CN114266168A - Load scheduling parameter optimization method for shared node - Google Patents
Load scheduling parameter optimization method for shared node Download PDFInfo
- Publication number
- CN114266168A CN114266168A CN202111622496.2A CN202111622496A CN114266168A CN 114266168 A CN114266168 A CN 114266168A CN 202111622496 A CN202111622496 A CN 202111622496A CN 114266168 A CN114266168 A CN 114266168A
- Authority
- CN
- China
- Prior art keywords
- load
- shared
- node
- nodes
- sharing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application belongs to the technical field of shared device scheduling control methods, and particularly relates to a shared node load scheduling parameter optimization method. The method comprises the following steps: determining basic data of a sharing node in a system; determining a sharing node needing load scheduling in a sharing system; constructing a load transfer optimal model under natural circulation; constructing a shared node utilization rate and current load rate calculation model; constructing a system effective load transfer target; constructing a load scheduling optimization model of the shared nodes; and solving the load scheduling optimization model of the shared nodes, and determining the load transfer target and the transfer path of each load node. The shared node load scheduling parameter optimization method based on the application can efficiently and simply obtain necessary parameters required by load scheduling, can effectively compress the data volume required to be processed and analyzed in the data calculation process, and reduces the system prediction analysis cost, so that the method is more convenient and efficient to apply and implement, and has smaller requirements on software and hardware.
Description
Technical Field
The application belongs to the technical field of shared device scheduling control methods, and particularly relates to a shared node load scheduling parameter optimization method.
Background
With the continuous development of the sharing industry, a large amount of sharing equipment is put into the market, and due to the influence of the utilization rate of the sharing equipment on the use of the sharing equipment and various external factors, the sharing equipment is difficult to be uniformly distributed and not utilized, in most sharing systems, a sharing node for providing service supply needs to be dynamically scheduled with an external sharing node except for the area where the sharing node is located, so as to meet the periodic use requirement, the load scheduling is generally scheduled by a sharing service provider according to the real-time requirement of the sharing equipment, so that hysteresis exists, and the increasingly competitive service market is difficult to achieve.
Disclosure of Invention
The application aims to provide a shared node load scheduling parameter optimization method for planning and designing load scheduling targets in advance and optimizing load scheduling data.
In order to achieve the purpose, the following technical scheme is adopted in the application.
A load scheduling parameter optimization method for a shared node comprises the following steps:
step one, determining basic data of shared nodes in a system
Specifically, according to a system scheduling range, coordinates are formulated, the coordinate positions of all sharing nodes are determined, and the maximum load capacity of each sharing node is determined;
step two, determining a sharing node needing load scheduling in the sharing system;
for the load state of any node j, constructing a load balance target in a period:
wherein f isjIs the current load rate of the sharing node; f. ofjmaxIs the maximum load rate of the sharing node;
empirical parameter θj，(0≤θjLess than or equal to 1) is obtained by counting the actual operation process of each sharing node j for a period of time, and the load rate of the sharing node is satisfied
The utilization rate index of the shared node can be ensured to meet the requirement in the range; based on the load balance target, screening the sharing nodes which do not meet the balance target as the sharing nodes needing load scheduling;
step three, constructing a load transfer optimal model under natural circulation
Specifically, the natural load circulation situation of each load node in the period is counted, and the probability mark k of the load circulation is usedij∈[0,1]Representing the probability of load transfer among different sharing nodes, and constructing an optimization target based on the minimum natural load circulation path among the sharing nodes
N is the set of all sharing nodes needing load transfer;
step four, constructing a shared node utilization rate and current load rate calculation model
Building shares based on load balancing objectives within a cycleUtilization ratio mu of node jjWith the current load factor fjThe relationship of (1) is:
step five, constructing a system effective load transfer target
System payload transfer rate target
Wherein R iskIs the total number of all shared devices in the load I/O k, J is the total number of shared nodes in the load I/O k, mujThe utilization rate of the shared node j inside the load input and output end k is obtained;
step six, constructing a load scheduling optimization model of the shared nodes
And seventhly, solving the load scheduling optimization model of the shared nodes, and determining load transfer targets and transfer paths of the load nodes.
A further improvement on the foregoing method for load scheduling parameters of shared nodes or the step three in the preferred embodiment further includes simplifying clustering of shared nodes that need load scheduling, specifically, including:
3.1, determining load transfer distances among the sharing nodes, and positioning the two sharing nodes with the largest load transfer distances; setting the two sharing nodes as a clustering center;
3.2, rejecting the sharing nodes which are set as the clustering centers, and determining the sharing nodes which maximize the product of the distance between the sharing nodes and the known clustering centers from the rest sharing nodes as new clustering centers;
3.3, repeating the step 3, and selecting a plurality of clustering centers;
3.4, after the shared node serving as the clustering center is determined, calculating the distance from the shared node of other non-clustering centers to the shared node corresponding to each clustering center, and dividing all the shared nodes of the non-clustering centers into a class corresponding to the nearest clustering center according to the standard of the minimum load transfer distance;
3.5, calculating the coordinates of all shared nodes in each category, carrying out averaging treatment to obtain new coordinates as a new clustering center, repeating the step 5, and clustering the shared nodes of all non-clustering centers again on the basis of the new clustering center until the maximum iteration times or the average error reaches the standard requirement;
and 3.6, acquiring a final clustering result, and taking the coordinates corresponding to each clustering center as a load node which needs to be subjected to load transfer finally.
The beneficial effects are that:
the shared node load scheduling parameter optimization method based on the application can efficiently and simply obtain necessary parameters required by load scheduling, can effectively compress the data volume required to be processed and analyzed in the data calculation process, can reply and reduce the pressure on system calculation power on the original basis by combining with a set improved clustering scheme, improves the data acquisition speed, and reduces the system prediction analysis cost, so that the method is more convenient and efficient to apply and implement, and has smaller requirements on software and hardware.
Drawings
FIG. 1 is a graph of utilization of a shared node j versus a current load rate.
Detailed Description
For any sharing system, the distance between the nodes directly affects the endurance of the matched sharing equipment, the size of the service range and the stability utilization rate of the service, and in most cases, under the condition that the number V of the sharing nodes in the service area and the area is known, the distance d between the sharing node i and the sharing node jijThe smaller the requirement the better, wherein there may be multiple load transfer paths between multiple shared nodes within the system, so on the basis of the foregoing, and thus on the basis of the foregoing, the shortest distance one of the multiple load transfer paths is first determined, for several load scheduling needs to be performedThe shared node is used for ensuring that the scheduling distance required by executing the load scheduling is minimum in order to ensure that the cost of the system is minimum when the load scheduling is carried out, and is essentially a TSP problem, and based on the TSP problem, the shared node load scheduling parameter optimization method mainly comprises the following steps:
step one, determining basic data of shared nodes in a system
The basic data comprises available load transfer paths based on the coordinate position of a sharing node of the area where the system is located;
determining n sharing nodes needing load scheduling in a sharing system;
in order to ensure that the load rate of each contributing node in the area is the highest as much as possible, for the load state of any node j, although the full load can ensure that the load rate of the current sharing node at a specific time point reaches the highest, due to the transfer-out and transfer-in of the load among the sharing nodes, when a certain sharing node is in the full load state, the load transferred in by the previous sharing node cannot be quickly accessed, and meanwhile, the load rate of the next sharing node is adversely affected; therefore, for the whole system, the optimal load rate of each sharing node is not the maximum load rate, and generally, when the real-time load is the maximum load fjmaxWhen the number of the shared nodes is half of the number of the shared nodes, the service utilization rate of the shared nodes can be the best;
in practice, because each sharing node in the sharing system is in a dynamic regulation process, the optimal load rate f of any single sharing node j is the whole systemjNot only the maximum load f related to the nodejmaxConsidering the optimal load factor f of the shared node jjInvolving continuous admittance of load between adjacent shared nodes, which is actually a dynamic regulation process, so that the empirical parameter theta obtained through data statisticsj，(0≤θj1) or less; thus for a load balancing goal within a cycle can be expressed as:
when in use
Indicating that the current load rate of the shared node j is small, and calling in the load to improve the system rate; when in use
Indicating that the current load rate of the shared node j is large and needs to be called out; when in use
Representing that the current load rate of the shared node j meets the requirement;
wherein the empirical parameter θj，(0≤θjLess than or equal to 1) is obtained by counting the actual operation process of each sharing node j for a period of time when the load rate of the sharing node is within
The technical indexes such as the utilization rate of the shared node and the like can meet the requirements in the range;
for a large-scale sharing system, because the number of sharing nodes is large, a large amount of data operation processing work can be caused, the data processing efficiency is influenced, considering that a large amount of sharing equipment put in the same sharing node is closely related to the activity range of a client when in use, the circulation use of the sharing equipment presents the characteristic of aggregation, and simultaneously influences the load transfer characteristics in each sharing node, which means that the probability of the load transfer executed by the two sharing nodes with longer load transfer distance is far lower than the probability of the load transfer between the two load nodes with smaller load transfer distance, therefore, in order to obtain the optimal solution of the optimization objective function, the cluster processing can be carried out on the shared nodes needing load scheduling, and the reasonability of a solving result is optimized while the data processing scale is reduced by considering the aggregation flow of the shared equipment in a specific shared node; specifically, the following clustering scheme may be employed:
1. determining load transfer distances among the n sharing nodes, and determining two sharing nodes with the largest load transfer distances; setting the two sharing nodes as a clustering center;
2. rejecting shared nodes set as cluster centers, and determining max (d) among the remaining shared nodes which is the product of the distance from the known cluster centers1×d2×d3...) as a new cluster center; diMeans the distance from each cluster center of the ith;
3. repeating the step 3, and selecting a plurality of clustering centers;
4. after the shared node serving as the clustering center is determined, calculating the distance from the shared node of other non-clustering centers to the shared node corresponding to each clustering center, and dividing all the shared nodes of the non-clustering centers into a class corresponding to the closest clustering center according to the standard of the minimum load transfer distance;
5. calculating the coordinates of all shared nodes in each category, carrying out averaging treatment to obtain new coordinates as a new clustering center, repeating the step 5, and re-clustering the shared nodes of all non-clustering centers on the basis of the new clustering center until the maximum iteration times or the average error reaches the standard requirement;
in actual operation, instead of natural circulation of load between any two sharing nodes, circulation with selectivity and directivity is performed according to actual conditions, and it is assumed that a load circulation probability flag k is usedij∈[0,1]Denotes k ij1 then indicates that there is a transition from shared node i to shared node j, kijIf the mark 0 indicates that the shared node i does not transfer to the shared node j, the optimization target based on the minimum load scheduling path between the shared nodes can be obtained
With the aforementioned range load rate as a standard, a curve of the utilization rate of the sharing node j and the current load rate can be plotted as shown in fig. 1, that is, the utilization rate of the sharing node j can be expressed as
In the sharing system, because of the homogeneity of the sharing equipment, the transferring-in and transferring-out operations of two sharing equipment with different IDs in the same sharing node j in a short time, when the continuous transferring of the load which can be regarded as a single sharing equipment is carried out and the interaction between the sharing equipment and the sharing node j is ignored, mu is at the momentjThe shared node corresponding to 1 can be regarded as balance without load transfer, and the optimal scheme of analyzing the system load scheduling is that only mu needs to be consideredjNot equal to 1 corresponding sharing node;
based on the steps, a plurality of clustering centers are obtained, the distribution and the characteristics of the clustering centers directly influence the result and the efficiency of the system during load transfer, on the basis, all shared nodes in each cluster can be fuzzified into a load input and output end, the input and output relations of the load input and output ends are internally self-consistent and are only relevant to other external load input and output ends; for the whole sharing system, the average system load rate of the load input and output ends can be adopted to simplify the effective utilization rate of the calculation sharing system; the load transferring-out operation usually only involves the position transfer of the sharing device, the load transferring-in operation needs to consider the availability of the sharing device and the content of replanning the service scheme, and the like, in a stable sharing system, for K load input and output ends needing load scheduling, the average load rate is calculated by stress to be the maximum, namely the effective load transfer rate target of the system
Wherein R iskIs the total number of all shared devices in the load I/O k, J is the total number of shared nodes in the load I/O k, mujThe utilization rate of the shared node j inside the load input and output end k is obtained;
then based on the foregoing, an optimization objective model for load scheduling can be constructed:
solving is carried out based on the optimization target model, so that the optimal load transfer parameter can be determined, namely the utilization rate mu corresponding to each load node after load transfer is carried outjAnd is based on mujCalculating reversely to obtain a target load rate target, and scheduling the load according to the difference between the target load rate and the previous load rate
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the protection scope of the present application, and although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application.
Claims (5)
1. A shared node load scheduling parameter optimization method is characterized by comprising the following steps:
step one, determining basic data of shared nodes in a system
Specifically, according to a system scheduling range, coordinates are formulated, the coordinate positions of all sharing nodes are determined, and the maximum load capacity of each sharing node is determined;
step two, determining a sharing node needing load scheduling in the sharing system;
for the load state of any node j, constructing a load balance target in a period:
wherein f isjIs the current load rate of the sharing node; f. ofjmaxIs the maximum load rate of the sharing node;
empirical parameter θj，(0≤θjLess than or equal to 1) is obtained by counting the actual operation process of each sharing node j for a period of time, and the load rate of the sharing node is satisfied
The utilization rate index of the shared node can be ensured to meet the requirement in the range; based on the load balance target, screening the sharing nodes which do not meet the balance target as the sharing nodes needing load scheduling;
step three, constructing a load transfer optimal model under natural circulation
Specifically, the natural load circulation situation of each load node in the period is counted, and the probability mark k of the load circulation is usedij∈[0,1]Representing the probability of load transfer among different sharing nodes, and constructing an optimization target based on the minimum natural load circulation path among the sharing nodes
N is the set of all sharing nodes needing load transfer;
step four, constructing a shared node utilization rate and current load rate calculation model
Establishing the utilization rate mu of the shared node j based on the load balance target in the periodjWith the current load factor fjThe relationship of (1) is:
step five, constructing a system effective load transfer target
System payload transfer rate target
Wherein R iskIs the total number of all shared devices in the load I/O k, J is the total number of shared nodes in the load I/O k, mujThe utilization rate of the shared node j inside the load input and output end k is obtained;
step six, constructing a load scheduling optimization model of the shared nodes
And seventhly, solving the load scheduling optimization model of the shared nodes, and determining load transfer targets and transfer paths of the load nodes.
2. The method for optimizing load scheduling parameters of shared nodes according to claim 1, wherein the third step further comprises cluster simplification of the shared nodes requiring load scheduling, specifically comprising:
3.1, determining load transfer distances among the sharing nodes, and positioning the two sharing nodes with the largest load transfer distances; these two shared nodes are set as cluster centers.
3. The method for optimizing load scheduling parameters of shared nodes according to claim 2, wherein the simplifying clustering of the shared nodes requiring load scheduling further comprises:
and 3.2, rejecting the shared nodes which are set as the cluster centers, and determining the shared nodes which maximize the product of the distances between the shared nodes and the known cluster centers as new cluster centers.
4. The method for optimizing load scheduling parameters of shared nodes according to claim 3, wherein the simplifying clustering of the shared nodes requiring load scheduling further comprises:
3.3, repeating the step 3.1, and selecting a plurality of clustering centers;
and 3.4, after the shared node serving as the clustering center is determined, calculating the distance from the shared node of other non-clustering centers to the shared node corresponding to each clustering center, and dividing all the shared nodes of the non-clustering centers into a class corresponding to the closest clustering center according to the standard of the minimum load transfer distance.
5. The method for optimizing load scheduling parameters of shared nodes according to claim 4, wherein the cluster simplification of the shared nodes requiring load scheduling further comprises:
3.5, calculating the coordinates of all shared nodes in each category, carrying out averaging treatment to obtain new coordinates as a new clustering center, repeating the step 3.4, and clustering the shared nodes of all non-clustering centers again on the basis of the new clustering center until the maximum iteration times or the average error reaches the standard requirement;
and 3.6, acquiring a final clustering result, and taking the coordinates corresponding to each clustering center as a load node which needs to be subjected to load transfer finally.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111622496.2A CN114266168A (en) | 2021-12-28 | 2021-12-28 | Load scheduling parameter optimization method for shared node |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111622496.2A CN114266168A (en) | 2021-12-28 | 2021-12-28 | Load scheduling parameter optimization method for shared node |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114266168A true CN114266168A (en) | 2022-04-01 |
Family
ID=80830702
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111622496.2A Pending CN114266168A (en) | 2021-12-28 | 2021-12-28 | Load scheduling parameter optimization method for shared node |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114266168A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115936240A (en) * | 2022-12-23 | 2023-04-07 | 暨南大学 | Shared bicycle demand forecasting and delivering scheduling method |
-
2021
- 2021-12-28 CN CN202111622496.2A patent/CN114266168A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115936240A (en) * | 2022-12-23 | 2023-04-07 | 暨南大学 | Shared bicycle demand forecasting and delivering scheduling method |
| CN115936240B (en) * | 2022-12-23 | 2024-03-15 | 暨南大学 | Shared bicycle demand prediction and delivery scheduling method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109885397B (en) | Delay optimization load task migration algorithm in edge computing environment | |
| CN109617826B (en) | Storm dynamic load balancing method based on cuckoo search | |
| CN107357652B (en) | Cloud computing task scheduling method based on segmentation ordering and standard deviation adjustment factor | |
| CN112817728B (en) | Task scheduling method, network device and storage medium | |
| CN111212106A (en) | Edge computing task processing and scheduling method and device in industrial internet environment | |
| CN103988179A (en) | Optimization mechanisms for latency reduction and elasticity improvement in geographically distributed datacenters | |
| CN114564312A (en) | Cloud edge-side cooperative computing method based on adaptive deep neural network | |
| CN113259469A (en) | Edge server deployment method, system and storage medium in intelligent manufacturing | |
| CN112261120A (en) | Cloud-side cooperative task unloading method and device for power distribution internet of things | |
| CN113553160A (en) | Task scheduling method and system for edge computing node of artificial intelligence Internet of things | |
| CN114266168A (en) | Load scheduling parameter optimization method for shared node | |
| Zhang et al. | Quantified edge server placement with quantum encoding in internet of vehicles | |
| CN114938376A (en) | Industrial Internet of things based on priority processing data and control method thereof | |
| CN114356531A (en) | Edge calculation task classification scheduling method based on K-means clustering and queuing theory | |
| CN117479049A (en) | Data transmission control method for intelligent fuel gas, internet of things system and medium | |
| CN117640634A (en) | Cloud edge end cooperation method and system for integrated management of intelligent monitoring system | |
| CN116795553A (en) | Method and device for scheduling computing power resources, storage medium and electronic device | |
| CN116089083A (en) | Multi-target data center resource scheduling method | |
| Prado et al. | On providing quality of service in grid computing through multi-objective swarm-based knowledge acquisition in fuzzy schedulers | |
| CN106295117A (en) | A kind of passive phased-array radar resource is dynamically queued up management-control method | |
| CN107562534B (en) | Load balancing method for weighted minimum data volume | |
| CN112954022A (en) | Multi-concurrency real-time communication method and device based on intelligent substation | |
| CN113806094B (en) | Cloud platform resource dynamic scheduling method based on deep learning | |
| CN116414726B (en) | Task dynamic allocation data parallel computing method based on memory and calculation integrated accelerator | |
| CN117596246B (en) | Method and system for scheduling workflow of computing power network based on heterogeneous resource measurement characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |