CN112668912B - Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network - Google Patents
Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network Download PDFInfo
- Publication number
- CN112668912B CN112668912B CN202011641675.6A CN202011641675A CN112668912B CN 112668912 B CN112668912 B CN 112668912B CN 202011641675 A CN202011641675 A CN 202011641675A CN 112668912 B CN112668912 B CN 112668912B
- Authority
- CN
- China
- Prior art keywords
- dynamic calculation
- terminal device
- condition
- artificial neural
- neural network
- 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.)
- Active
Links
- 238000004364 calculation method Methods 0.000 title claims abstract description 78
- 230000011218 segmentation Effects 0.000 title claims abstract description 36
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012549 training Methods 0.000 title claims abstract description 25
- 230000005540 biological transmission Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 6
- 238000007689 inspection Methods 0.000 description 7
- 238000012423 maintenance Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 239000011521 glass Substances 0.000 description 2
- 238000011897 real-time detection Methods 0.000 description 2
- 238000009877 rendering Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a training method of an artificial neural network, a dynamic calculation segmentation scheduling method, a storage medium and a system, wherein the training method respectively executes sample acquisition steps on each terminal device to obtain a plurality of groups of learning samples, and the sample acquisition steps executed on each terminal device comprise: acquiring the network condition, the cruising condition and the operation load of the terminal equipment; acquiring the dynamic calculation task quantity which can be processed by the terminal equipment; taking the network condition, the cruising condition and the operation load of the terminal equipment as input signals and taking the dynamic calculation task quantity which can be processed by the terminal equipment as output signals to form a group of learning samples for the artificial neural network to perform dynamic calculation, segmentation and scheduling training; and carrying out dynamic calculation segmentation scheduling training on the artificial neural network by adopting the plurality of groups of learning samples until the artificial neural network has the capability of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a training method of an artificial neural network, a dynamic calculation segmentation scheduling method, a storage medium and a system.
Background
At present, the field equipment inspection mainly adopts a manual inspection and manual recording mode, and the possibility of illegal inspection, missed inspection and false inspection exists. In the repair work of equipment maintenance, the problems of accident occurrence, untimely repair and the like caused by negligence and insufficient experience of operators can exist.
The AR intelligent operation and detection system is mainly used for field equipment inspection and repair in electric power operation and maintenance, and through AR intelligent terminal equipment, safety operation, efficient operation and intelligent operation of field operation and maintenance personnel are assisted, and individual operation capacity of the operation and maintenance personnel is improved. The AR intelligent operation and detection system has the main functions of workflow visualization, operation process recording, abnormal state alarming, operation position navigation, equipment data real-time viewing, visual operation guidance, remote collaborative operation, remote expert guidance and the like.
In order to adapt to different application scenes and application requirements of enterprises, the AR intelligent operation and detection system can support multiple terminal devices such as AR intelligent glasses, intelligent head-mounted terminals, intelligent mobile phones, tablets and computers. Because the AR intelligent operation and detection system needs to perform various real-time detection, identification and rendering tasks, a plurality of terminal devices are required to support relatively large dynamic calculation tasks, and the dynamic calculation tasks need to be reasonably segmented and scheduled according to each terminal device, so that the amount of the dynamic calculation tasks processed by each terminal device is reasonable, but the network condition, the cruising condition and the operation load of each terminal device are different, so that the amount of the dynamic calculation tasks which can be processed by each terminal device is different, and the segmentation and scheduling of the dynamic calculation tasks are difficult to reasonably segment and schedule.
Disclosure of Invention
The technical problem to be solved by the invention is how to reasonably segment and schedule the dynamic calculation task.
In order to solve the technical problems, the invention provides a training method for enabling an artificial neural network to have the capability of dynamically calculating segmentation scheduling, which comprises the following steps:
P, under the condition that a plurality of terminal devices which are matched with each other to execute dynamic calculation tasks exist and the amount of the dynamic calculation tasks which can be processed by each terminal device is known, each terminal device is executed with a sample acquisition step to obtain a plurality of groups of learning samples, wherein the sample acquisition step executed for each terminal device comprises the following A, B, C:
a, acquiring the network condition, the cruising condition and the operation load of the terminal equipment;
B, obtaining the dynamic calculation task quantity which can be processed by the terminal equipment;
C, taking the network condition, the cruising condition and the operation load of the terminal equipment as input signals, and taking the dynamic calculation task quantity which can be processed by the terminal equipment as output signals to form a group of learning samples for the artificial neural network to perform dynamic calculation, segmentation and scheduling training;
and Q, carrying out dynamic calculation segmentation scheduling training on the artificial neural network by adopting the plurality of groups of learning samples until the artificial neural network has the capability of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment, so that the artificial neural network can carry out segmentation scheduling on the dynamic calculation task according to the dynamic calculation task quantity which can be processed by the terminal equipment.
Preferably, in the step a, a network transmission speed of the terminal device is obtained, and a network condition of the terminal device is analyzed according to the network transmission speed.
Preferably, in the step a, a remaining power of the terminal device is obtained, and a endurance condition of the terminal device is analyzed according to the remaining power.
Preferably, in the step a, a performance parameter of a central processing unit of the terminal device is obtained, and an operation load of the terminal device is analyzed according to the performance parameter of the central processing unit.
The invention also provides a dynamic calculation segmentation scheduling method, which comprises the following steps:
a. acquiring network conditions, endurance conditions and operation loads of a plurality of terminal devices;
b. inputting the network condition, the cruising condition and the operation load of each terminal device into a trained artificial neural network, and analyzing the dynamic calculation task quantity which can be processed by each terminal device by the artificial neural network;
c. and carrying out segmentation scheduling on the dynamic calculation tasks according to the dynamic calculation task quantity which can be processed by each terminal device.
Preferably, in the step a, a network transmission speed of each terminal device is obtained, and a network condition of each terminal device is analyzed according to the network transmission speed.
Preferably, in the step a, the remaining power of each terminal device is obtained, and the endurance condition of each terminal device is analyzed according to the remaining power.
Preferably, in the step a, the performance parameters of the central processing unit of each terminal device are obtained, and the operation load of each terminal device is analyzed according to the performance parameters of each central processing unit.
The present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements steps in a training method as described above and/or steps in a dynamic computing cut scheduling method as described above.
The invention also provides a dynamic computing segmentation scheduling system, which comprises a server and a plurality of terminal devices which are mutually matched to execute dynamic computing tasks, wherein the server comprises a computer readable storage medium and a processor which are mutually connected, and the computer readable storage medium is described above.
The invention has the following beneficial effects: the trained artificial neural network has the capability of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment, so that the artificial neural network can segment and schedule the dynamic calculation task according to the dynamic calculation task quantity which can be processed by the terminal equipment, and the segmentation and scheduling of the dynamic calculation task are more reasonable.
Detailed Description
Exemplary embodiments of the present application will be described in detail below. While exemplary embodiments of the application are described below, it should be understood that the application can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art.
In the process of inspecting, repairing and repairing field devices in power operation and maintenance, the AR intelligent operation and inspection system needs to perform various real-time detection, identification and rendering tasks, so that a plurality of terminal devices (such as AR intelligent glasses, intelligent head-mounted terminals, intelligent mobile phones, tablets, computers and the like) are needed to support relatively large dynamic calculation tasks, and a dynamic calculation segmentation scheduling system is needed to reasonably segment and schedule the dynamic calculation tasks, so that the amount of the dynamic calculation tasks processed by each terminal device is reasonable. The dynamic computing segmentation scheduling system comprises a server and a plurality of terminal devices which are mutually matched to execute dynamic computing tasks, wherein the server comprises a computer readable storage medium and a processor which are mutually connected, and a computer program is stored in the computer readable storage medium. Under the condition that a plurality of terminal devices which are matched with each other to execute dynamic calculation tasks exist and the amount of the dynamic calculation tasks which can be processed by each terminal device is known, the dynamic calculation segmentation scheduling system enables a processor to execute a computer program in a computer readable storage medium to realize a training method for enabling an artificial neural network to have the dynamic calculation segmentation scheduling capability, the training method respectively executes a sample acquisition step on each terminal device to obtain a plurality of groups of learning samples, wherein the sample acquisition step executed on each terminal device comprises the following A, B, C:
A. acquiring the network transmission speed of the terminal equipment, and analyzing the network condition of the terminal equipment according to the network transmission speed; acquiring the residual electric quantity of the terminal equipment, and analyzing the endurance condition of the terminal equipment according to the residual electric quantity; and acquiring the performance parameters of the central processing unit of the terminal equipment, and analyzing the operation load of the terminal equipment according to the performance parameters of the central processing unit.
B. Acquiring the dynamic calculation task quantity which can be processed by the terminal equipment;
C. The network condition, the cruising condition and the operation load of the terminal equipment are used as input signals, and the dynamic calculation task quantity which can be processed by the terminal equipment is used as an output signal to form a group of learning samples for the dynamic calculation, segmentation and scheduling training of the artificial neural network. Thus, each group of learning samples establishes a corresponding relation between the network condition, the endurance condition and the operation load and the dynamic calculation task quantity which can be processed by the terminal equipment.
The plurality of groups of learning samples are adopted to carry out dynamic calculation segmentation scheduling training on the artificial neural network until the artificial neural network has the capacity of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment, so that the artificial neural network can carry out segmentation scheduling on the dynamic calculation task according to the dynamic calculation task quantity which can be processed by the terminal equipment, and the dynamic calculation segmentation scheduling system can enable a processor to execute a computer program in a computer readable storage medium to realize a dynamic calculation segmentation scheduling method, and the following steps a, b and c are detailed:
a. acquiring network transmission speeds of a plurality of terminal devices, and analyzing the network condition of each terminal device according to each network transmission speed; obtaining the residual electric quantity of a plurality of terminal devices, and analyzing the endurance condition of each terminal device according to each residual electric quantity; acquiring the performance parameters of central processing units of a plurality of terminal devices, and analyzing the operation load of each terminal device according to the performance parameters of each central processing unit;
b. inputting the network condition, the cruising condition and the operation load of each terminal device into a trained artificial neural network, and analyzing the dynamic calculation task quantity which can be processed by each terminal device by the artificial neural network;
c. The segmentation scheduling is carried out on the dynamic calculation tasks according to the dynamic calculation task quantity which can be processed by each terminal device, so that the dynamic calculation task quantity which is finally scheduled by each terminal device is adapted to the dynamic calculation task quantity which can be processed by the terminal device, and the segmentation scheduling of the dynamic calculation tasks is more reasonable.
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the scope of the present application, and although the present application has been 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 to the technical solution of the present application without departing from the spirit and scope of the technical solution of the present application.
Claims (10)
1. The training method for enabling the artificial neural network to have the capability of dynamically calculating segmentation scheduling is characterized by comprising the following steps:
P, under the condition that a plurality of terminal devices which are matched with each other to execute dynamic calculation tasks exist and the amount of the dynamic calculation tasks which can be processed by each terminal device is known, each terminal device is executed with a sample acquisition step to obtain a plurality of groups of learning samples, wherein the sample acquisition step executed for each terminal device comprises the following A, B, C:
A. Acquiring the network condition, the cruising condition and the operation load of the terminal equipment;
B. Acquiring the dynamic calculation task quantity which can be processed by the terminal equipment;
C. taking the network condition, the cruising condition and the operation load of the terminal equipment as input signals and taking the dynamic calculation task quantity which can be processed by the terminal equipment as output signals to form a group of learning samples for the artificial neural network to perform dynamic calculation, segmentation and scheduling training;
and Q, carrying out dynamic calculation segmentation scheduling training on the artificial neural network by adopting the plurality of groups of learning samples until the artificial neural network has the capability of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment, so that the artificial neural network can carry out segmentation scheduling on the dynamic calculation task according to the dynamic calculation task quantity which can be processed by the terminal equipment.
2. The training method according to claim 1, wherein in the step a, a network transmission speed of the terminal device is obtained, and the network condition of the terminal device is analyzed according to the network transmission speed.
3. The training method according to claim 1, wherein in the step a, a remaining power of the terminal device is obtained, and a cruising situation of the terminal device is analyzed according to the remaining power.
4. The training method of claim 1, wherein in the step a, a performance parameter of a central processing unit of the terminal device is obtained, and an operation load of the terminal device is analyzed according to the performance parameter of the central processing unit.
5. The dynamic calculation segmentation scheduling method is characterized by comprising the following steps of:
a. acquiring network conditions, endurance conditions and operation loads of a plurality of terminal devices;
b. inputting the network condition, the cruising condition and the operation load of each terminal device into a trained artificial neural network, and analyzing the dynamic calculation task quantity which can be processed by each terminal device by the artificial neural network;
c. The method comprises the steps that segmentation scheduling is carried out on dynamic calculation tasks according to the amount of the dynamic calculation tasks which can be processed by each terminal device;
the training method adopted by the trained artificial neural network comprises the following steps:
P, under the condition that a plurality of terminal devices which are matched with each other to execute dynamic calculation tasks exist and the amount of the dynamic calculation tasks which can be processed by each terminal device is known, each terminal device is executed with a sample acquisition step to obtain a plurality of groups of learning samples, wherein the sample acquisition step executed for each terminal device comprises the following A, B, C:
A. Acquiring the network condition, the cruising condition and the operation load of the terminal equipment;
B. Acquiring the dynamic calculation task quantity which can be processed by the terminal equipment;
C. taking the network condition, the cruising condition and the operation load of the terminal equipment as input signals and taking the dynamic calculation task quantity which can be processed by the terminal equipment as output signals to form a group of learning samples for the artificial neural network to perform dynamic calculation, segmentation and scheduling training;
and Q, carrying out dynamic calculation segmentation scheduling training on the artificial neural network by adopting the plurality of groups of learning samples until the artificial neural network has the capability of analyzing the dynamic calculation task quantity which can be processed by the terminal equipment according to the network condition, the cruising condition and the operation load of the terminal equipment, so that the artificial neural network can carry out segmentation scheduling on the dynamic calculation task according to the dynamic calculation task quantity which can be processed by the terminal equipment.
6. The method according to claim 5, wherein in the step a, a network transmission speed of each terminal device is obtained, and a network condition of each terminal device is analyzed according to the network transmission speed.
7. The method for dynamically calculating and slicing the scheduling according to claim 5, wherein in the step a, the remaining power of each terminal device is obtained, and the endurance of each terminal device is analyzed according to the remaining power.
8. The method according to claim 5, wherein in the step a, the cpu performance parameters of each terminal device are obtained, and the operation load of each terminal device is analyzed according to the cpu performance parameters.
9. Computer readable storage medium, having stored thereon a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the training method according to any of claims 1 to 4 and/or the steps of the dynamic computing segmentation scheduling method according to any of claims 5 to 8.
10. A dynamic computing slicing scheduling system, comprising a server and a plurality of terminal devices that cooperate to perform dynamic computing tasks, the server comprising a computer readable storage medium and a processor coupled to each other, the computer readable storage medium being as recited in claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011641675.6A CN112668912B (en) | 2020-12-31 | 2020-12-31 | Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011641675.6A CN112668912B (en) | 2020-12-31 | 2020-12-31 | Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112668912A CN112668912A (en) | 2021-04-16 |
| CN112668912B true CN112668912B (en) | 2024-06-14 |
Family
ID=75412308
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011641675.6A Active CN112668912B (en) | 2020-12-31 | 2020-12-31 | Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112668912B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115551105B (en) * | 2022-09-15 | 2023-08-25 | 公诚管理咨询有限公司 | Task scheduling method, device and storage medium based on 5G network edge calculation |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109976911A (en) * | 2019-03-25 | 2019-07-05 | 哈尔滨工程大学 | A kind of adaptive resource dispatching method |
| CN111079921A (en) * | 2019-11-29 | 2020-04-28 | 杭州电子科技大学舟山同博海洋电子信息研究院有限公司 | Efficient neural network training and scheduling method based on heterogeneous distributed system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9672064B2 (en) * | 2015-07-13 | 2017-06-06 | Palo Alto Research Center Incorporated | Dynamically adaptive, resource aware system and method for scheduling |
| KR20190076693A (en) * | 2017-12-22 | 2019-07-02 | 부산대학교 산학협력단 | Automatic distributing and predicting usage for workload in heterogeneous cloud environment |
| CN111126594B (en) * | 2019-11-25 | 2023-08-04 | 北京邮电大学 | Neural network model dynamic segmentation method and device based on edge calculation |
| CN111158918B (en) * | 2019-12-31 | 2022-11-11 | 深圳大学 | Supporting point parallel enumeration load balancing method, device, equipment and medium |
| CN111915142B (en) * | 2020-07-07 | 2024-04-12 | 广东工业大学 | Unmanned aerial vehicle auxiliary resource allocation method based on deep reinforcement learning |
-
2020
- 2020-12-31 CN CN202011641675.6A patent/CN112668912B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109976911A (en) * | 2019-03-25 | 2019-07-05 | 哈尔滨工程大学 | A kind of adaptive resource dispatching method |
| CN111079921A (en) * | 2019-11-29 | 2020-04-28 | 杭州电子科技大学舟山同博海洋电子信息研究院有限公司 | Efficient neural network training and scheduling method based on heterogeneous distributed system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112668912A (en) | 2021-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110020967B (en) | Information processing method and device for intelligent alarm of power grid dispatching end transformer substation | |
| CN110221145B (en) | Power equipment fault diagnosis method and device and terminal equipment | |
| Nascimento et al. | An integrated inspection system for belt conveyor rollers | |
| CN112668912B (en) | Training method, dynamic calculation segmentation scheduling method, storage medium and system for artificial neural network | |
| CN111123873B (en) | Production data acquisition method and system based on stream processing technology | |
| CN115562144A (en) | Major hazard source safety monitoring and early warning management system based on 5G technology | |
| CN115660288A (en) | Analysis management system based on internet big data | |
| CN114995302A (en) | Intelligent control method and system | |
| CN111339933A (en) | Transformer substation safety monitoring method and device based on deep learning | |
| Huang et al. | A method of transmission conductor-loosened detect based on image sensors | |
| CN112101125B (en) | Method and device for detecting stacking degree of express goods | |
| CN117726164A (en) | Risk fault situation awareness system based on multisource data fusion | |
| CN112306722B (en) | Method, apparatus, device and computer readable medium for identifying faults | |
| CN116185757B (en) | Intelligent monitoring system for energy consumption of machine room | |
| CN112785109A (en) | Power grid equipment fault analysis method and system based on regulation cloud | |
| CN115567563A (en) | Comprehensive transportation hub monitoring and early warning system based on end edge cloud and control method thereof | |
| CN115187086A (en) | Intelligent management and control system for emergency management | |
| CN206237041U (en) | A kind of clustering system of power transmission and transforming equipment electric power data | |
| CN114265891A (en) | Intelligent workshop system and method based on multi-source data fusion and storage medium | |
| CN115239071A (en) | Industrial production environment risk assessment system based on big data | |
| CN112116108A (en) | Expert system and method for failure analysis of power equipment | |
| CN108107874A (en) | A kind of multi-stage scheduling automated system | |
| CN113837110A (en) | Centralized monitoring and alarming method and system for running state of power grid station equipment | |
| CN114139747A (en) | AIOps intelligent operation and maintenance system based on artificial intelligence technology | |
| CN112711508A (en) | Intelligent operation and maintenance service system facing large-scale client system |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240517 Address after: 5-C, Jiahai Building, No. 17 Jinpan Road, Longhua District, Haikou City, Hainan Province, 570216 Applicant after: China Software Technology (Hainan) Information Technology Co.,Ltd. Country or region after: China Address before: 510620 room 2807a12, 140-148 TIYU East Road, Tianhe District, Guangzhou City, Guangdong Province Applicant before: Guangdong Zhongke nengrei Information Technology Co.,Ltd. Country or region before: China |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |