CN114356502A - Unstructured data marking, training and publishing system and method based on edge computing technology - Google Patents
Unstructured data marking, training and publishing system and method based on edge computing technology Download PDFInfo
- Publication number
- CN114356502A CN114356502A CN202111667932.8A CN202111667932A CN114356502A CN 114356502 A CN114356502 A CN 114356502A CN 202111667932 A CN202111667932 A CN 202111667932A CN 114356502 A CN114356502 A CN 114356502A
- Authority
- CN
- China
- Prior art keywords
- model
- training
- management controller
- edge computing
- data
- 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.)
- Granted
Links
- 238000012549 training Methods 0.000 title claims abstract description 91
- 238000005516 engineering process Methods 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000011156 evaluation Methods 0.000 claims abstract description 40
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 238000013473 artificial intelligence Methods 0.000 claims description 14
- 238000002372 labelling Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 4
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical compound OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 description 8
- 230000002085 persistent effect Effects 0.000 description 5
- 230000003068 static effect Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000003466 anti-cipated effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an unstructured data marking, training and publishing system and method based on an edge computing technology. The edge computing physical terminal is used for acquiring unstructured data of a power operation field, loading a model to perform edge analysis and computation on the acquired data, returning an analysis result to a cloud control center, receiving cloud issuing or pushing information and automatically completing deployment of an issuing model; the cloud control center comprises a container cluster management controller, a model training evaluation and data marking management controller and a distributed message management controller. The invention realizes the sharing and releasing of the video stream data collected by the terminal by using the distributed message management controller, generates the virtual terminal by using the container pool technology, greatly improves the training, evaluation and updating management efficiency of the model, shortens the upgrading and optimizing period of the edge computing physical terminal model, and continuously improves the analysis capability of the whole system.
Description
Technical Field
The invention relates to the field of edge computing and artificial intelligence, in particular to an unstructured data marking, training and publishing system and method based on an edge computing technology.
Background
The internet of things (IOT) technology is widely used at present, and supports applications including internet of vehicles and smart cities, and is gradually developing in China, and it is predicted that in the future, 5 hundred million edge devices in the world are expected to achieve internet of things access before 2030 years, and such applications based on data flow will continue to increase, and a huge expansion is expected in recent years.
In the application scenario, the continuous information generated by the internet of things needs to be processed, and the application scenario is from the source to various application purposes. For example, large internet social media requires processing millions of pictures each day, checking for non-compliant content. Machine Learning (ML) and Artificial Intelligence (AI) are playing a key decision-making role in improving the work of business operations and personal life by converting the raw information into useful predictions and recommendations. For this reason, it is anticipated that the above scenario ultimately requires an unstructured data tagging, training and publishing that supports continuous data flow information for ML/AI algorithms and systems.
Most artificial energy intelligent standards and training frameworks in the industry at present are based on the design and development of ML/AI algorithm training platforms of persistent data sets and static data, and are not designed to cooperate with the work of data flow. To date, the popular Python framework is similar to Pytorch, thano, tensrflow, providing no or only partial support for data streaming systems. This situation exists not only in training Machine Learning (ML) models only, but also in subsequent application steps of ML/AI pipelining, testing and evaluation. The problem that a lasting data set and static data cannot be suitable for training a continuously-changing production operation scene model due to the timeliness problem, the existing artificial energy intelligent standard and a training framework cannot well support massive video stream processing based on the Internet of things technology, the problems that the training model period is long, different application scenes cannot be quickly adapted and the like exist, and the problem that a trained general model is completed and the model can be suitable for application in a specific scene is solved.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides an unstructured data marking, training and publishing system and method based on an edge computing technology, which can support the processing of massive video streams based on the Internet of things technology and solve the problem that the existing persistent data set and static data cannot be applied to a specific production scene; the virtual terminal generated by using the container pool technology can obviously improve the model training, evaluation and release automation level and the model multiplexing compatibility so as to solve the problems that a general model can be applied to a specific scene.
An unstructured data marking, training and publishing system and method based on an edge computing technology comprises an edge computing physical terminal, a cloud control center and a computing node cluster;
the edge computing physical terminal is used for acquiring unstructured data of an electric power operation field, issuing the unstructured data to the cloud control center as a shared data set, loading a model which is loaded on the cloud control center to finish training and evaluation, carrying out edge analysis and computation on the acquired unstructured data, returning an analysis result to the cloud control center, receiving a message issued or pushed by the cloud control center, and automatically finishing deployment of an issued model;
the cloud control center comprises a container cluster management controller, a model training evaluation and data marking management controller and a distributed message management controller;
the container cluster management controller is used for creating a Docker container in the computing node cluster aiming at each edge computing physical terminal, generating a corresponding virtual terminal corresponding to the mirror image file of the edge computing physical terminal, and the virtual terminal is used for training and evaluating the model;
the model training evaluation and data marking management controller is used for managing a model, marking a data set and evaluating the model;
the distributed message management controller is used for managing video stream sharing, control message publishing and subscribing among the physical edge terminal, the cloud control center and the virtual terminal;
the computing node cluster comprises a mirror image library, a container pool and a device area,
the mirror image library is used for storing mirror image files corresponding to each physical edge computing physical terminal;
the container pool is used for operating the virtual terminal and calculating model training, marking, evaluating and issuing tasks created by the data marking management controller;
and the equipment area is used for storing basic information and running state information of the edge computing physical terminal sent by the cloud control center.
Further, the container cluster management controller is deployed by Kubernets, generates a virtual terminal in the container pool according to a mirror image file in a mirror image library corresponding to the edge computing physical terminal, and mounts GPU computing resources in the computing node cluster to the virtual terminal by a CUDA Driver.
Further, the container cluster management controller is further configured to destroy the edge computing physical terminal and recover the computing resources according to the disabled state of the edge computing physical terminal.
Further, the model training evaluation and data labeling management controller is used for managing models, labeled data sets and model evaluation, and specifically includes: designing and defining an artificial intelligence training model, creating model training, marking, evaluating and issuing tasks, configuring a shared continuous video stream data set issued by a distributed message management controller for the tasks to serve as a training set, a test set or a marking set of the model, issuing the training and evaluating tasks of the model to a virtual terminal for training and evaluating, and issuing the evaluated model to the distributed message management controller to be pushed to an edge computing physical terminal.
Further, the basic information comprises an identity ID, a security certificate and a digital certificate of the equipment; the running state information comprises log-in and return message log records of the equipment.
An unstructured data marking, training and publishing system and method based on an edge computing technology are carried out by adopting the system, and the method comprises the following steps:
the method comprises the steps that an edge computing physical terminal collects unstructured data of a power operation field, shares and issues the unstructured data to a cloud control center, and generates a message theme of a continuous video stream data set;
designing and defining an artificial intelligence training model by a model training evaluation and data marking management controller, configuring the continuous video stream data set, and establishing tasks of model training, marking, evaluating and issuing;
the container cluster management controller calls the computing node cluster according to the created task, takes out the mirror image file from the mirror image library, distributes computing resources to generate a deployment virtual terminal in a container pool, and runs model training, marking, evaluating and releasing tasks;
the distributed message management controller pushes the trained model to an edge computing physical terminal;
the edge computing physical terminal starts a model issued by the distributed message management controller, is used for carrying out edge analysis computation on the acquired unstructured data, analyzing and marking video stream data, and returns the result to the distributed message management controller.
Further, the method also comprises the following steps: and the model training evaluation and data marking management controller evaluates the returned analysis result to generate a model analysis evaluation report, generates a strategy for optimizing and improving the model according to the evaluation report, and starts a new round of model training and issuing tasks.
The invention realizes the sharing and releasing of the video stream data collected by the terminal by using the distributed message management controller, generates the virtual terminal by using the container pool technology, greatly improves the training, evaluation and updating management efficiency of the model, shortens the training and optimizing period of the edge computing physical terminal model, and continuously improves the integral analysis capability of the system. In the system disclosed by the invention, in the real-time deployment and application of a 500kV transformer substation operation site in Hubei, 119 total-station video acquisition terminals are used as dynamic continuous data source access systems to cover areas such as 3 total-station 500kV main transformers and disconnecting links, and aiming at the training and application of 12 models such as target detection, personnel posture recognition and the like in the overhaul scene of an operating transformer substation, the period of model training and tuning is shortened from 7 days to 1 day, and the model reasoning precision (recognition accuracy index) is improved by about 5%.
Drawings
FIG. 1 is a schematic structural diagram of an unstructured data labeling, training and publishing system based on edge computing technology according to an embodiment of the present invention;
FIG. 2 is a flow chart of one embodiment of the unstructured data labeling, training and publishing method based on edge computing technology;
FIG. 3 is a schematic diagram of an edge computing physical terminal marking function interface according to the present invention.
In fig. 1: 10-edge computing physical terminal, 20-cloud control center, 30-computing node cluster, 21-container cluster management controller, 22-model training evaluation and data marking management controller, 23-distributed message management controller, 31-mirror library, 32-container pool and 33-device area.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
Fig. 1 is a schematic structural diagram of an unstructured data labeling, training and publishing system based on an edge computing technology according to an embodiment of the present invention, where the system includes an edge computing physical terminal 10, a cloud control center 20, and a compute node cluster 30.
The edge computing physical terminal 10 is configured to collect unstructured data such as video streams of power operation sites (e.g., substations, cables, towers, and the like), and issue the unstructured data to the cloud control center 30 as a shared data set; the system is further configured to load a model of the cloud control center 20 completing training and evaluation to perform edge analysis calculation on the acquired unstructured data, analyze and mark video stream data, and return an analysis result to the cloud control center 30, where the video analysis result includes a video key frame and a JSON format mark file; the edge computing physical terminal 10 may also be configured to receive a message issued or pushed by the cloud control center 30, and automatically complete the deployment of the issuing model. The edge computing physical terminal 10 may be a mobile edge computing physical terminal.
The cloud control center 20 includes a container cluster management controller 21, a model training evaluation and data tagging management controller 22, and a distributed message management controller 23.
The container cluster management controller 21 is deployed by kubernets, a Docker container is created in the computing node cluster 30 for each edge computing physical terminal 10, the Docker container and the edge computing physical terminal 10 share the same mirror image file, a virtual terminal is generated in a container pool 32 according to the mirror image file in a mirror image library 31 corresponding to the edge computing physical terminal 10, GPU computing resources in the computing node cluster 30 are mounted to the virtual terminal by a CUDA Driver, and training and evaluation of a model are achieved in the virtual terminal. The container cluster management controller 21 can perform destruction and recovery of computing resources according to the deactivated state of the edge computing physical terminal 10.
The model training evaluation and data tagging management controller 22 is used for managing models, tagging data sets and model evaluation, is compatible with mainstream artificial intelligence algorithm support libraries such as Pytorch and tenserflow, specifically, designs and defines an artificial intelligence training model, creates model training, tagging, evaluation and publishing tasks, configures a shared continuous video stream data set published by the distributed message management controller 23 for the tasks to be used as a training set, a test set or a tagging set of the model, publishes a training evaluation task of the model to a virtual terminal for training and evaluation, and publishes the evaluated model to the distributed message management controller 23 to be pushed to the edge computing physical terminal 10, so as to update a local model.
The distributed message management controller 23 is deployed by Kafka, and is configured to share, publish and subscribe the control message to and manage the video stream between the edge computing physical terminal 10, the cloud control center 20, and the virtual terminal, for example, receive the video stream data acquired by the edge computing physical terminal 10, and publish the video stream data as a shared video stream data set (OpenCV processes the video stream into a single frame of picture, and publishes the video stream data through a local Kafka Producer). The video stream data set will be published as a Topic (Topic) by an agent (Broker) of the cloud control center 20 for shared persistent video stream data set for model training and evaluation.
The computing node cluster comprises a mirror image library 31, a container pool 32 and a device area 33,
the mirror image library 31 is used for storing mirror image files corresponding to each physical edge computing physical terminal 10, allocating resources by the container cluster management controller 21, and then generating virtual terminals in the container pool 32.
The container pool 32 is used for operating virtual terminals, and the virtual terminals execute model training evaluation and model training, marking, evaluation and publishing tasks created by the data marking management controller 22.
The device area 33 is configured to store basic information and running state information of the edge computing physical terminal 10 sent by the cloud control center 20, and an Apache IoTDB database is used. Wherein the basic information comprises the identity ID, security authentication, digital certificate and the like of the equipment; the running state information includes logging on of the device, returning of message log records, and the like.
As shown in fig. 2, an embodiment of the present invention further provides an unstructured data labeling, training, and publishing method based on an edge computing technology, which is performed by using the foregoing system, and the method includes the following steps:
the edge computing physical terminal 10 collects unstructured data such as video streams of the power operation site, shares and issues the unstructured data to the cloud control center 20, and generates a message theme of a continuous video stream data set.
The model training evaluation and data tagging management controller 22 designs and defines an artificial intelligence training model, configures a continuous video stream data set generated in the previous step, and creates model training, tagging, evaluating and publishing tasks.
The container cluster management controller 21 calls the computing node cluster 30 according to the task created in the previous step, takes out the mirror image file from the mirror image library 31, allocates computing resources to generate a deployment virtual terminal in the container pool 32, and runs the tasks of model training, marking, evaluating and issuing.
After the task operation in the previous step is completed, the distributed message management controller 23 pushes the trained model to the edge computing physical terminal 10.
The edge computing physical terminal 10 enables a model issued by the distributed message management controller 23 for performing edge analysis computation on the collected unstructured data, analyzing and marking video stream data (as shown in fig. 3), and transmitting the result back to the distributed message management controller 23.
The model training evaluation and data marking management controller 22 evaluates the analysis result returned in the previous step, and generates a model analysis evaluation report. And generating a strategy for optimizing and improving the model according to the evaluation report, and starting a new round of model training and issuing tasks.
The invention utilizes distributed message distribution, container clustering, virtualization terminal technology and continuous video stream data set sharing technology, improves the efficiency of model training, marking, evaluating and distributing of the edge computing device, shortens the period of upgrading and optimizing the edge computing physical terminal model and continuously improves the analysis capability of the whole system. The system is applied to the edge calculation analysis of the digital safety control terminal of the electric power operation field in many places and cities in Hubei province.
Under the traditional method, the current artificial energy intelligence standard and training framework of the ML/AI algorithm training platform based on the design and development of persistent data sets and static data are not suitable for the AI edge computing technology under the application scene of the Internet of things. The period from the collection, labeling, model training, evaluation of the persistent data set to the model release is up to several weeks, but with the present invention, the period of completing the training release of a model only needs less than 1 day. The new method actually changes the original static data set configuration mode into a dynamic continuous video stream data set mode, realizes distributed concurrent processing of model training and release by combining a virtual terminal technology, and fully utilizes data and computing resources, thereby greatly improving the efficiency.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (7)
1. An unstructured data labeling, training and publishing system based on edge computing technology, characterized by: the method comprises the following steps of calculating a physical terminal, a cloud control center and a calculation node cluster;
the edge computing physical terminal is used for acquiring unstructured data of an electric power operation field, issuing the unstructured data to the cloud control center as a shared data set, loading a model which is loaded on the cloud control center to finish training and evaluation, carrying out edge analysis and computation on the acquired unstructured data, returning an analysis result to the cloud control center, receiving a message issued or pushed by the cloud control center, and automatically finishing deployment of an issued model;
the cloud control center comprises a container cluster management controller, a model training evaluation and data marking management controller and a distributed message management controller;
the container cluster management controller is used for creating a Docker container in the computing node cluster aiming at each edge computing physical terminal, generating a corresponding virtual terminal corresponding to the mirror image file of the edge computing physical terminal, and the virtual terminal is used for training and evaluating the model;
the model training evaluation and data marking management controller is used for managing a model, marking a data set and evaluating the model;
the distributed message management controller is used for managing video stream sharing, control message publishing and subscribing among the physical edge terminal, the cloud control center and the virtual terminal;
the computing node cluster comprises a mirror image library, a container pool and a device area,
the mirror image library is used for storing mirror image files corresponding to each physical edge computing physical terminal;
the container pool is used for operating the virtual terminal and calculating model training, marking, evaluating and issuing tasks created by the data marking management controller;
and the equipment area is used for storing basic information and running state information of the edge computing physical terminal sent by the cloud control center.
2. An unstructured data tagging, training and publishing system based on edge computing technology as claimed in claim 1 wherein: the container cluster management controller is deployed by Kubernets, generates a virtual terminal in a container pool according to a mirror image file in a mirror image library corresponding to an edge computing physical terminal, and mounts GPU computing resources in a computing node cluster to the virtual terminal by a CUDA Driver.
3. An unstructured data tagging, training and publishing system based on edge computing technology as claimed in claim 1 wherein: the container cluster management controller is also used for destroying and recovering computing resources according to the stopping state of the edge computing physical terminal.
4. An unstructured data tagging, training and publishing system based on edge computing technology as claimed in claim 1 wherein: the model training evaluation and data marking management controller is used for managing models, marking data sets and model evaluation, and specifically comprises the following steps: designing and defining an artificial intelligence training model, creating model training, marking, evaluating and issuing tasks, configuring a shared continuous video stream data set issued by a distributed message management controller for the tasks to serve as a training set, a test set or a marking set of the model, issuing the training and evaluating tasks of the model to a virtual terminal for training and evaluating, and issuing the evaluated model to the distributed message management controller to be pushed to an edge computing physical terminal.
5. An unstructured data tagging, training and publishing system based on edge computing technology as claimed in claim 1 wherein: the basic information comprises an identity ID, security authentication and a digital certificate of the equipment; the running state information comprises log-in and return message log records of the equipment.
6. An unstructured data labeling, training and publishing method based on edge computing technology, performed by the system of any one of claims 1-5, characterized in that the method comprises the steps of:
the method comprises the steps that an edge computing physical terminal collects unstructured data of a power operation field, shares and issues the unstructured data to a cloud control center, and generates a message theme of a continuous video stream data set;
designing and defining an artificial intelligence training model by a model training evaluation and data marking management controller, configuring the continuous video stream data set, and establishing tasks of model training, marking, evaluating and issuing;
the container cluster management controller calls the computing node cluster according to the created task, takes out the mirror image file from the mirror image library, distributes computing resources to generate a deployment virtual terminal in a container pool, and runs model training, marking, evaluating and releasing tasks;
the distributed message management controller pushes the trained model to an edge computing physical terminal;
the edge computing physical terminal starts a model issued by the distributed message management controller, is used for carrying out edge analysis computation on the acquired unstructured data, analyzing and marking video stream data, and returns the result to the distributed message management controller.
7. The unstructured-data labeling, training and publishing method based on edge computing technology as recited in claim 6, wherein: further comprising: and the model training evaluation and data marking management controller evaluates the returned analysis result to generate a model analysis evaluation report, generates a strategy for optimizing and improving the model according to the evaluation report, and starts a new round of model training and issuing tasks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111667932.8A CN114356502B (en) | 2021-12-31 | 2021-12-31 | Unstructured data marking, training and publishing system and method based on edge computing technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111667932.8A CN114356502B (en) | 2021-12-31 | 2021-12-31 | Unstructured data marking, training and publishing system and method based on edge computing technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114356502A true CN114356502A (en) | 2022-04-15 |
| CN114356502B CN114356502B (en) | 2024-02-13 |
Family
ID=81104834
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111667932.8A Active CN114356502B (en) | 2021-12-31 | 2021-12-31 | Unstructured data marking, training and publishing system and method based on edge computing technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114356502B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117969789A (en) * | 2024-03-29 | 2024-05-03 | 山东省地矿工程勘察院(山东省地质矿产勘查开发局八〇一水文地质工程地质大队) | Water regime prediction method and monitoring system for underground water resource |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018234741A1 (en) * | 2017-06-23 | 2018-12-27 | Qio Technologies Ltd | Systems and methods for distributed systemic anticipatory industrial asset intelligence |
| CN112102111A (en) * | 2020-09-27 | 2020-12-18 | 华电福新广州能源有限公司 | Intelligent processing system for power plant data |
| CN112214541A (en) * | 2020-09-04 | 2021-01-12 | 江苏方天电力技术有限公司 | Deep decoupling and data cooperation method for power monitoring data |
| CN113179190A (en) * | 2021-06-29 | 2021-07-27 | 深圳智造谷工业互联网创新中心有限公司 | Edge controller, edge computing system and configuration method thereof |
| US20210392049A1 (en) * | 2020-06-15 | 2021-12-16 | Cisco Technology, Inc. | Machine-learning infused network topology generation and deployment |
-
2021
- 2021-12-31 CN CN202111667932.8A patent/CN114356502B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018234741A1 (en) * | 2017-06-23 | 2018-12-27 | Qio Technologies Ltd | Systems and methods for distributed systemic anticipatory industrial asset intelligence |
| US20210392049A1 (en) * | 2020-06-15 | 2021-12-16 | Cisco Technology, Inc. | Machine-learning infused network topology generation and deployment |
| CN112214541A (en) * | 2020-09-04 | 2021-01-12 | 江苏方天电力技术有限公司 | Deep decoupling and data cooperation method for power monitoring data |
| CN112102111A (en) * | 2020-09-27 | 2020-12-18 | 华电福新广州能源有限公司 | Intelligent processing system for power plant data |
| CN113179190A (en) * | 2021-06-29 | 2021-07-27 | 深圳智造谷工业互联网创新中心有限公司 | Edge controller, edge computing system and configuration method thereof |
Non-Patent Citations (1)
| Title |
|---|
| 徐楚轲;吴昌宏;舒杰;杨湛晔;张虎润;: "一种面向分布式发电微电网的边缘计算架构与应用", 新能源进展, no. 05, 15 October 2020 (2020-10-15) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117969789A (en) * | 2024-03-29 | 2024-05-03 | 山东省地矿工程勘察院(山东省地质矿产勘查开发局八〇一水文地质工程地质大队) | Water regime prediction method and monitoring system for underground water resource |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114356502B (en) | 2024-02-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114362367B (en) | Cloud-edge-cooperation-oriented power transmission line monitoring system and method, and cloud-edge-cooperation-oriented power transmission line identification system and method | |
| US11392733B2 (en) | Multi-dimensional event model generation | |
| Hao et al. | The design of intelligent transportation video processing system in big data environment | |
| CN105512297A (en) | Distributed stream-oriented computation based spatial data processing method and system | |
| Liu et al. | Research and application of artificial intelligence service platform for the power field | |
| Jiang et al. | Application Research of Key Frames Extraction Technology Combined with Optimized Faster R‐CNN Algorithm in Traffic Video Analysis | |
| CN112291367B (en) | Data processing method and device | |
| CN112332981B (en) | Data processing method and device | |
| CN111131771A (en) | Video monitoring system | |
| CN112233428B (en) | Traffic flow prediction method, device, storage medium and equipment | |
| CN112328585A (en) | Data processing method and device | |
| CN113516331A (en) | Building data processing method and device | |
| Embarak | Smart Cities New Paradigm Applications and Challenges | |
| Jiang et al. | Research and design of infrastructure monitoring platform of intelligent high speed railway | |
| KR20230132350A (en) | Joint perception model training method, joint perception method, device, and storage medium | |
| CN114356502A (en) | Unstructured data marking, training and publishing system and method based on edge computing technology | |
| CN114418360A (en) | Smart city operation sign big data analysis method and device | |
| CN112486676B (en) | Data sharing and distributing device based on edge calculation | |
| Gunawardena et al. | Real-time Uber data analysis of popular Uber locations in Kubernetes environment | |
| CN115567563B (en) | Comprehensive transportation hub monitoring and early warning system based on end edge cloud and control method thereof | |
| CN112037103A (en) | Government affair management system | |
| CN112306985A (en) | Digital retina multi-modal feature combined accurate retrieval method | |
| CN114493206B (en) | Global airport management system and method based on BIM | |
| CN112383435B (en) | Fault processing method and device | |
| Ding et al. | Performance modeling and evaluation of real-time traffic status query for intelligent traffic systems |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |