WO2022160414A1 - 一种电厂监控系统 - Google Patents

一种电厂监控系统 Download PDF

Info

Publication number
WO2022160414A1
WO2022160414A1 PCT/CN2021/079288 CN2021079288W WO2022160414A1 WO 2022160414 A1 WO2022160414 A1 WO 2022160414A1 CN 2021079288 W CN2021079288 W CN 2021079288W WO 2022160414 A1 WO2022160414 A1 WO 2022160414A1
Authority
WO
WIPO (PCT)
Prior art keywords
server
power plant
monitoring terminal
application
image
Prior art date
Application number
PCT/CN2021/079288
Other languages
English (en)
French (fr)
Inventor
汪志强
陈满
张豪
卢勇
刘涛
李建辉
吕志鹏
林恺
韩玉麟
巩宇
陆传德
黄发满
景增明
Original Assignee
南方电网调峰调频发电有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 南方电网调峰调频发电有限公司 filed Critical 南方电网调峰调频发电有限公司
Publication of WO2022160414A1 publication Critical patent/WO2022160414A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Definitions

  • the utility model relates to the field of safety monitoring of power plant production equipment, more particularly, to a power plant monitoring system.
  • the traditional video surveillance system has the problem that the accuracy of fault judgment is not high, and it is prone to misjudgment and missed judgment.
  • the embodiment of the present utility model provides a power plant monitoring system, including an industrial television system and a monitoring terminal.
  • the industrial television system includes a plurality of image acquisition devices, and the image acquisition devices are arranged in each production area of the power plant and are used to collect images of each production area.
  • the monitoring terminal is connected to the industrial TV system through the local area network in the factory, and is used to display the image signals of each production area.
  • the power plant monitoring system further includes: an AI server and a cloud platform server;
  • the in-plant local area network is connected to each image acquisition device, which is used for AI image recognition of image signals through the AI image signal recognition algorithm model;
  • the cloud platform server includes a control cloud server and an application cloud server, and the control cloud server is connected to the AI server through the in-plant local area network. It is used to send the AI image signal recognition algorithm model to the AI server;
  • the application cloud server is connected to the AI server through the factory local area network to obtain and store the recognition results of the AI server on the image signal;
  • the application cloud server is also connected to the monitoring terminal through the public network , which is used to send the identification result to the monitoring terminal.
  • the image acquisition device includes a plurality of image acquisition units
  • the power plant monitoring system further includes a convergence switch set in each production area of the power plant; each image acquisition unit is connected to the convergence switch, and is connected to the plant through the convergence switch Intranet.
  • the application cloud platform server is further configured to connect to the production business system of the power plant, and obtain the equipment status log stored in the production business system.
  • the power plant monitoring system further includes an application server for data forwarding, and the application cloud platform server is connected to the AI server through the application server to receive the identification result forwarded by the application server; the application cloud platform server also passes the application server. Connect to the production business system to receive the device status log forwarded by the application server.
  • the AI server and the application server are deployed in the machine room of the power plant.
  • the monitoring terminal includes an in-plant monitoring terminal and an out-of-plant monitoring terminal; the in-plant monitoring terminal is connected to an industrial television system through an in-plant local area network, and is used to display image signals of each production area; the in-plant monitoring terminal is also The public network is connected to the application cloud server to receive the identification result; the off-site monitoring terminal is connected to the application cloud server through the public network to receive the identification result.
  • the monitoring terminal includes a computer device and a mobile terminal.
  • the AI server is a server cluster.
  • the image acquisition unit includes a high-definition camera.
  • the aggregation switch includes a 100M switch and a gigabit switch.
  • the embodiment of the utility model is based on the original industrial television system and monitoring terminal in the power plant, and an AI server for image recognition is added to perform image recognition on the image signals collected by the image acquisition device of the industrial television system, and obtain information about each production area of the power plant.
  • the identification result of the fault state is sent to the monitoring terminal through the application cloud server.
  • the power plant monitoring system is constructed in combination with the original industrial TV system, which reduces the cost of system construction and has a high cost-effectiveness advantage.
  • the AI server can perform AI intelligent recognition of image signals, realizing fast and accurate identification of abnormal conditions of equipment in the production area, and reducing the occurrence of misjudgments and missed judgments.
  • FIG. 1 is a schematic structural diagram of a power plant monitoring system in one embodiment
  • FIG. 2 is a schematic diagram of the connection between an image acquisition unit and an in-plant local area network in one embodiment
  • FIG. 3 is a schematic structural diagram of a power plant monitoring system in another embodiment
  • FIG. 4 is a main data path diagram of the power plant monitoring system in one embodiment.
  • connection in the following embodiments should be understood as “electrical connection”, “communication connection” and the like if there is transmission of electrical signals or data between the objects to be connected.
  • the traditional video surveillance system in the prior art has the problem of easy misjudgment and missed judgment.
  • the applicant found that the reason for this problem is that the traditional video surveillance system relies on manpower to monitor the monitoring screen. It is judged that a large amount of image information will be generated in the video monitoring process, so it will consume a lot of manpower and easily lead to misjudgments and missed judgments due to fatigue. Therefore, it is necessary to design a monitoring system for intelligent perception of abnormal status of production equipment to carry out the production process of power plants. Safety monitoring reduces manpower and material resources, improves fault early warning capability and fault judgment efficiency, and shortens fault handling time.
  • an embodiment of the present invention provides a power plant monitoring system 10 , including an industrial television system 100 , a monitoring terminal 300 , an AI server 500 and a cloud platform server 700 .
  • the industrial television system 100 includes a plurality of image acquisition devices 110, and the image acquisition devices 110 are arranged in various production areas of the power plant, such as the main transformation room, the busbar hole, the main workshop, the GIS room, the boiler area and other production areas of the power plant, and are used for acquisition. Image signals for each production area.
  • the industrial TV system 100 is currently widely used in power plants, and the staff does not need to be in the production area to observe the production, so as to avoid the personal harm to the staff caused by the harsh production environment such as toxic, harmful and high temperature.
  • the industrial television system 100 can also be used for streaming media processing and forwarding of the image signals.
  • the video can be compressed by setting a video processing server or the like in the industrial television system 100 and then transmitted based on a video transmission protocol, such as an RTP transmission protocol.
  • a video transmission protocol such as an RTP transmission protocol.
  • Commonly used video compression standards include H.265, H.264 or MPEG-4.
  • the industrial television system 100 may also be provided with a storage server for storing the collected image signals, so that the staff can call up the image signals at historical moments when needed.
  • the monitoring terminal 300 is connected to the industrial television system 100 through the intra-factory local area network, and is used for receiving image signals of each production area. Please refer to FIG. 4 , the industrial television system 100 collects and sends the image signals of each production area of the power plant to the monitoring terminal 300 , so that the power plant staff can simultaneously obtain the image signals of the production areas distributed in various places from the monitoring terminal 300 . The operating status of each production area. It can be understood that the intra-plant local area network can connect various devices in the power plant to each other through network connectors, network cables, etc., so that the devices in the power plant can exchange data and share resources.
  • the AI (artificial intelligence) server 500 is connected to each image acquisition device 110 through the factory local area network, and is used to perform AI image recognition on the image signal through the AI image signal recognition algorithm model.
  • the AI server 500 is a server that uses a processing module formed by a GPU (graphics processing unit, graphics processing unit), a TPU (tensor processing unit, tensor processing unit) and other processors combined with a CPU as a core computing unit.
  • GPU graphics processing unit, graphics processing unit
  • TPU tensor processing unit, tensor processing unit
  • AI server 500 has great performance advantages in AI image recognition.
  • AI image recognition is developing rapidly. Using a mature convolutional neural network model as the AI image signal recognition algorithm model, AI image recognition can be performed on the image signals of each production area of the power plant, and the operation status of each production area can be obtained.
  • the convolutional layer in the convolutional neural network model can perform feature extraction on the image signal. Then, the image signal after feature extraction is compressed through the pooling layer in the convolutional neural network model, and the parameters to be analyzed are reduced to improve the operation speed. Finally, the fully connected layer can comprehensively analyze the results obtained by the Chihua layer to obtain the predicted value of each recognition result, and then select the largest predicted value as the output of the recognition result.
  • the cloud platform server 700 includes a control cloud server 710 and an application cloud server 730 .
  • the control cloud server 710 is connected to the AI server 500 through the factory local area network, and is used for sending the AI image signal recognition algorithm model to the AI server 500 .
  • the control cloud server 710 may also be used to perform key authentication on the AI server 500 . After the key authentication of the AI server 500 is passed, the control cloud server 710 distributes the algorithm model to the AI server 500 .
  • the control cloud server 710 is also used to obtain the running status of the AI server 500 . Referring to FIG.
  • data exchange is performed between the control cloud server 710 and the application cloud server 730 through wireless communication.
  • the control cloud server 710 can quickly know it, and send the corresponding AI server 500 fault signal to the application cloud server 730.
  • the monitoring terminal 300 can receive the AI server 500 fault signal through the application cloud server 730, and the power plant works. The personnel can repair the AI server 500 in time according to the fault signal.
  • the application cloud server 730 is connected to the AI server 500 through the factory local area network, and is used for acquiring and storing the recognition result of the image signal by the AI server 500 .
  • the identification results include the fault type and location of the faulty equipment in each production area, as well as the type and severity of abnormal conditions in the production area. Uploading the recognition result to the application cloud server 730 can save memory for the AI server 500 and reduce the load of the AI server 500 .
  • the application cloud server 730 can also be connected to the industrial television system 100 for receiving and storing the image signals collected by the industrial television system 100 . It can be understood that the application cloud server 730 can be flexibly and flexibly configured according to the power plant's requirements for performance, memory, and the like.
  • the monitoring terminal 300 can be connected to the application cloud server 730 through the public network to obtain the identification result.
  • the monitoring terminal 300 can obtain identification results through simple operations, thereby reducing the cost of system maintenance.
  • the embodiment of the present invention adds an AI server 500 for image recognition to perform image recognition on the image signals collected by the image acquisition device 110 of the industrial television system 100, and obtain The identification result of the fault state of each production area of the power plant is sent to the monitoring terminal 300 through the application cloud server 730 .
  • the power plant monitoring system is constructed in combination with the original industrial television system 100, which reduces the construction cost of the system and has a high cost-effectiveness advantage.
  • the AI server 500 can perform AI intelligent recognition on image signals, so as to realize fast and accurate recognition of abnormal conditions of equipment in the production area, and reduce the occurrence of misjudgments and missed judgments.
  • the image acquisition device 110 includes a plurality of image acquisition units 111, and the power plant monitoring system 10 further includes a convergence switch 200 disposed in each production area of the power plant; each image acquisition unit 111 is connected to the convergence
  • the switch 200 is connected to the intra-plant local area network through the aggregation switch 200.
  • the aggregation switch 200 receives the data of all the image acquisition units 111 deployed in each production area, the aggregation switch 200 exports the data uniformly, and transmits the data to the industrial TV system 100 and the AI server 500 through the factory local area network.
  • the image acquisition device 110 may further include an auxiliary acquisition device, which is connected to the image acquisition unit 111 , such as a controllable pan/tilt, for adjusting the angle of the image acquisition unit 111 .
  • an auxiliary acquisition device which is connected to the image acquisition unit 111 , such as a controllable pan/tilt, for adjusting the angle of the image acquisition unit 111 .
  • the controllable pan-tilt is connected to the monitoring terminal 300, and the monitoring terminal 300 is used to issue a control command to adjust the angle of the image acquisition unit 111 to the controllable pan-tilt, so as to facilitate the power plant staff to monitor different positions in the production area.
  • the image acquisition unit 111 includes a high-definition camera.
  • a high-definition camera to obtain a clearer image signal is beneficial to the AI server 500 for AI image recognition and improves the accuracy of the recognition result.
  • the high-definition camera can be a high-definition industrial camera, such as a full-frame CCD industrial camera.
  • the high-definition industrial camera can also adapt to the complex environment in the power plant and work stably for a long time.
  • the aggregation switch 200 includes a Fast Switch and a Gigabit switch.
  • a gigabit switch is preferably used to prevent data loss due to insufficient bandwidth.
  • the application cloud platform server 700 is further configured to connect to the production business system 400 of the power plant, and obtain the equipment status log stored in the production business system 400 .
  • the production business system 400 is a system used by the power plant to manage and adjust the operating status of the production equipment.
  • the production business system 400 includes a status acquisition device that is connected to the production equipment.
  • the status acquisition device can collect real-time actions and status information of the equipment.
  • the production business system 400 further includes a log server, and the log server is used to obtain real-time actions and status information of devices, and generate and store device status logs of each device based on these.
  • the application cloud platform server 700 After the application cloud platform server 700 obtains the device status log from the production business system 400 , the monitoring terminal 300 has been connected to the application cloud platform server 700 through the public network, and can directly learn the status of any device from the application cloud platform server 700 . In addition, when the AI server 500 recognizes that a certain device is abnormal, the application cloud platform server 700 can also analyze the state log of the device and the identification result to further confirm whether the device is abnormal and improve the system monitoring accuracy.
  • the power plant monitoring system 10 further includes an application server 600 for data forwarding, and the application cloud platform server 700 is connected to the AI server 500 through the application server 600 for receiving the data forwarded by the application server 600 .
  • the application cloud platform server 700 is further connected to the production business system 400 through the application server 600 to receive the device status log forwarded by the application server 600 .
  • the application server 600 receives and forwards the data of each access system or server to the application cloud server 730, thereby saving network bandwidth and reducing network bottleneck and server bottleneck.
  • the AI server 500 and the application server 600 are deployed in the machine room of the power plant.
  • the AI server 500 is a server cluster.
  • a server cluster composed of multiple AI servers 500 can be used to jointly perform AI image recognition work to improve computing power and speed up AI image recognition.
  • the monitoring terminal 300 includes an on-site monitoring terminal 300 and an off-site monitoring terminal 300 .
  • the in-plant monitoring terminal 300 is connected to the industrial television system 100 through the in-plant local area network, and is used for receiving real-time image signals of each production area.
  • the external monitoring terminal 300 and the internal monitoring terminal 300 are respectively connected to the application cloud server 730 through the public network, and are used for receiving the AI image recognition result obtained by the AI server 500 .
  • the in-plant monitoring terminal 300 is set in the power plant, and the off-plant monitoring terminal 300 is set in the power grid company. Referring to FIG.
  • both the in-plant monitoring terminal 300 and the out-of-plant monitoring terminal 300 can be connected to the application cloud platform server 700 through the public network, and perform data interaction with the application cloud platform server 700 .
  • the in-plant monitoring terminal 300 and the out-of-plant monitoring terminal 300 can also obtain data such as equipment status logs through the application cloud platform server 700, so as to remotely perform business analysis and business management on the power plant.
  • the in-plant monitoring terminal 300 can also acquire and directly receive real-time image signals from each production area, so as to directly observe the situation in the power plant. This distinguishes the authority of the monitoring terminal 300 inside and outside the power plant to obtain power plant data, so as to improve system security.
  • the monitoring terminal 300 includes a computer device and a mobile terminal.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Multimedia (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

本申请涉及一种电厂监控系统。所述电厂监控系统包括：工业电视系统以及监控终端，工业电视系统包括多个设置于电厂的各生产区域内图像采集装置；监控终端用于显示各生产区域的图像信号，电厂监控系统还包括AI服务器以及云平台服务器。本申请基于电厂内原有的工业电视系统和监控终端，增设用于图像识别的AI服务器，对工业电视系统的图像采集装置所采集到的图像信号进行图像识别，获得有关电厂各生产区域的故障状态的识别结果，并通过应用云服务器将识别结果发送到监控终端。系统构建成本较低，具有较高的性价比优势。此外，AI服务器可对图像信号进行AI智能识别，实现对生产区域内设备异常情况的快速与精确识别，减少错判、漏判的情况发生。

Description

一种电厂监控系统 技术领域
本实用新型涉及电厂生产设备安全监控领域，更具体地，涉及一种电厂监控系统。
背景技术
近年来我国对能源电力的需求大幅度增加，对电力供应和电网建设也提高了重视，电厂作为我国重要的电力生产企业，安全生产是保证电网长久运行和电力企业健康发展的重要手段。在电厂生产过程中，现场环境具有电力设备体积大、电气设备多、高温高压设备多、易燃易爆和有毒物品多等特点，常发生异常情况有电力设备故障、管道破损、零部件脱落、漏油、漏水、漏气等，因此要保障电厂生产区域各项设备稳定运行或者保证故障诊断能被及时发现，需要对其进行有效监控。
传统视频监控系统存在着故障判断准确度不高，易产生误判、漏判的问题。
实用新型内容
基于此，有必要针对上述技术问题，提供一种电厂监控系统。
本实用新型实施例提供一种电厂监控系统，包括工业电视系统以及监控终端，工业电视系统包括多个图像采集装置，图像采集装置设置在电厂的各生产区域内，用于采集各生产区域的图像信号；所述监控终端通过厂内局域网与工业电视系统连接，用于显示各生产区域的所述图像信号，其特征在于，所述电厂监控系统还包括：AI服务器以及云平台服务器；AI服务器通过厂内局域网连接各图像采集装置，用于通过AI图像信号识别算法模型对图像信号进行AI图 像识别；云平台服务器包括控制云服务器与应用云服务器，控制云服务器通过厂内局域网与AI服务器连接，用于将AI图像信号识别算法模型发送到AI服务器中；应用云服务器通过厂内局域网连接AI服务器，用于获取并存储AI服务器对图像信号的识别结果；应用云服务器还通过公网连接监控终端，用于将识别结果发送至监控终端。
在其中一个实施例中，图像采集装置中包括多个图像采集单元，电厂监控系统还包括设置于电厂各生产区域内的汇聚交换机；各图像采集单元连接到汇聚交换机，并通过汇聚交换机接入厂内局域网。
在其中一个实施例中，应用云平台服务器还用于连接电厂的生产业务系统，获取生产业务系统存储的设备状态日志。
在其中一个实施例中，电厂监控系统还包括用于数据转发的应用服务器，应用云平台服务器通过应用服务器连接AI服务器，用于接收由应用服务器转发的识别结果；应用云平台服务器还通过应用服务器连接生产业务系统，用于接收由应用服务器转发的设备状态日志。
在其中一个实施例中，AI服务器以及应用服务器部署在电厂的机房内。
在其中一个实施例中，监控终端包括厂内监控终端以及厂外监控终端；厂内监控终端通过厂内局域网与工业电视系统连接，用于显示各生产区域的图像信号；厂内监控终端还通过公网连接应用云服务器，用于接收识别结果；厂外监控终端通过公网连接应用云服务器，用于接收识别结果。
在其中一个实施例中，监控终端包括计算机设备以及移动终端。
在其中一个实施例中，AI服务器为服务器集群。
在其中一个实施例中，图像采集单元包括高清摄像头。
在其中一个实施例中，汇聚交换机包括百兆交换机以及千兆交换机。
本实用新型实施例基于电厂内原有的工业电视系统和监控终端，增设用于图像识别的AI服务器，对工业电视系统的图像采集装置所采集到的图像信号进行图像识别，获得有关电厂各生产区域的故障状态的识别结果，并通过应用云服务器将识别结果发送到监控终端。本电厂监视系统结合原工业电视系统进行建设，降低系统构建成本，具有较高的性价比优势。此外，AI服务器可对图像信号进行AI智能识别，实现对生产区域内设备异常情况的快速与精确识别，减少错判、漏判的情况发生。
附图说明
为了更清楚地说明本申请实施例或传统技术中的技术方案，下面将对实施例或传统技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为一个实施例中电厂监控系统结构示意图；
图2为一个实施例中图像采集单元与厂内局域网的连接示意图；
图3为另一个实施例中电厂监控系统结构示意图；
图4为一个实施例中电厂监控系统主要数据路径图。
具体实施方式
为了便于理解本申请，下面将参照相关附图对本申请进行更全面的描述。附图中给出了本申请的实施例。但是，本申请可以以许多不同的形式来实现，并不限于本文所描述的实施例。相反地，提供这些实施例的目的是使本申请的公开内容更加透彻全面。
除非另有定义，本文所使用的所有的技术和科学术语与属于本申请的技术领域的技术人员通常理解的含义相同。本文中在本申请的说明书中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本申请。
需要说明的是，当一个元件被认为是“连接”另一个元件时，它可以是直接连接到另一个元件，或者通过居中元件连接另一个元件。此外，以下实施例中的“连接”，如果被连接的对象之间具有电信号或数据的传递，则应理解为“电连接”、“通信连接”等。
在此使用时，单数形式的“一”、“一个”和“所述/该”也可以包括复数形式，除非上下文清楚指出另外的方式。还应当理解的是，术语“包括/包含”或“具有”等指定所陈述的特征、整体、步骤、操作、组件、部分或它们的组合的存在，但是不排除存在或添加一个或更多个其他特征、整体、步骤、操作、组件、部分或它们的组合的可能性。同时，在本说明书中使用的术语“和/或”包括相关所列项目的任何及所有组合。
正如背景技术所述，现有技术中的传统视频监控系统有易产生误判、漏判的问题，经申请人研究发现，出现这种问题的原因在于，传统视频监控系统依靠人力对监控画面进行判断，视频监控过程中会产生大量的图像信息，所以会耗费大量人力且容易因疲劳而产生误判、漏判，因此有必要设计一种生产设备异常状态智能感知的监控系统，进行电厂生产过程安全监测，减少人力物力，提高故障预警能力和故障判断效率，缩短故障处理时间。
基于以上原因，请参阅图1，本实用新型实施例提供了一种电厂监控系统10，包括工业电视系统100、监控终端300、AI服务器500以及云平台服务器700。工业电视系统100包括多个图像采集装置110，图像采集装置110设置在电厂的各生产区域内，例如电厂的主变室、母线洞、主厂房、GIS室、锅炉区等 生产区域，用于采集各生产区域的图像信号。工业电视系统100目前广泛运用在电厂中，工作人员可不必身处生产区域对生产进行观察，避免有毒、有害、高温等恶劣生产环境对工作人员的人身危害。此外，图像采集装置110采集到的图像信号直接进行传输会占用很大的网络带宽以及内存，因此，工业电视系统100还可用于对图像信号进行流媒体处理与转发。具体地，可通过在工业电视系统100设置视频处理服务器等对视频进行压缩处理再基于视频传输协议，如RTP传输协议，进行传输。常用的视频压缩标准包括H.265、H.264或MPEG-4。进一步地，工业电视系统100还可设置存储服务器，用于将采集到的图像信号进行存储，以便工作人员在需要时可以调用历史时刻的图像信号。
监控终端300通过厂内局域网与工业电视系统100连接，用于接收各生产区域的图像信号。请参阅图4，工业电视系统100将电厂的各生产区域的图像信号采集并发送到监控终端300，使电厂工作人员可以从监控终端300中同时获得分布在各处的生产区域的图像信号，了解各生产区域的运行状况。可以理解，厂内局域网可将电厂内的各种设备通过网络连接器、网络电缆等互相连接，以使得电厂内的设备可以进行数据交互、资源共享等。
AI(人工智能)服务器500通过厂内局域网连接各图像采集装置110，用于通过AI图像信号识别算法模型对所述图像信号进行AI图像识别。AI服务器500是一种以GPU(graphics processing unit，图形处理器)、TPU(tensor processing unit,张量处理器)等处理器组合CPU形成的处理模组为核心计算单元的服务器。相较于传统服务器，AI服务器500在进行AI图像识别方面具有很大的性能优势。同时，目前AI图像识别发展迅速，采用成熟的卷积神经网络模型作为AI图像信号识别算法模型即可对电厂各生产区域的图像信号进行AI图像识别，得到各生产区域的运行状况。具体地，卷积神经网络模型中的卷积层可以对图像信号 的进行特征提取。再通过卷积神经网络模型中的池化层对经过特征提取处理后的图像信号进行压缩，减少待分析的参数，以提高运算速度。最后全连接层可以将池华层得到的结果进行综合分析，得到各个识别结果的预测值，然后选取预测值最大的作为识别结果输出。
云平台服务器700包括控制云服务器710与应用云服务器730。控制云服务器710通过厂内局域网与AI服务器500连接，用于将AI图像信号识别算法模型发送到AI服务器500中。通过控制云服务器710为AI服务器500分发识别算法模型，便于系统的开发与升级，提高了系统的可拓展性。进一步地，为提高系统安全性，控制云服务器710还可用于对AI服务器500进行密钥认证。当AI服务器500的密钥认证通过后，控制云服务器710才向AI服务器500分发算法模型。此外，控制云服务器710还用于获取AI服务器500的运行状态。请参阅图4，控制云服务器710与应用云服务器730之间采用无线通信方式进行数据交互。在AI服务器500出现故障时，控制云服务器710即可快速获知，并向应用云服务器730发送相应的AI服务器500故障信号，监控终端300可通过应用云服务器730接收AI服务器500故障信号，电厂工作人员即可根据故障信号及时对AI服务器500进行维修。
应用云服务器730通过厂内局域网连接AI服务器500，用于获取并存储AI服务器500对图像信号的识别结果。识别结果包括各生产区域中故障设备的故障类型、所处位置等以及生产区域的异常情况的类型、严重程度等。将识别结果上传到应用云服务器730中，可以为AI服务器500节约内存，降低AI服务器500的载荷。进一步地，应用云服务器730还可与工业电视系统100连接，用于接收并存储工业电视系统100采集到的图像信号。可以理解，应用云服务器730的可以根据电厂对性能、内存等需要进行灵活、弹性配置。将图像信号 存储到应用云服务器730中，相较于设置专用的存储服务器，后续升级更为方便，成本也较低。此外，监控终端300可以通过公网与应用云服务器730连接，获得识别结果。监控终端300经过简单的操作即可获得识别结果，降低系统维护的成本。
本实用新型实施例基于电厂内原有的工业电视系统100和监控终端300，增设用于图像识别的AI服务器500，对工业电视系统100的图像采集装置110所采集到的图像信号进行图像识别，获得有关电厂各生产区域的故障状态的识别结果，并通过应用云服务器730将识别结果发送到监控终端300。本电厂监视系统结合原工业电视系统100进行建设，降低系统构建成本，具有较高的性价比优势。此外，AI服务器500可对图像信号进行AI智能识别，实现对生产区域内设备异常情况的快速与精确识别，减少错判、漏判的情况发生。
在一个实施例中，请参阅图2，图像采集装置110中包括多个图像采集单元111，电厂监控系统10还包括设置于电厂各生产区域内的汇聚交换机200；各图像采集单元111连接到汇聚交换机200，并通过汇聚交换机200接入厂内局域网。汇聚交换机200接收部署在每个生产区域的所有图像采集单元111的数据，汇聚交换机200将这些数据统一出口，并通过厂内局域网将这些数据传输到工业电视系统100以及AI服务器500中。进一步地，图像采集装置110还可包括采集辅助设备，采集辅助设备与图像采集单元111连接，例如可控云台，用于调整图像采集单元111的角度。在图像采集单元111监视范围不足时，通过可控云台调整图像采集单元111的角度，以覆盖生产区域的不同位置，实现大范围扫描监视。此外，可控云台与监控终端300连接，监控终端300用于向可控云台发出调整图像采集单元111角度的控制命令，便于电厂工作人员监视生产区域的不同位置。
在一个实施例中，图像采集单元111包括高清摄像头。采用高清摄像头获得更为清晰的图像信号，有利于AI服务器500进行AI图像识别，提高识别结果的精度。进一步地，高清摄像头可以为高清工业摄像头，如全幅面CCD工业相机等。高清工业摄像头除了可获得高清晰度与分辨率的图像信号以外，还可适应电厂内复杂的环境，能长时间稳定地工作。
在一个实施例中，汇聚交换机200包括百兆交换机以及千兆交换机。当每个生产区域设置的图像采集单元111的数量超过100台时，优先选用千兆交换机，以防带宽不足造成数据遗失。
在一个实施例中，请参阅图3，应用云平台服务器700还用于连接电厂的生产业务系统400，获取生产业务系统400存储的设备状态日志。可以理解，生产业务系统400是电厂用于管理以及调整生产设备的运行状况的系统，生产业务系统400包括状态与生产设备连接的状态采集装置，状态采集装置可以采集设备的实时动作、状态信息等。生产业务系统400中还包括日志服务器，日志服务器用于获取设备的实时动作、状态信息等并以此为依据，生成并存储各台设备的设备状态日志。应用云平台服务器700从生产业务系统400中获取设备状态日志后，监控终端300已通过公网与应用云平台服务器700连接，可直接从应用云平台服务器700中获知任意一台设备的情况。此外，当AI服务器500识别出某台设备出现异常，应用云平台服务器700还可结合这台设备的状态日志和识别结果进行分析，进一步确认设备是否发生异常情况，提高系统监测准确性。
在一个实施例中，请继续参阅图3，电厂监控系统10还包括用于数据转发的应用服务器600，应用云平台服务器700通过应用服务器600连接AI服务器500，用于接收由应用服务器600转发的识别结果，应用云平台服务器700还通 过应用服务器600连接生产业务系统400，用于接收由应用服务器600转发的设备状态日志。请参阅图4，应用服务器600将各接入系统或服务器的数据接收并转发到应用云服务器730中，从而节省了网络带宽，减小了网络瓶颈与服务器瓶颈。
在一个实施例中，AI服务器500以及应用服务器600部署在电厂的机房内。
在一个实施例中，AI服务器500为服务器集群。在某些对AI图像识别速度有较高要求的应用场景下，可以采用多台AI服务器500组成的服务器集群共同进行AI图像识别工作，以提高计算能力，加快AI图像识别速度。
在一个实施例中，监控终端300包括厂内监控终端300以及厂外监控终端300。厂内监控终端300通过厂内局域网与工业电视系统100连接，用于接收各生产区域的实时图像信号。厂外监控终端300与厂内监控终端300分别通过公网与应用云服务器730连接，用于接收由AI服务器500得到的AI图像识别结果。其中，厂内监控终端300设置在电厂内，厂外监控终端300设置在电网公司。请参阅图4，厂内监控终端300与厂外监控终端300都可通过公网与应用云平台服务器700连接，与应用云平台服务器700进行数据交互。基于此，厂内监控终端300与厂外监控终端300还可通过应用云平台服务器700获取如设备的状态日志等数据，从而可以远程对电厂进行业务分析、业务管理等工作。而厂内监控终端300还可获取直接接收由各生产区域的实时图像信号，对电厂内的情况进行直接观察。这区分了电厂内外监控终端300获取电厂数据的权限，以提高系统安全性。
在一个实施例中，监控终端300包括计算机设备以及移动终端。
在本说明书的描述中，参考术语“有些实施例”、“其他实施例”、“理想实施例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或 者特征包含于本实用新型的至少一个实施例或示例中。在本说明书中，对上述术语的示意性描述不一定指的是相同的实施例或示例。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对实用新型专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种电厂监控系统，包括工业电视系统以及监控终端，所述工业电视系统包括多个图像采集装置，所述图像采集装置设置在电厂的各生产区域内，用于采集各所述生产区域的图像信号；所述监控终端通过厂内局域网与所述工业电视系统连接，用于显示各所述生产区域的所述图像信号，其特征在于，所述电厂监控系统还包括：AI服务器以及云平台服务器；
    所述AI服务器通过所述厂内局域网连接各所述图像采集装置，用于通过AI图像信号识别算法模型对所述图像信号进行AI图像识别；
    所述云平台服务器包括控制云服务器与应用云服务器，所述控制云服务器通过所述厂内局域网与所述AI服务器连接，用于将所述AI图像信号识别算法模型发送到所述AI服务器中；所述应用云服务器通过所述厂内局域网连接AI服务器，用于获取并存储所述AI服务器对图像信号的识别结果；所述应用云服务器还通过公网连接所述监控终端，用于将所述识别结果发送至所述监控终端。
  2. 根据权利要求1所述的电厂监控系统，其特征在于，所述图像采集装置中包括多个图像采集单元，所述电厂监控系统还包括设置于电厂各生产区域内的汇聚交换机；
    各所述图像采集单元连接到所述汇聚交换机，并通过所述汇聚交换机接入所述厂内局域网。
  3. 根据权利要求1所述的电厂监控系统，其特征在于，所述应用云平台服务器还用于连接电厂的生产业务系统，获取所述生产业务系统存储的设备状态日志。
  4. 根据权利要求3所述的电厂监控系统，其特征在于，所述电厂监控系统还包括用于数据转发的应用服务器，所述应用云平台服务器通过所述应用服务 器连接所述AI服务器，用于接收由所述应用服务器转发的所述识别结果；所述应用云平台服务器还通过所述应用服务器连接所述生产业务系统，用于接收由所述应用服务器转发的所述设备状态日志。
  5. 根据权利要求4所述的电厂监控系统，其特征在于，所述AI服务器以及所述应用服务器部署在所述电厂的机房内。
  6. 根据权利要求1所述的电厂监控系统，其特征在于，所述监控终端包括厂内监控终端以及厂外监控终端；
    所述厂内监控终端通过厂内局域网与所述工业电视系统连接，用于接收各所述生产区域的所述图像信号；所述厂内监控终端还通过公网连接所述应用云服务器，用于接收所述识别结果；
    所述厂外监控终端通过公网连接所述应用云服务器，用于接收所述识别结果。
  7. 根据权利要求1所述的电厂监控系统，其特征在于，所述监控终端包括计算机设备或移动终端。
  8. 根据权利要求1所述的电厂监控系统，其特征在于，所述AI服务器为服务器集群。
  9. 根据权利要求2所述的电厂监控系统，其特征在于，所述图像采集单元包括高清摄像头。
  10. 根据权利要求2所述的电厂监控系统，其特征在于，所述汇聚交换机包括百兆交换机或千兆交换机。
PCT/CN2021/079288 2021-01-29 2021-03-05 一种电厂监控系统 WO2022160414A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202120260331.4U CN214177452U (zh) 2021-01-29 2021-01-29 一种电厂监控系统
CN202120260331.4 2021-01-29

Publications (1)

Publication Number Publication Date
WO2022160414A1 true WO2022160414A1 (zh) 2022-08-04

Family

ID=77594353

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/079288 WO2022160414A1 (zh) 2021-01-29 2021-03-05 一种电厂监控系统

Country Status (2)

Country Link
CN (1) CN214177452U (zh)
WO (1) WO2022160414A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116482714A (zh) * 2023-03-29 2023-07-25 国家电投集团郑州燃气发电有限公司 基于安全双重预防体系的电厂卫星定位系统

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114758296B (zh) * 2022-04-19 2023-02-24 国网湖北省电力有限公司黄石供电公司 一种基于vr技术的电网设备远程监测方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110160922A1 (en) * 2009-12-30 2011-06-30 Eduardo Pedrosa Santos Decentralized system and architecture for remote real time monitoring of power transformers, reactors, circuit breakers, instrument transformers, disconnect switches and similar high voltage equipment for power plants and electric power substations
CN211293956U (zh) * 2019-11-14 2020-08-18 北京国电通网络技术有限公司 基于ai的施工现场异常行为人的识别和告警系统
CN211406109U (zh) * 2020-04-08 2020-09-01 深圳江行联加智能科技有限公司 一种具备ai识别功能的便携视频移动监控系统
CN112188164A (zh) * 2020-09-29 2021-01-05 爱动超越人工智能科技（北京）有限责任公司 一种基于ai视觉的违规行为实时监测系统及方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110160922A1 (en) * 2009-12-30 2011-06-30 Eduardo Pedrosa Santos Decentralized system and architecture for remote real time monitoring of power transformers, reactors, circuit breakers, instrument transformers, disconnect switches and similar high voltage equipment for power plants and electric power substations
CN211293956U (zh) * 2019-11-14 2020-08-18 北京国电通网络技术有限公司 基于ai的施工现场异常行为人的识别和告警系统
CN211406109U (zh) * 2020-04-08 2020-09-01 深圳江行联加智能科技有限公司 一种具备ai识别功能的便携视频移动监控系统
CN112188164A (zh) * 2020-09-29 2021-01-05 爱动超越人工智能科技（北京）有限责任公司 一种基于ai视觉的违规行为实时监测系统及方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116482714A (zh) * 2023-03-29 2023-07-25 国家电投集团郑州燃气发电有限公司 基于安全双重预防体系的电厂卫星定位系统
CN116482714B (zh) * 2023-03-29 2024-06-11 国家电投集团郑州燃气发电有限公司 基于安全双重预防体系的电厂卫星定位系统

Also Published As

Publication number Publication date
CN214177452U (zh) 2021-09-10

Similar Documents

Publication Publication Date Title
CN104581087B (zh) 一种基于物联网的输电线路智能远程监控系统
CN110733038A (zh) 工业机器人远程监控及数据处理系统
WO2022160414A1 (zh) 一种电厂监控系统
CN109617240A (zh) 一种电力用户配电室运维系统
CN202143171U (zh) 基于移动机器人的智能化变电站视频综合监控系统
CN104079885A (zh) 无人监守联动跟踪的网络摄像方法及装置
CN106843183B (zh) 基于vpn技术的干式水泥线调试的远程监控系统及控制方法
CN104218679B (zh) 一种基于遥视系统的变电站远程自动巡检方法
CN111061171A (zh) 配电房监控系统、配电房数据处理、调用方法及其装置
CN114243932B (zh) 一种变电站视频及环境监控站端系统智能运维终端
CN104375485A (zh) 一种用于变配电安全生产的辅助监控系统及其监控方法
CN101958582A (zh) 变电站视频及环境监控省级主站系统
CN101976888A (zh) 变电站视频及环境监控主站系统
CN202535476U (zh) 一种电力设备视频监控系统
CN105740993A (zh) 变电站运维巡检系统及其实现方法
CN102801960A (zh) 电力系统远程设备视频监控系统及其数据处理方法
CN214314723U (zh) 一种基于5g云控平台的智能配电站监控系统
CN109495563A (zh) 智能水务可视化神经网络监控系统
CN210380926U (zh) 一种泛在电力物联网智能网关设备
CN113258677A (zh) 一种含低碳调控与预警抢修功能的智能配电柜(箱)装置及系统
CN209895464U (zh) 一种变电站异常监测报警系统
CN204155134U (zh) 一种光伏组件发电集中监测系统
CN110676935B (zh) 一种变电站智能运维闭环操作的系统及其方法
CN205070609U (zh) 一种新型配电自动化视频在线监控系统
CN204887249U (zh) 一种无线视频监控系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21922013

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21922013

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 21922013

Country of ref document: EP

Kind code of ref document: A1