Task time sequence tight coupling cloud manufacturing energy consumption and service quality collaborative scheduling optimization method

Abstract

The invention discloses a task time sequence tight coupling cloud manufacturing energy consumption and service quality collaborative scheduling optimization method, belongs to the field of cloud computing, intelligent manufacturing and resource scheduling optimization, and aims to solve the problems of preheating energy consumption waste, service occupation conflict, supply and demand benefit imbalance, insufficient optimization performance and the like in the prior art. Comprising the following steps: decomposing a customized task into independent sub-tasks and matching a candidate service set; constructing a bilevel planning model in which the preheating energy consumption of the upper layer focuses on the minimum supplier and the lower layer is based on the QoS weighting preference of the demand side; designing an energy perception scheduling generation scheme containing service occupation scheduling and connection degree scheduling; and solving an optimal Pareto solution through an adaptive NSGA-II algorithm integrated with MEOS, and outputting a service combination and a subtask execution plan. On the premise that QoS constraints are met, preheating energy consumption is averagely reduced by 2%-8%, service occupation conflicts are effectively avoided, benefits of a supply party and a demand party are balanced, the Pareto frontier is better, the method is suitable for cloud manufacturing scenes with preheating processes, and diversified requirements of personalized manufacturing tasks can be met.

Description

Technical Field

[0001] This invention relates to the fields of cloud computing, intelligent manufacturing, and resource scheduling optimization, and particularly to the collaborative scheduling optimization of cloud manufacturing energy consumption and service quality with tight task timing coupling. Background Technology

[0002] Cloud manufacturing (CMfg), as an intelligent and interconnected manufacturing paradigm, relies on cloud computing, the Internet of Things, and information technology to achieve resource sharing, collaboration among SMEs, and customized product production, playing a crucial role in reducing investment risks and addressing resource constraints. Cloud Manufacturing Service Selection and Scheduling (CMSSS) is a core component of the cloud manufacturing model, comprising four main steps: task decomposition, service discovery and matching, service selection, and task scheduling. It directly impacts resource allocation efficiency, production costs, and Quality of Service (QoS).

[0003] However, existing CMSSS technology has the following shortcomings: Ignoring the preheating process of manufacturing equipment: In multi-task collaboration scenarios, the intermittent nature of production tasks makes it impossible for manufacturing services to process in batches continuously, resulting in frequent waste of equipment preheating energy. However, most existing studies have not included preheating energy consumption in the optimization scope. Inadequate handling of service decentralization: In real production environments, service reservations and supplier autonomy can easily lead to the decentralization of manufacturing services, resulting in fragmented subtask execution times. Existing scheduling schemes are unable to effectively avoid service conflicts. Imbalance between supply and demand: Demand side focuses on QoS (time, cost, reliability), while suppliers focus on energy economy. However, existing optimization models often confuse the indicators of the interests of both parties, making it difficult to achieve a balance between the two objectives. Limited performance optimization: CMSSS is an NP-hard problem. While traditional service occupation scheduling methods (MOS) can shorten the total task time, they are not good at energy saving and lack optimality of Pareto front solutions. Although energy-efficient scheduling methods (EOS) take into account energy economy, they often come at the cost of longer task completion time, thus leading to the risk of delayed delivery.

[0004] Therefore, a cloud manufacturing energy consumption and service quality collaborative scheduling optimization method with tight task timing is needed to reduce preheating energy consumption while meeting QoS requirements, thereby maximizing the benefits for both supply and demand sides. Summary of the Invention

[0005] This invention aims to solve problems such as wasted preheating energy, service occupancy conflicts, supply and demand imbalances, and insufficient performance optimization in existing cloud manufacturing service selection and scheduling technologies.

[0006] Technical Solution: To solve the above-mentioned technical problems, according to one aspect of the present invention, more specifically, a method for optimizing the selection and scheduling of cloud manufacturing services, the specific technical solution is as follows: (I) Core Definitions and Assumptions Cloud manufacturing tasks: Production tasks submitted by the demand side can be broken down into multiple sub-tasks. Each subtask requires cloud manufacturing services to complete; Cloud manufacturing services: Suppliers package manufacturing equipment or resources as services and publish them to the cloud platform to complete corresponding sub-tasks; Preheating process: The preheating preparation process for equipment before the execution of cloud manufacturing services, including preheating time. and cooldown time It is a constant value; Service occupancy: The state in which the target cloud manufacturing service is occupied or reserved by other subtasks during the execution of a subtask; Assumptions: Subtasks and services are matched to form a candidate set after task decomposition; services are reserved on a first-come, first-served basis by the cloud platform; subtask execution cannot be interrupted; devices are shut down immediately when there are no tasks to be executed; logistics time and cost are proportional to the distance to the service location.

[0007] (ii) Multi-objective bilevel programming model Upper-level objective function (supplier energy economy): ,in This is the standardized value for total preheating energy consumption; Lower-level objective function (demand-side QoS): Min( × + × +(1- )×(1- )),in , , These are the standardized values ​​for total duration, total cost, and average reliability, respectively. , , For the corresponding weights; Constraints: The subtask service time is within the idle period of the cloud manufacturing service: , ( , (Left and right boundaries of the idle time slot); Total duration ≤ (The maximum time limit set by the demander); Total cost ≤ (The maximum budget set by the demander); Average reliability ≥ (Minimum reliability set by the demand side).

[0008] (III) Energy Sensing Dispatch Generation Scheme (MEOS) Service occupancy scheduling: For the four service occupancy scenarios, the idle time slot set is traversed in reverse to ensure that subtasks are scheduled to idle time slots that meet the service cycle requirements, thus avoiding execution conflicts; Continuity scheduling: Optimize the execution time arrangement based on the continuity of subtasks to make the execution time of the current subtask as continuous as possible with the preceding and following subtasks, shorten the warm-up time and reduce the warm-up energy consumption.

[0009] (iv) Adaptive NSGA-II Algorithm Integrating MEOS Encoding and initialization: An initial population is randomly generated using a vector encoding service combination. Subtask scheduling: Service occupancy scheduling and connectivity scheduling are executed sequentially, and various evaluation indicators are calculated; Fitness calculation: After standardization, input the bi-level objective function and solve; Crossover mutation: Generates new individuals according to a set probability; Elite selection: Select non-dominated solutions to form a new population, and output the optimal solution after iterating to a preset number of times.

[0010] The beneficial effects of this invention in sensing and optimizing the selection and scheduling of cloud manufacturing services are as follows: Significantly reduced preheating energy consumption: At the same QoS satisfaction level, compared with MOS and EOS, preheating energy consumption is reduced by an average of 2% to 8%, demonstrating a clear advantage in energy economy; Effectively avoid service contention conflicts: MEOS's service contention scheduling mechanism can handle the decentralized contention problem under multi-task collaboration and ensure the continuous execution of subtasks; Balancing the interests of both supply and demand: The two-tiered planning model focuses on both the energy-saving goals of suppliers and the QoS requirements of demanders, achieving optimization of both objectives; Optimized Pareto Front: Under seven QoS weight conditions, MEOS's Pareto front is superior to MOS and EOS, providing a better set of solutions; High applicability: It is suitable for cloud manufacturing scenarios with preheating processes and can meet the diverse needs of personalized manufacturing tasks. Attached Figure Description

[0011] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0012] Figure 1 A framework diagram for collaborative optimization of energy consumption and service quality in cloud manufacturing with tightly coupled task timing; Figure 2 A graph showing one availability and four occupancy statuses for cloud manufacturing services; Figure 3 Flowchart of subtask scheduling under cloud manufacturing service occupancy conditions; Figure 4 Correlation graph between current subtasks and service usage time for cloud manufacturing services; Figure 5 A flowchart for optimizing subtask scheduling for improved coherence; Figure 6 Flowchart of the adaptive NSGA-II algorithm for integration with MEOS, EOS, and MOS; Figure 7 Pareto optimal solutions for MEOS, MOS, and EOS under seven different reliability, cost, and time weights are plotted. Figure 8 A graph showing the average energy savings of MEOS, MOS, and EOS under seven reliability, cost, and time weights; Figure 9 Gantt plot of the optimal Pareto front solution for MEOS and EOS when completing the 19th manufacturing task; Detailed Implementation

[0013] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0014] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] Reference Figure 1-9 It can perceive the selection and scheduling optimization of cloud manufacturing services, including the following implementation steps: (a) Experimental parameter settings Core parameter configuration: number of subtasks n=10, capacity of candidate service set for each subtask m=5; number of idle time slots for manufacturing services l∈[0,n], task order time is 60; QoS constraints: Maximum time = 800, Maximum cost = 7000, Minimum reliability = 0.6; Algorithm parameters: Population size = 50, crossover probability = 0.9, mutation probability = 0.05, number of iterations = 2000; Dataset source: Simulated based on real energy consumption data from Hunan Valin Xiangtan Iron & Steel Co., Ltd., including key information such as the reliability of cloud manufacturing services, production time, manufacturing cost, preheating / cooling time, preheating energy consumption, service occupancy time, and service location.

[0016] (II) Implementation Steps S1. Task decomposition and candidate service matching For a specific customized machining and manufacturing task, it is broken down into 10 independent sub-tasks according to the cloud manufacturing task decomposition rules. Each sub-task needs to be executed sequentially and requires logistical coordination.

[0017] refer to Figure 1 (Diagram illustrating the interaction process between the demand side, supply side, and cloud platform in cloud manufacturing) After receiving a task, the cloud platform matches five candidate cloud manufacturing services that meet the capability requirements for each sub-task, forming a candidate service set. Each candidate service set contains all cloud manufacturing services available for that subtask. (like Include , , , , ).

[0018] S2. Construction of a Multi-Objective Bi-level Programming Model Based on the actual needs of the demand side, the QoS weight allocation is set as follows: time weight. =0.33, cost weight =0.33, reliability weight =0.34, which satisfies the condition. + + =1 weight constraint.

[0019] Constructing the upper-level objective function (supplier energy economy): Lower-level objective function (demand-side QoS): Meanwhile, the following constraints are specified: the subtask service time must be within the idle period of the cloud manufacturing service, the total duration must not exceed 800, the total cost must not exceed 7000, and the average reliability must not be less than 0.6.

[0020] S3, MEOS scheduling execution (including service occupancy scheduling and connectivity scheduling) Service usage scheduling: Reference Figure 2 (Diagram showing one available and four occupied scenarios in cloud manufacturing services) A ​​scheduling strategy that reverses the traversal of idle time slots is adopted to address the four service occupancy scenarios that may be encountered during subtask execution. For example... Figure 3 (The flowchart for subtask scheduling under cloud manufacturing service occupancy) shows that the system first determines the occupancy status of the target service when the subtask arrives. If there is an occupancy conflict, it searches backwards for an idle time slot that meets the service cycle requirements of the subtask to ensure that the subtask execution is conflict-free. For example, when... arrive At that time, if In Figure 2 In Case 2, the system reverses its idle time slot set to determine the occupancy status. Reschedule to the next available time slot that meets the service duration requirements.

[0021] Connectivity scheduling: based on Figure 4 The connectivity model shown in the correlation graph between the current subtask and the service's occupied time period on the cloud manufacturing service calculates the forward connectivity of the subtask. and backward connectivity This leads to the degree of connection. . refer to Figure 5 (Flowchart of subtask scheduling optimization), by adjusting the service start time of subtasks. and end time This improves the temporal continuity between subtasks and their adjacent counterparts, and shortens equipment warm-up time. For example, optimizing... exist The execution time should be adjusted so that its end time is as close as possible. The start time is adjusted to reduce the preheating energy consumption between the two processes.

[0022] S4. Adaptive NSGA-II algorithm optimization integrating MEOS Encoding and Initialization: Cloud manufacturing service compositions are encoded using vector encoding. The vector length is 10 (corresponding to 10 subtasks), and the vector element values ​​are 1-5 (corresponding to the 5 candidate service indices for each subtask). Fifty service composition vectors conforming to the encoding rules are randomly generated as the initial population.

[0023] Fitness calculation: For each individual in the population, perform the service occupancy scheduling and connectivity scheduling as described above in sequence, and calculate the total preheating energy consumption (according to the formula). Total duration (according to formula) Total cost (according to the formula) and average reliability (according to the formula) After standardizing these evaluation indicators, they are substituted into the bi-level objective function to calculate the fitness value.

[0024] Crossover, mutation, and elite selection: Individuals in the population are paired with each other at a crossover probability of 0.9 and crossover is performed. Mutation points are randomly selected from individuals at a mutation probability of 0.05 to generate the offspring population. (Reference) Figure 6 (Flowchart of the adaptive NSGA-II algorithm integrated with MEOS, EOS, and MOS) Merge the parent and offspring populations, select non-dominated solutions based on Pareto rank and crowding distance, retain elite individuals to form a new population, and repeat the iteration 2000 times.

[0025] S5. Output and Verification of the Optimal Solution After the iteration is complete, the optimal Pareto solution set is output, from which the scheme with the best overall performance (e.g., ...) is selected. Figure 9The MEOS scheduling scheme for the 19th manufacturing task in the process. The corresponding cloud manufacturing service combination is: , , , , , , , , , .

[0026] refer to Figure 9 (Gantt chart of the optimal Pareto front solution of MEOS and EOS when completing the 19th manufacturing task), outputting the execution order of subtasks, the start time and end time of each subtask on the corresponding cloud manufacturing service, and the logistics connection plan between subtasks. The specific performance indicators of this solution are: preheating energy consumption 518.74, average reliability 0.79, total cost 4955.89, and total duration 671.48, all of which meet the preset QoS constraints and energy-saving targets.

[0027] (III) Experimental Verification and Result Analysis Verification environment: Based on the above parameter settings, 20 repeated experiments were conducted under seven QoS weight conditions (including single index, dual index, and three index combination), and the best results of each experiment were selected for statistical analysis.

[0028] Result comparison: Pareto Front Comparison: Reference Figure 7 (Pareto optimal solutions for MEOS, MOS, and EOS under seven different reliability, cost, and time weights) MEOS outperforms traditional MOS and EOS in the Pareto front under all weight combinations, providing a better set of solutions.

[0029] Preheating energy consumption comparison: Reference Figure 8 (Average energy consumption savings under seven different reliability, cost, and time weights) At the same QoS satisfaction level, MEOS reduces preheating energy consumption by an average of 2% to 8% compared to MOS. Among them, the energy consumption reduction is as high as 6.38% in the dual-index (cost + time) scenario, and 3.3% in the three-index scenario with a satisfaction range of 0.4-0.6.

[0030] Conclusion: This embodiment verifies the effectiveness of the cloud manufacturing energy consumption and service quality collaborative scheduling optimization method with tightly coupled task timing. While meeting the QoS requirements of the demand side, it significantly improves the energy economy of the service provider and is applicable to cloud manufacturing scenarios with preheating processes.

[0031] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for collaborative scheduling and optimization of energy consumption and service quality in cloud manufacturing with tightly coupled task timelines, characterized in that, Includes the following steps: S1. Task decomposition and candidate service matching: Decompose the customized tasks submitted by the cloud manufacturing demander into n independent sub-tasks, and match each sub-task to form a candidate service set CMSS. Each candidate service set contains m cloud manufacturing services CMS that can complete the corresponding sub-task. S2. Construct a multi-objective bi-level programming model: The upper objective function aims to minimize the preheating energy consumption of the service provider's equipment, while the lower objective function optimizes based on the demand side's QoS weighted preferences (time, cost, reliability). Constraints include matching the subtask service time with the idle time of the cloud manufacturing service, the total time not exceeding the maximum time limit, the total cost not exceeding the maximum budget, and the reliability not lower than the minimum standard. S3. Design an Energy Sensing Dispatch Generation Scheme (MEOS): Service occupancy scheduling: For scenarios where cloud manufacturing services are occupied by multiple tasks in a distributed manner, the service idle time slot set is traversed in reverse to schedule subtasks to idle time slots that meet the service cycle requirements, thereby avoiding execution conflicts; Connectivity Scheduling: Quantify preheating energy consumption by measuring subtask connectivity, and calculate the forward connectivity between the current subtask and its preceding and following adjacent subtasks. and backward connectivity Optimize the temporal continuity of subtask execution to shorten warm-up time; S4. Solving using the adaptive NSGA-II algorithm integrated with MEOS: Encoding: The cloud manufacturing service combination is encoded in vector form, where the element length corresponds to the number of subtasks n, and the element value corresponds to the candidate service number; Initialization: Randomly generate an initial population of size 50, where each individual in the population is a cloud manufacturing service composite vector; Subtask scheduling and execution: For each individual in the population, service occupancy scheduling and connectivity scheduling are executed sequentially, and the total preheating energy consumption, total duration, total cost and average reliability are calculated; Fitness calculation: After standardizing the evaluation indicators, the fitness values ​​are calculated by substituting them into the two-level objective function; Crossover and mutation: Perform crossover and mutation operations with a crossover probability of 0.9 and a mutation probability of 0.05 to generate new individuals; Elite selection: Non-dominated solutions are selected based on Pareto rank and crowding distance, parent and offspring populations are merged to form a new population, and the optimal Pareto solution is output after 2000 iterations; S5. Output the optimal service selection and scheduling scheme: Based on the optimal Pareto solution, determine the cloud manufacturing service combination and the execution order and time arrangement of sub-tasks to ensure that QoS requirements and energy saving goals are met.

2. The cloud manufacturing service selection and scheduling optimization according to claim 1, characterized in that, The degree of connection between subtasks The calculation method is as follows ,in It is forward connectivity. It is backward connectivity: and These represent the preheating time and cooling time before and after the service process, respectively. and These are the start and end times of the service.

3. The cloud manufacturing service selection and scheduling optimization as described in claim 1, characterized in that, The total preheating energy consumption The calculation method is as follows ,in Serving cloud manufacturing Preheating energy consumption Serving cloud manufacturing The actual preheating energy consumption For subtasks in The degree of connection between the two.

4. The cloud manufacturing service selection and scheduling optimization as described in claim 1, characterized in that, The QoS weighted preference Including time weights Cost weight Reliability weight ,satisfy + + =1, the underlying objective function achieves multi-index aggregation optimization through weighted summation.

Images