Integrated Scheduling Method for Device Networks Considering Migration

Abstract

Integrated scheduling method for device network considering migration. The method of the present invention includes: the method uses a time beam algorithm to obtain a preselected set of network devices, and performs density clustering analysis according to requirements to divide areas where devices are relatively concentrated; in each area, a set of scheduling devices that are locally optimal for key migration paths is selected respectively; a reverse recursion strategy is adopted to determine the node scheduling order of non-key migration paths; a unified rendering strategy is adopted to determine the scheduling devices of processes in combination with the information of the already scheduled nodes.

Description

Technical Field

[0001] The present invention relates to a comprehensive scheduling method for a device network considering migration. Background Art

[0002] Aiming at the problem that the previous comprehensive scheduling methods did not consider the device network with common migration factors, a solution is proposed starting from the goal of obtaining a better scheduling result with less time used to complete the processing process tree: first, obtain a reference of the device set according to the basic information of the processing process tree; perform clustering analysis on the device set with coordinates according to the reference to divide regions; screen out a locally optimal scheduling device set within the regions; in this scheduling device set, schedule for the key migration path nodes and non-key migration path nodes respectively; compare the results of different regions to obtain the optimal solution. Summary of the Invention

[0003] The object of the present invention is to provide a comprehensive scheduling method for a device network considering migration. Aiming at the problem that the previous comprehensive scheduling methods did not consider the device network with common migration factors, a comprehensive scheduling method for a device network considering migration is proposed. The time beam algorithm strategy is used to analyze the device types and the corresponding device quantity requirements under the condition of as much parallelism as possible according to the information of the processing process tree; density clustering analysis is performed according to the requirements to divide regions where devices are relatively concentrated; in each region, a locally optimal scheduling device set for the key migration path is selected respectively; the reverse recursion strategy is adopted to determine the node scheduling order of the non-key migration path; the unified rendering strategy is adopted to determine the specific scheduling selection among the same devices in combination with the information of the already scheduled nodes.

[0004] The time beam algorithm strategy is to make the subsequent density clustering more purposeful and complete. According to the information of the process processing tree, the specific requirements for various devices under the condition of as much parallel processing as possible are deduced in advance. Advantage analysis: Aiming at the requirement of refined screening of the network device preselection set, the efficiency of the subsequent screening and the process of determining the device set is effectively improved, while reducing the overall algorithm complexity and avoiding wasting time due to ineffective screening.

[0005] The reverse recursion strategy is to determine the overall working hours for the operation set on the key migration path of the forward scheduling, and then increase the total duration by making up time as much as possible to reduce the conflict of the non-key migration path nodes. Advantage analysis: For the non-key path, giving priority to processing those with more migration times is beneficial to reducing the overall time consumption increased by making up time for the non-key path. Prioritizing the selection of the operation far from the completion node is beneficial to achieving the load balance of the device and reducing conflicts.

[0006] The integrated rendering strategy is to select devices with less conflict possibility and less migration time as much as possible when there are several devices of the same type to choose from. Advantage analysis: For the same type of processing devices that can be selected, preferentially select the devices with short migration time, which is beneficial to making the non-critical path plus the long migration time within the time of the critical migration path, reducing the make-up time. Preferentially select the devices with fewer occurrences of easily conflicting processes in the neighborhood, which is beneficial to reducing the time increase caused by conflicts and also beneficial to achieving the load balance of the devices.

[0007] The above object is achieved by the following technical solutions:

[0008] A comprehensive scheduling method for a device network considering migration, which mainly includes the following steps: Using the time beam algorithm strategy to analyze the device types and the corresponding device quantity requirements under the condition of as much parallelism as possible according to the information of the processing technology tree; Conducting density clustering analysis according to the requirements to divide the areas where the devices are relatively concentrated; Selecting the locally optimal scheduling device set for the critical migration path in each area respectively; Using the reverse recursion strategy to determine the node scheduling order of the non-critical migration path; Using the integrated rendering strategy to determine the specific scheduling selection among the same devices in combination with the information of the scheduled nodes.

[0009] For the comprehensive scheduling method for a device network considering migration, the specific implementation steps of the scheduling method are as follows:

[0010] Step 1: Obtain the processing technology tree, and calculate the critical migration path and the critical migration device type according to the specific information of the processing technology tree;

[0011] Step 2: And use the time beam algorithm to obtain the reference of the preselected set of network devices, so as to obtain the complete quantity of devices for the clustering condition;

[0012] Step 3: After collecting the information of the device set, use the complete quantity of devices as the clustering density for area division;

[0013] Step 4: Conduct scheduling within a single area, select the core nodes of the critical migration device type for device completeness inspection, and after screening out the qualified nodes, form the preselected node set of network devices with their neighboring nodes;

[0014] Step 5: For the processing procedures other than the critical migration path, conduct reverse recursion according to the descending order of the route migration times as the priority, and screen the same type of devices according to the integrated rendering strategy during the process;

[0015] Step 6: Obtain the optimal scheduling plan within the area, and obtain the final plan by comparing the plans of each area;

[0016] Step 7: Output the Gantt chart of the scheduling result.

[0017] The device network comprehensive scheduling method considering migration and the time bundle algorithm are based only on the information of the processing technology tree, do not limit the number of processing equipment, rely on the timeline, and perform virtual scheduling under the premise of satisfying parallelism as much as possible.

[0018] The device network comprehensive scheduling method considering migration, and the reverse recursive strategy, after forward scheduling the process set on the critical migration path, begin reverse scheduling the process set on the non-critical migration path. The scheduling order priority is arranged in descending order of the number of migrations of the migration path.

[0019] The aforementioned integrated device network scheduling method, which considers migration, and the integrated rendering strategy, are designed to dynamically select the appropriate device to minimize overall scheduling time when there are multiple devices of the same type available. After determining the rendering factor, the one with the largest coefficient is prioritized unless there is a time conflict.

[0020] Beneficial effects:

[0021] 1. The present invention analyzes the processing technology tree information and adopts a time bundle algorithm strategy. It derives in advance the specific requirements for various types of equipment while processing in parallel as much as possible, and targets the requirements of fine-grained screening of the pre-selected set of network equipment. It effectively improves the efficiency of subsequent screening and equipment set determination processes, reduces the overall algorithm complexity, and avoids wasting time due to invalid screening. It can indirectly advance the completion time of the process, thereby shortening the completion time of the entire product.

[0022] The present invention adopts a reverse recursive strategy for non-critical path processes in the scheduling process, and reduces the total time increased by overtime due to conflicts in non-critical migration path nodes as much as possible. For non-critical paths, those with a large number of migrations are given priority, which is beneficial to reducing the overall time consumption increased by overtime on non-critical paths, and giving priority to processes far away from completion nodes, which is beneficial to achieving load balancing of equipment and reducing conflicts, solving the process selection problem of non-critical paths, and making the overall completion time as early as possible.

[0023] The present invention adopts an integrated rendering strategy for the same type of equipment in the scheduling process, giving priority to equipment with low conflict possibility, short migration time, and few neighboring conflict-prone processes, thereby reducing the impact of processes on non-critical migration paths in the process tree on product completion time, shortening the product completion time overall, and also solving the problem of selecting the same equipment at a certain moment. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Attachment Figure 1 It is a process tree diagram of the present invention.

[0025] Attachment Figure 2 It is the device information of the present invention.

[0026] Appendix Figure 3 is the clustering analysis result of the device information shown in the appendix of the present invention. Figure 2

[0027] Appendix Figure 4 is the Gantt chart of the scheduling result of the process tree shown in the appendix of the present invention. Figure 1 Detailed implementation manners

[0028] Example 1:

[0029] A comprehensive scheduling method for a device network considering migration, characterized in that: first, obtain a reference of a device set according to the basic information of a processing process tree; divide regions by performing a clustering analysis on the device set with coordinates according to the reference; screen out a locally optimal scheduling device set within the region; in the scheduling device set, schedule for critical migration path nodes and non-critical migration path nodes respectively; compare the results of different regions to obtain an optimal solution.

[0030] Example 2:

[0031] For the above-mentioned comprehensive scheduling method for a device network considering migration, the specific implementation steps of the scheduling method are as follows:

[0032] Step 1: Obtain a processing process tree, and calculate a critical migration path and critical migration device types according to the specific information of the processing process tree;

[0033] Step 2: And use a time beam algorithm to obtain a reference of a preselected set of network devices, so as to obtain the number of device completeness of clustering conditions;

[0034] Step 3: After collecting device set information, divide regions with the number of device completeness as the clustering density;

[0035] Step 4: Perform scheduling within a single region, select core nodes of critical migration device types to check device completeness, and after screening out qualified nodes, form a preselected set of network device nodes with their neighboring nodes;

[0036] Step 5: For the processing procedures other than the critical migration path, perform reverse recursion with the number of route migrations in descending order as the priority, and screen the devices of the same type according to the integrated rendering strategy during the process;

[0037] Step 6: Obtain the optimal scheduling plan within the region, and obtain the final plan by comparing the plans of each region;

[0038] Step 7: Output the Gantt chart of the scheduling result.

[0039] Example 3:

[0040] The above-mentioned comprehensive scheduling method for equipment network considering migration and the time bundle algorithm are based only on the information of the processing technology tree, do not limit the number of processing equipment, rely on the timeline, and perform virtual scheduling under the premise of satisfying parallelism as much as possible.

[0041] In the aforementioned integrated scheduling method for equipment networks that consider migration, the reverse recursive strategy involves forward scheduling of the set of processes on the critical migration path, followed by reverse scheduling of the set of processes on the non-critical migration path. Scheduling priority is based on descending order of the number of migrations along the migration path. For processes with the same status, priority should be given to processes that started earlier in the process tree, i.e., those farther away from the completion process. This is because the time span of such nodes is less flexible than that of nodes of the same status closer to the completion process, and the cost of not prioritizing them is higher.

[0042] The aforementioned integrated device network scheduling method, which considers migration, includes a unified rendering strategy. To minimize overall scheduling time when there are multiple devices of the same type available, the scheduling strategy should centralize processes while minimizing conflicts. This strategy dynamically selects appropriate devices. After determining the rendering factor, the one with the highest coefficient is prioritized unless there is a time conflict.

[0043] Example 4:

[0044] The aforementioned comprehensive scheduling method for equipment networks that considers migration, and the time-bundle algorithm, are based solely on information from the process tree, without limiting the number of processing equipment. They rely on timelines and perform virtual scheduling while ensuring maximum parallelism. The overall scheduling process begins with the critical migration path, considering only the constraints of the processing equipment type. Once a conflict arises between non-serial sequential processes, new processing equipment of the corresponding equipment type is immediately assigned. The scheduling priority of non-critical path processes is implemented according to a reverse recursive strategy until the end of the process tree. After all processes have been scheduled, the scheduling time for non-critical migration path processes may be longer, requiring a time-compensation operation, where the start time of the earliest scheduled process is used as the overall start time.

[0045] Example 5:

[0046] The described comprehensive scheduling method for device networks considering migration, the reverse recursion strategy. After scheduling the operation sets on the critical migration paths in the forward direction, start scheduling the operation sets on the non-critical migration paths in the reverse direction. The scheduling order priority is arranged in descending order according to the number of migrations of the migration paths. In the case of having the same status, the operations that start earlier in the processing technology tree should be scheduled first, that is, the operations far from the completed operations, because the time selection span of such nodes is not as flexible as that of the same-status nodes close to the completed operations. If not scheduled first, once a conflict occurs, the cost will be relatively high. During the scheduling process, the integrated rendering strategy is used as a reference for selecting devices. First, try to schedule in sequence. If all result in conflicts and cannot be directly scheduled, then select a plan with less impact on the zero scale of the original scheduling plan. After all operations are scheduled, a time compensation operation is required.

[0047] Example 6:

[0048] The described comprehensive scheduling method for device networks considering migration, the integrated rendering strategy. In the case of having several devices of the same type available for selection, in order to minimize the overall scheduling time as much as possible, the scheduling strategy should let the operations be processed as concentrated as possible while reducing the occurrence of conflicts. For this purpose, this strategy is designed to dynamically select appropriate devices. The current scheduling operation is Ai, and currently there are k devices of the same type, X1, X2,..., Xk available for selection, and the rendering factor is: [[ID=Y]]

[0049] μi = w0X0 + ∑[w1(1 / Sij) - w2(nij / ni) + w3(Nij / Ni)]Xj

[0050] Where wi (i = 0, 1, 2, 3) are the weights corresponding to the evaluations. The specific weights need to be analyzed and determined according to the specific purpose. The first item is the situation of selecting the same processing device for the subsequent operation. Then, for the remaining devices, calculate according to the reciprocal of the scheduling time between devices, that is, the first item of the accumulation, the proportion of the number of operations prone to conflicts in the neighborhood in the device, that is, the second item of the accumulation, and the processing proportion of the whole on the device, that is, the third item of the accumulation. After obtaining the rendering factor, select the one with a large priority selection coefficient, unless a time conflict occurs.

[0051] Example 7:

[0052] The above-mentioned comprehensive scheduling method for device networks considering migration, as shown in the appendix Figure 1 is a legend of the process tree of a complex product. There are 20 operations in the figure, and 3 types of processing devices are required. Each operation can be processed on the corresponding device type, and the attribute information of each operation is different. The meaning of the symbols in the rectangular frame is: product operation name / processing device type / processing time. The product needs to be processed on the device set distributed in a scattered manner.

[0053] The following will be combined with the appendix Figure 1The process tree legend in it is used to illustrate the specific execution process of this method.

[0054] Example 8:

[0055] For the above-mentioned integrated scheduling method of device network considering migration, the following will use this scheduling method to perform scheduling on the process tree legend in the appendix. The specific operation steps of the method are as follows: Figure 1 The process tree legend in it is used to perform scheduling. The specific operation steps of the method are as follows:

[0056] According to the process tree, we can obtain that the critical migration paths are A17, A12, A8, A4, A2, A1, and the critical migration device type is 1. The results of using the time beam algorithm strategy are that the numbers of type 1, 2, and 3 devices are 2, 3, and 2 respectively. Based on the above results as the density and a suitable radius (this time, 0.13 length units are selected as the clustering radius) for density clustering. The clustering results are divided into two node sets, and the clustering results are as shown in the appendix. Figure 3 as shown.

[0057] Process the nodes in set 1. The neighborhood node set of 28 meets the requirements, and its scheduling takes 11.32 unit time. Recursively backward the processes on the non-critical path, and use the integrated rendering strategy to calculate the rendering factor. The coefficients used are 0.3, 0.3, 0.2, and 0.2 respectively. First, process the devices from A3 to A1, and there are devices 15, 23, 24 of type 3, and their corresponding coefficients are 0.0300, 0.0359, and 0.0600 respectively. Since the coefficient of device 24 is large, it is preferentially selected as the scheduling option. Continue to calculate the subsequent nodes until all processes are scheduled. The Gantt chart is as shown in the appendix. Figure 4 as shown.

[0058] Perform the same process on set 2. The shortest path of the critical migration route in set 2 takes 18.46 unit time. At this time, it has exceeded the time used for the complete scheduling of set 1, so the result obtained from set 1 is the result.

[0059] Example 9:

[0060] For the above-mentioned integrated scheduling method of device network considering migration, example comparison:

[0061] Since there has been no previous research on the integrated scheduling method of device network considering migration proposed by the present invention, no suitable examples have been found for comparative analysis. However, it can be seen from the scheduling results in different regions in Example 8 that the total processing time of the product in set 1 is 29.32, while the path length of the product process tree calculated in set 2 is already 36.46 without additional time. And 29.32 is less than 36.46. Therefore, this method has obtained better scheduling results, and this method has solved the problem of integrated scheduling of device network to a certain extent.

[0062] Therefore, the scheduling method proposed by the present invention is to solve the problem of considering migration in the comprehensive scheduling of device networks.

Claims

1. An integrated scheduling method for a device network considering migration, characterized in that: A time-bundle algorithm is used to derive a pre-selected set of network devices, and a density cluster analysis method is used to divide dense areas. Within the key area, the locally optimal set of scheduling devices is screened based on the scheduling time of the key migration path. For the problem of non-critical migration path nodes, a reverse recursive strategy is used to determine the scheduling order of non-critical migration path nodes. An integrated rendering strategy is used to determine the process equipment. The specific implementation steps of the scheduling method are as follows: Step 1: Obtain the processing technology tree and calculate the key migration path and key migration equipment type based on the specific information of the processing technology tree; Step 2: Use the time beam algorithm to obtain the pre-selected set of network devices, thereby obtaining the number of devices with clustering conditions; Step 3: After collecting the device set information, the area is divided using the number of device completeness as the cluster density; Step 4: Scheduling is performed within a single region. Core nodes of key migration device types are selected for device integrity checks. After qualified nodes are screened, they are combined with their domain nodes to form a pre-selected set of network device nodes. Virtual scheduling is performed on the key migration paths within these nodes. The device set with the shortest migration time is selected as the pre-selected set of network devices. Step 5: For the processing steps other than the key migration path, reverse recursion is performed based on the descending order of the number of migrations as the priority. During this process, the same type of equipment is screened according to the integrated rendering strategy; Step 6: Obtain the optimal scheduling plan within the region and obtain the final plan by comparing the plans of each region; Step 7: Output the scheduling result Gantt chart.

2. The device network integrated scheduling method considering migration according to claim 1, characterized in that: The time bundle algorithm is based only on the information of the processing technology tree, does not limit the number of processing equipment, relies on the timeline, and performs virtual scheduling under the premise of satisfying parallelism as much as possible. The overall scheduling process starts from the critical migration path and only considers the constraints of the processing equipment type. Once a conflict between non-serial sequential and sequential processing procedures occurs, new processing equipment of the corresponding equipment type is immediately allocated. The scheduling priority of non-critical path procedures is implemented according to the reverse recursive strategy until the processing of the processing technology tree is completed. After all procedures are scheduled, the scheduling time of non-critical migration path procedures may be longer. At this time, a time-compensation operation is required, that is, the start time of the earliest scheduled procedure is used as the overall start time.

3. The device network integrated scheduling method considering migration according to claim 1, characterized in that: The reverse recursive strategy described above forward schedules the set of processes on the critical migration path, and then starts to reversely schedule the set of processes on the non-critical migration path. The scheduling order priority is arranged in descending order according to the number of migrations of the migration path. During the scheduling process, the integrated rendering strategy is used as a reference for selecting equipment, and scheduling is tried in sequence first.

4. The device network integrated scheduling method considering migration according to claim 1, characterized in that: The integrated rendering strategy mentioned above is designed to dynamically select appropriate devices to meet the established scheduling goals in the case of having several options. The calculation method of the rendering factor is as follows: Assume that the current process is A i , at this time, there are X1, X2,..., X k available for selection. First, prioritize the devices where the subsequent processes are located, that is, in the case of continuous processing. Then, for the remaining devices, calculate according to the reciprocal of the scheduling time between devices, the proportion of the number of neighboring conflict-prone processes on the devices, and the overall processing proportion on the devices. Each item has its own weight. After obtaining the rendering factor, the one with a larger priority selection coefficient is preferred, unless there is a time conflict

Images