, the deadline of the manufacturing task is comprehensively considered. , Remaining execution time Total execution time , the number of service nodes required , arrival status , execution status The calculation formula for the comprehensive evaluation index of manufacturing task urgency is constructed based on factors such as:

[0032] (1)

[0033] in, , , It is the weight coefficient of each factor, indicating the relative importance of different factors to the urgency of the task.

[0034] When a manufacturing task arrives at the platform ( =1), the urgency of the manufacturing task is measured by formula (1) to determine whether the task needs to be handled urgently or there is still buffer time;

[0035] Step 2.2: Based on the perceived service node status information, evaluate the load status and shortage level of the service node.

[0036] (1) In At this moment, each service node Load status Can be assessed as:

[0037] , , (2)

[0038] in, yes The waiting queue length of the service node at this moment, is the maximum number of tasks allowed to be carried by the node;

[0039] (2) In At this moment, comprehensively consider the number of service nodes available on the platform and the load of each service node , free time and task completion The calculation formula for the comprehensive evaluation index of manufacturing service node scarcity is constructed based on factors such as:

[0040] (3)

[0041] in, Is a service node exist The number of tasks completed at any given moment and satisfying ; Service node per unit time The task load is the execution rate; , , , It is the weight coefficient of each factor, indicating the relative importance of different factors to the urgency of the task.

[0042] By constructing the above evaluation formula, we can measure the degree of scarcity of manufacturing service nodes and determine whether manufacturing resources are in a state of scarcity and difficulty in supporting the execution of manufacturing tasks, or are relatively abundant and can be flexibly allocated.

[0043] Step 3: Testing the manufacturing service scheduling algorithm, including: constructing constraints for platform manufacturing task allocation and manufacturing service scheduling based on the status of the platform's manufacturing service nodes; constructing an optimization objective function, which is further decomposed into two sub-problems for task allocation and service scheduling; based on the objective function and constraints, selecting appropriate scheduling algorithms from a pre-built algorithm library for testing based on the real-time perception and evaluation of the platform's operating status and manufacturing task requirements. The specific implementation method is as follows:

[0044] Step 3.1: Based on the platform manufacturing task attributes and service node status, construct the constraints of the platform manufacturing task allocation and manufacturing service scheduling problem:

[0045] (1) When matching and executing subtasks through service nodes, The first subtask QoS attribute value Should meet , , ,in For subtask QoS attribute requirements, A set of QoS indicators, including manufacturing service availability , execution rate , load status ,reliability and other indicators;

[0046] (2) Considering the limited number of manufacturing service nodes, in order to reduce service queue congestion, each service queue in the service cluster should meet the time average stability constraint:

[0047] (4)

[0048] in, For service nodes The average length of the waiting queue is For service nodes The maximum allowable upper bound of the average waiting queue length is E, which is the expected value. is the set of positive real numbers;

[0049] Step 3.2: To ensure the long-term throughput performance of the platform under the dynamic operation environment, construct a multi-objective function that considers the throughput of different nodes and the fairness of service utilization Consider two decision variables and The independence of , can further decompose the optimization problem into two sub-optimization problems with low complexity, namely, the sub-optimization problem for task allocation and the sub-optimization problem for service scheduling. On the basis of ensuring the long-term throughput performance of the platform, other optimization objective functions are further set according to the type of manufacturing tasks and the status of service nodes. For example, in a scenario where manufacturing tasks are relatively tight, in order to ensure that all tasks can be completed in the shortest time, a minimum total task completion time function can be constructed. In the scenario where service nodes are relatively scarce, in order to ensure that limited resources are used most efficiently, a function that maximizes resource utilization can be constructed. ;

[0050] Step 3.3: Based on the objective function and constraints, and according to the real-time perception and evaluation of the platform operation status and manufacturing task requirements, select the appropriate scheduling algorithm from the pre-built algorithm library for testing. The algorithm library includes a variety of algorithms for solving scheduling problems, such as simple optimization algorithms, heuristic optimization algorithms, intelligent optimization algorithms, etc. When the number of manufacturing tasks arriving at the platform is small and the service node load is relatively light, a simple traditional scheduling algorithm may be selected; when the manufacturing task demand is high and the service node load is heavy, a heuristic or intelligent optimization algorithm that is good at handling multi-constraint nonlinear complex optimization problems tends to be selected. After selecting a suitable scheduling algorithm based on the platform operation status and manufacturing task requirements, the above sub-optimization problem is solved for each service community in the manufacturing service aggregation network to obtain the decision variables The optimal solution , ,according to The value of generates the platform manufacturing task allocation and service scheduling sequence table, for example, for the found ,if , then it is a service node Assign subtasks, otherwise do not assign them. Configure service nodes for manufacturing tasks according to the sequence table, so as to promote the service scheduling optimization process for different service nodes in the service community.

[0051] Step 4: Algorithm evaluation, including: Based on the operating results of the manufacturing service scheduling algorithm, evaluate the scheduling solutions generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing; and evaluate the performance of different algorithms themselves in terms of time complexity, memory consumption, and scalability. The specific implementation methods are as follows:

[0052] Step 4.1: Based on the results of the manufacturing service scheduling algorithm, evaluate the scheduling schemes generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing. Evaluate the resource utilization of each scheme by calculating the ratio of the number of utilized manufacturing service nodes to the total number of aggregated network service nodes at a specific time. Evaluate the response rate and efficiency of each scheme by calculating the waiting time of each manufacturing task arriving at the platform at a specific time. Evaluate the load balancing of service nodes under each scheme by comparing the occupancy time of each service node with the number of assigned manufacturing tasks during a specific time period.

[0053] Step 4.2: Based on the results of the manufacturing service scheduling algorithm, evaluate the performance of different algorithms in terms of time complexity, memory consumption, and scalability. By analyzing the operational steps of each algorithm component, calculate the time complexity of the scheduling algorithm. The lower the complexity, the more efficient the algorithm. Evaluate the algorithm's memory consumption by measuring the memory space used during runtime at different manufacturing task scales, especially for large-scale manufacturing task allocation and service scheduling. The lower the memory consumption, the more efficient the system's resource utilization. Evaluate the algorithm's scalability by increasing the number of manufacturing tasks and service nodes and comparing the algorithm's runtime and memory consumption. The better the scalability, the more stable the algorithm's performance.

[0054] Step 5: Algorithm recommendation, including: conducting an in-depth analysis of the current platform operation status, evaluating the current platform operation status by sensing the manufacturing task attributes and manufacturing service status; and making algorithm recommendations based on the platform operation status and the algorithm evaluation data obtained from the algorithm evaluation. The specific implementation method is as follows:

[0055] Step 5.1: In-depth analysis of the current platform operation state, by sensing the manufacturing task attributes and manufacturing service state, the current platform operation state is evaluated, which is usually divided into normal state and emergency state. In the normal state, the manufacturing task demand and service node load state remain relatively stable, the task allocation is relatively balanced, and the utilization rate of the service node is moderate; in the emergency state, the manufacturing task demand increases sharply, the service node load is close to saturation, and the resources are relatively scarce;

[0056] Step 5.2: Based on the platform operation state, algorithm recommendation is made in combination with the algorithm evaluation data of the algorithm evaluation module. For example, in the normal operation state, it is necessary to maintain the efficient use of resources and ensure the smooth execution of various tasks, and a high-efficiency and stable scheduling algorithm is recommended to ensure the smooth progress of manufacturing task allocation and service scheduling; in the emergency operation state, it is necessary to prioritize the processing of emergency manufacturing tasks, and a scheduling algorithm that can quickly respond is recommended, and through dynamic resource allocation, load balancing and other ways, the interruption and delay of manufacturing tasks are minimized. Through timely sensing and analysis of different operation states, combined with algorithm evaluation data, the most suitable algorithm is selected from the algorithm library to regulate the task allocation and service scheduling among service nodes.

[0057] In summary, the present application discloses a distributed platform manufacturing service scheduling algorithm test system. The system includes a platform information sensing module, a platform state evaluation module, a manufacturing service scheduling algorithm test module, an algorithm evaluation module, and an algorithm recommendation module. The present application can sense the manufacturing platform state in real time, evaluate the adaptability of multiple algorithms in different platform operation states, test and compare multiple algorithms, select the optimal algorithm to regulate the task allocation and service scheduling among service nodes, achieve the balance between the maximum time average throughput of platform service community and utilization fairness, and reduce the occurrence of service queue congestion, providing support for the continuous and stable operation of industrial internet platform.

[0058] The contents not described in detail in the specification of the present application belong to the prior art known to those skilled in the art. The above is only the preferred embodiment of the present application, and it should be pointed out that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should be regarded as the protection scope of the present application.

Claims

1. A distributed platform manufacturing service scheduling algorithm testing method, characterized in that: The following steps are involved: Step 1: Platform information perception, including: describing the dynamic characteristics of manufacturing service node clustering and combined collaboration relationships within and between different manufacturing service communities on the platform based on the established manufacturing service aggregation network topology; perceiving manufacturing task attributes after receiving manufacturing task arrival signals; and monitoring the status of the platform's manufacturing service nodes in real time to obtain real-time information about the service nodes. Step 2: Platform status assessment, including: assessing the urgency of manufacturing tasks based on perceived manufacturing task attributes; assessing the load status and shortage of manufacturing service nodes based on perceived manufacturing service node status information; Step 3: Testing the manufacturing service scheduling algorithm, including: constructing constraints for platform manufacturing task allocation and manufacturing service scheduling based on the status of the platform's manufacturing service nodes; constructing an optimization objective function, which is further decomposed into two sub-problems for task allocation and service scheduling; based on the objective function and constraints, selecting appropriate scheduling algorithms from a pre-built algorithm library for testing according to the real-time perception and evaluation of the platform's operating status and manufacturing task requirements; Step 4: Algorithm evaluation, including: Based on the operational results of the manufacturing service scheduling algorithm, evaluating the scheduling solutions generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing; and evaluating the performance of different algorithms themselves in terms of time complexity, memory consumption, and scalability; Step 5: Algorithm recommendation, including: in-depth analysis of the current platform operation status, evaluating the current platform operation status by perceiving the manufacturing task attributes and manufacturing service status; and making algorithm recommendations based on the platform operation status and the algorithm evaluation data obtained by the algorithm evaluation.

2. A distributed platform manufacturing service scheduling algorithm testing method according to claim 1, characterized in that: The step 1 comprises: Step 1.1: Abstract and encapsulate the platform manufacturing resources and manufacturing capabilities into manufacturing service nodes, and describe the manufacturing service aggregation network topology as a graph. ,satisfy ,in Represents the set of all manufacturing service nodes in the network, Manufacturing service community The nodes, Manufacturing service community The number of nodes in The number of all manufacturing service communities in the network, , ; It is the set of all edges in the network, representing the clustering or combined collaboration relationship between manufacturing service nodes, that is, clustering or combined collaboration edges. The number of manufacturing service nodes in different service communities is the same. Each manufacturing service node is the minimum service unit and has one or more functional attributes. After node function decomposition and replication, it belongs to different manufacturing service communities: within the same manufacturing service community, all are clustered collaboration edges and are undirected, while between different manufacturing service communities, all are combined collaboration edges and are directed; several nodes in the same manufacturing service community can cluster and collaborate to complete a certain large-scale task, and nodes in different manufacturing service communities can combine and collaborate to complete a task with multiple functional requirements; Step 1.2: After receiving the manufacturing task signal, the attributes of the manufacturing task are perceived and a certain period of time during the platform operation is divided into several moments. , T represents the length of the time period, every two adjacent moments and The unit time length between ,exist At the moment, the manufacturing tasks arriving at the platform are decomposed according to the functional requirements and divided into different subtask sets to represent the current moment The following has The subtask set of class functional requirements is , the number of subtasks that arrive is , the number of subtasks that are completed and left is ,exist At this moment, Represents a subtask set The mth subtask in the task, the specific task attributes are described as ,in, Indicates the functional requirement type, Represents the load of each subtask, and the subset of candidate service nodes that meet the QoS requirements is ; According to the carrying capacity limit of the platform during operation, the number of tasks that reach the subset The maximum number of tasks allowed should not be exceeded , Represents a manufacturing subtask Deadline; express Moment subtask set Manufacturing subtasks in Total execution time; express Moment subtask set Manufacturing subtasks in Length of time it has been executed; express Moment subtask set Manufacturing subtasks in The remaining execution time of ; Indicates the completion of the manufacturing subtask The number of manufacturing service nodes required in each service cluster, To complete the manufacturing subtask The number of manufacturing service nodes required in the jth service cluster. The number of nodes required to be called in each manufacturing service cluster does not exceed the total number of nodes in the cluster. , express Moment subtask set Manufacturing subtasks in Arrival, =1 means the task has arrived, =0 means not arrived; express Moment subtask set Manufacturing subtasks in Will it be completed on time? =1 means the task will be completed on time, =0 means the task cannot be completed on time; express Moment subtask set Manufacturing subtasks in Is it being executed? =1 means the task is being executed. =0 means the task has not been executed. Manufacturing tasks include regular manufacturing tasks and emergency manufacturing tasks. By sensing the attributes of manufacturing tasks, service nodes are allocated and scheduled according to the specific circumstances of the manufacturing tasks to ensure efficient execution of tasks. Step 1.3: Monitor the status of the manufacturing service nodes on the platform in real time and obtain real-time information of the manufacturing service nodes. At this moment, manufacturing service nodes The available status is , Indicates that the manufacturing service node is available. Indicates that the manufacturing service node is unavailable. If and only if the manufacturing service node is available, the manufacturing service node is matched with the subtask with the same functional requirements; the manufacturing service node The matching is , Indicates that the manufacturing service node matches a subtask. Indicates that the manufacturing service node does not match any subtask; represent Assigned to the manufacturing service node at this moment The number of subtasks, Time .

3. A distributed platform manufacturing service scheduling algorithm testing method according to claim 2, characterized in that: The step 2 includes: Step 2.1: Based on the perceived manufacturing task attributes, evaluate the urgency of the manufacturing task, quantify the urgency of the manufacturing task, and determine the priority of task allocation and service scheduling based on the urgency of the manufacturing task, taking into account the deadline of the manufacturing task. , Remaining execution time Total execution time , the number of service nodes required , arrival status , execution status The calculation formula for the comprehensive evaluation index of manufacturing task urgency is constructed based on various factors: （1） in, , , is the weight coefficient of each factor, indicating the relative importance of different factors to the urgency of the task; When a manufacturing task arrives at the platform =1, the urgency of the manufacturing task is measured by formula (1) to determine whether the task needs to be handled urgently or there is still buffer time; Step 2.2: Based on the perceived manufacturing service node status information, evaluate the load status and shortage level of the service node; including: (1) In At this moment, each manufacturing service node Load status Evaluates to: , , （2） in, yes The waiting queue length of the manufacturing service node at the moment, The maximum number of tasks allowed to be carried by the manufacturing service node; (2) In At this moment, comprehensively consider the number of service nodes available on the platform and the load of each service node , free time and task completion The calculation formula for the comprehensive evaluation index of manufacturing service node scarcity is constructed based on various factors: （3） in, It is a manufacturing service node exist The number of tasks completed at any given moment and satisfying ; Service node per unit time The task load is the execution rate; , , , is the weight coefficient of each factor, indicating the relative importance of different factors to the urgency of the task; By constructing the above evaluation formula, we can measure the degree of scarcity of manufacturing service nodes and determine whether manufacturing resources are in a state of scarcity and difficulty in supporting the execution of manufacturing tasks, or are relatively abundant and can be flexibly allocated.

4. A distributed platform manufacturing service scheduling algorithm testing method according to claim 3, characterized in that: The step 3 includes: Step 3.1: Based on the platform manufacturing task attributes and manufacturing service node status, construct the platform manufacturing task allocation and manufacturing service scheduling problem constraints, including: (1) When matching and executing subtasks through manufacturing service nodes, The first subtask QoS attribute value Should meet , , ,in For subtask QoS attribute requirements, A set of QoS indicators, including manufacturing service availability , execution rate , load status ,reliability Each indicator; (2) Considering the limited number of manufacturing service nodes, in order to reduce service queue congestion, each service queue in the manufacturing service cluster should meet the time average stability constraint: （4） in, For service nodes The average length of the waiting queue is For service nodes The maximum allowable upper bound of the average waiting queue length is E, which is the expected value. is the set of positive real numbers; Step 3.2: Construct a multi-objective function that considers different node throughput and service utilization fairness: ; Consider two decision variables and Based on the independence of the manufacturing tasks, the optimization problem is further decomposed into two sub-optimization problems with low complexity, namely the sub-optimization problem for task allocation and the sub-optimization problem for service scheduling. According to the types of manufacturing tasks and the status of manufacturing service nodes, other optimization objective functions are further set. In the scenario where the manufacturing tasks are relatively tight, in order to ensure that all tasks can be completed in the shortest time, a minimum total task completion time function is constructed. In the scenario where service nodes are relatively scarce, in order to ensure that limited resources are used most efficiently, a function that maximizes resource utilization is constructed. ; Step 3.3: Based on the objective function and constraints, select appropriate scheduling algorithms from the pre-built algorithm library according to the real-time perception and evaluation of the platform operation status and manufacturing task requirements for testing; after selecting the appropriate scheduling algorithm based on the platform operation status and manufacturing task requirements, solve the above sub-optimization problem for each service community in the manufacturing service aggregation network to obtain the decision variables The optimal solution , ,according to The value of generates the platform manufacturing task allocation and service scheduling sequence table, for the found ,if , then it is a manufacturing service node Assign subtasks, otherwise do not assign them, and configure manufacturing service nodes for manufacturing tasks according to the sequence table.

5. A distributed platform manufacturing service scheduling algorithm testing method according to claim 4, characterized in that: The step 4 comprises: Step 4.1: Based on the results of the manufacturing service scheduling algorithm, evaluate the scheduling schemes generated by different algorithms under different operating conditions from four aspects: resource utilization, response rate, efficiency, and load balance. Evaluate the resource utilization of the scheme by calculating the ratio of the number of utilized manufacturing service nodes at a specific time to the total number of manufacturing service nodes in the aggregation network. Evaluate the response rate and efficiency of the scheme by calculating the waiting time of each manufacturing task arriving at the platform at time t. Evaluate the load balance of the manufacturing service nodes under the scheme by comparing the occupancy time of each service node and the number of assigned manufacturing tasks during time period t. Step 4.2: Based on the running results of the manufacturing service scheduling algorithm, evaluate the performance of different algorithms from three aspects: time complexity, memory consumption, and scalability; calculate the time complexity of the scheduling algorithm by analyzing the operation steps of each part of the algorithm; measure the memory space used by the algorithm when running under different manufacturing task scales to determine the manufacturing task allocation and service scheduling under different scales, and evaluate the algorithm's memory consumption; evaluate the algorithm's scalability by increasing the number of manufacturing tasks and manufacturing service nodes and comparing the changes in algorithm running time and memory consumption.

6. A distributed platform manufacturing service scheduling algorithm testing method according to claim 5, characterized in that: The step 5 comprises: Step 5.1: Conduct an in-depth analysis of the current platform operating status. By sensing the attributes of manufacturing tasks and the status of manufacturing service nodes, the current platform operating status is assessed and divided into normal and emergency states. In the normal state, manufacturing task demand and service node load remain relatively stable, task distribution is balanced, and service node utilization is moderate. In the emergency state, manufacturing task demand surges, service node load approaches saturation, and resources are relatively tight. Step 5.2: Recommend an algorithm based on the platform operating status and the algorithm evaluation data obtained from the algorithm evaluation. This includes: recommending an efficient and stable scheduling algorithm under normal operating conditions to ensure the smooth allocation of manufacturing tasks and service scheduling; recommending a fast-responding scheduling algorithm under emergency operating conditions; selecting the most appropriate algorithm from the algorithm library through timely perception and analysis of different operating conditions, combined with algorithm evaluation data, to regulate task allocation and service scheduling between service nodes.

7. A distributed platform manufacturing service scheduling algorithm testing system, characterized in that: include: Platform information perception module; The platform information perception module is based on the established manufacturing service aggregation network topology structure and describes the dynamic characteristics of the manufacturing service node clustering and combination collaboration relationship within and between different manufacturing service communities on the platform; After receiving the manufacturing task arrival signal, the manufacturing task attributes are perceived; Monitor the status of the platform's manufacturing service nodes in real time and obtain real-time information about the service nodes; Platform status assessment module; The platform status assessment module assesses the urgency of manufacturing tasks based on the perceived attributes of manufacturing tasks; and assesses the load status and shortage of manufacturing service nodes based on the perceived status information of manufacturing service nodes; Manufacturing service scheduling algorithm testing module: This module constructs constraints for platform manufacturing task allocation and manufacturing service scheduling based on the status of the platform's manufacturing service nodes. It also constructs an optimization objective function, which is further decomposed into two sub-problems for task allocation and service scheduling. Based on the objective function and constraints, it selects appropriate scheduling algorithms from a pre-built algorithm library for testing, according to the real-time perception and evaluation of the platform's operating status and manufacturing task requirements. Algorithm evaluation module: Based on the operating results of the manufacturing service scheduling algorithm, the algorithm evaluation module evaluates the scheduling solutions generated by different algorithms under different operating conditions in terms of resource utilization, response rate, efficiency, and load balancing; and evaluates the performance of different algorithms themselves in terms of time complexity, memory consumption, and scalability; Algorithm recommendation module: The algorithm recommendation module conducts an in-depth analysis of the current platform operation status, and evaluates the current platform operation status by perceiving the manufacturing task attributes and manufacturing service status; based on the platform operation status, the algorithm evaluation data obtained by the algorithm evaluation is combined to make algorithm recommendations.

Images