[go: up one dir, main page]

CN115526871A - A method for detecting the state of an isolating switch and a camera - Google Patents

A method for detecting the state of an isolating switch and a camera Download PDF

Info

Publication number
CN115526871A
CN115526871A CN202211225772.6A CN202211225772A CN115526871A CN 115526871 A CN115526871 A CN 115526871A CN 202211225772 A CN202211225772 A CN 202211225772A CN 115526871 A CN115526871 A CN 115526871A
Authority
CN
China
Prior art keywords
isolating switch
state
detection
mosaic
diagram
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211225772.6A
Other languages
Chinese (zh)
Inventor
佘楚云
牛犇
郑润蓝
王兵
张晶焯
杨丰阁
高德民
郝越
李姝玉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Power Supply Bureau Co Ltd
Original Assignee
Shenzhen Power Supply Bureau Co Ltd
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 Shenzhen Power Supply Bureau Co Ltd filed Critical Shenzhen Power Supply Bureau Co Ltd
Priority to CN202211225772.6A priority Critical patent/CN115526871A/en
Publication of CN115526871A publication Critical patent/CN115526871A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0004Industrial image inspection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/20Analysis of motion
    • G06T7/246Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G06V10/765Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/52Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2200/00Indexing scheme for image data processing or generation, in general
    • G06T2200/32Indexing scheme for image data processing or generation, in general involving image mosaicing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30108Industrial image inspection
    • G06T2207/30164Workpiece; Machine component

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Quality & Reliability (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a method for detecting the state of an isolating switch, which comprises the steps that a camera collects real-time videos generated when the isolating switch operates and carries out image interception according to frames to obtain a plurality of isolating switch field moving images based on a frame sequence; splicing the field motion image and the opening and closing template graph of each isolating switch to obtain opening and closing splicing graphs of the field motion images of all isolating switches, and performing target detection by combining a trained isolating switch state detection model to obtain detection results of the field motion images of all isolating switches based on the opening and closing splicing graphs respectively; and determining the operation state of the isolating switch corresponding to each isolating switch field motion image based on the frame sequence according to two detection results of the same isolating switch field motion image based on the switching-closing splicing diagram. By implementing the invention, the state of the isolating switch can be rapidly detected based on the samples acquired by the camera at any angle, the detection precision is improved, and the number of samples is reduced to reduce the workload.

Description

一种隔离开关状态检测方法及摄像头A method for detecting the state of an isolating switch and a camera

技术领域technical field

本发明涉及隔离开关技术领域,尤其涉及一种隔离开关状态检测方法及摄像头。The invention relates to the technical field of isolating switches, in particular to a method for detecting the state of an isolating switch and a camera.

背景技术Background technique

变电站作为电网的重要组成部分,其运行状态是决定电网是否能够安全运行的重要因素之一。过去老式的变电站是通过人工检查来识别设备的实际运行状况,导致存在着不及时、不安全等隐患和缺陷。为了顺应智能变电站的未来发展方向,变电站的无人值守化已逐渐成为当今电网调度自动化的发展趋势。因此,有必要对变电站开关设备的开合状态进行实时监控,用以保障变电站的正常运行。As an important part of the power grid, the substation's operating status is one of the important factors that determine whether the power grid can operate safely. In the past, the old-fashioned substations used manual inspections to identify the actual operating conditions of the equipment, resulting in hidden dangers and defects such as untimely and unsafe conditions. In order to comply with the future development direction of smart substations, unattended substations have gradually become the development trend of today's power grid dispatching automation. Therefore, it is necessary to monitor the opening and closing status of the substation switchgear in real time to ensure the normal operation of the substation.

目前,隔离开关是电网系统中使用最广泛及安装量最大的设备。隔离开关若长时间的运行,可能会出现锈蚀卡涩的现象,从而容易造成分合闸未到位情况。因此,隔离开关的运行状态是决定变电站能够安全运行的关键因素之一,其状态的自动识别对电网生产运行监控有着重要意义。At present, the isolating switch is the most widely used and installed equipment in the power grid system. If the isolating switch is operated for a long time, it may be corroded and jammed, which may easily cause the opening and closing not in place. Therefore, the operating state of the isolating switch is one of the key factors that determine the safe operation of the substation, and its automatic identification of the state is of great significance to the production and operation monitoring of the power grid.

在现有技术中,利用深度学习算法对隔离开关进行状态检测及识别。但是,该基于深度学习算法的隔离开关状态检测方法完全依赖于样本数据,需采集大量的隔离开关现场运动图像并标注,导致工作量较大。另外,在摄像机拍摄角度变化后,隔离开关现场运动图像中隔离开关的开关臂的开合角度也会显现出明显变化,会严重影响到模型识别结果,无法做到精确检测,这就造成隔离开关现场运动图像采集时需要固定摄像头角度。In the prior art, a deep learning algorithm is used to detect and identify the state of the isolating switch. However, this method of isolating switch state detection based on deep learning algorithm is completely dependent on sample data, and a large number of on-site moving images of isolating switches need to be collected and labeled, resulting in a large workload. In addition, after the shooting angle of the camera changes, the opening and closing angle of the switch arm of the isolator in the on-site moving image of the isolator will also show obvious changes, which will seriously affect the model recognition results and cannot be accurately detected. The camera angle needs to be fixed when capturing live motion images.

因此,有必要提出一种新的隔离开关状态检测方法,不仅可以基于摄像头任意角度采集的样本进行隔离开关状态快速检测,提高了检测精度,还减少了样本数量来降低工作量。Therefore, it is necessary to propose a new isolation switch state detection method, which can not only quickly detect the isolation switch state based on the samples collected by the camera at any angle, improve the detection accuracy, but also reduce the number of samples to reduce the workload.

发明内容Contents of the invention

本发明实施例所要解决的技术问题在于,提供一种隔离开关状态检测方法及摄像头,不仅可以基于摄像头任意角度采集的样本进行隔离开关状态快速检测,提高了检测精度,还减少了样本数量来降低工作量。The technical problem to be solved by the embodiments of the present invention is to provide a method for detecting the state of an isolating switch and a camera, which can not only quickly detect the state of the isolating switch based on samples collected from any angle of the camera, improve the detection accuracy, but also reduce the number of samples to reduce the workload.

为了解决上述技术问题,本发明实施例提供了一种隔离开关状态检测方法,其在摄像头上实现,所述方法包括以下步骤:In order to solve the above technical problems, an embodiment of the present invention provides a method for detecting the state of an isolating switch, which is implemented on a camera, and the method includes the following steps:

所述摄像头采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;The camera collects the real-time video of the isolating switch in operation and intercepts the image frame by frame, so as to obtain a plurality of on-site moving images of the isolating switch based on the frame sequence;

将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;Splicing the on-site moving image of each isolating switch with the preset opening template diagram and the preset closing template diagram to obtain the corresponding opening mosaic diagram and closing mosaic diagram of all the on-site moving images of the isolating switch, and Combining the trained isolating switch state detection model for target detection, to obtain the detection results of all the on-site moving images of the isolating switches based on the opening mosaic diagram and the closing mosaic diagram; wherein, the detection results include the result being changed and the result Is empty;

根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。According to the two detection results obtained based on the opening mosaic diagram and the closing mosaic diagram of the same isolating switch on-site moving image, the operating state of the isolating switch corresponding to each isolating switch on-site moving image based on the frame sequence is determined; wherein the operation The state is one of the opening state, closing state and intermediate state.

其中,所述根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态的具体步骤包括:Wherein, according to the two detection results obtained based on the on-site moving image of the same isolating switch based on the opening mosaic diagram and the closing mosaic diagram, the specific method of determining the operating state of the isolating switch corresponding to each on-site moving image of the isolating switch based on the frame sequence Steps include:

在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态;When the trained isolating switch state detection model is used for target detection, if the results obtained by the detection of the mosaic diagram of the current isolating switch site moving image are the same as those detected by the mosaic diagram of the closing mosaic diagram, and both detection results are If there is a change, it is determined that the current operating state of the on-site moving image of the isolating switch is an intermediate state;

在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;反之,待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。When the trained isolating switch state detection model performs target detection, if the opening mosaic diagram of the current isolating switch scene moving image is different from the result detected by its closing mosaic diagram, and one of the detection results is that the result is empty And another detection result is that the result is changed, then after it is judged that the mosaic diagram corresponding to the detection result is empty is the mosaic diagram of opening, determine that the running state of the current moving image of the isolating switch is the opening state; otherwise, to be judged After the mosaic diagram corresponding to the detection result is empty is the closing mosaic diagram, it is determined that the current running state of the on-site moving image of the isolating switch is the closing state.

其中,所述方法进一步包括:Wherein, the method further includes:

在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;From the live moving images of multiple isolating switches based on the frame sequence, the live moving images of the isolating switches whose operation status is in the intermediate state are sequentially extracted;

将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合所述已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,所述运动状态为静止或正在运动。Splicing the extracted on-site moving image of each isolating switch in the intermediate state with the on-site moving image of the isolating switch in the next frame to obtain the corresponding front and rear frame mosaic images of the on-site moving images of the isolating switch in the intermediate state, and combining the The trained isolating switch state detection model performs target detection to obtain the detection results of all intermediate state isolating switch live moving images based on the front and rear frame mosaic images, and further according to the detection results of all intermediate state isolating switch live moving images based on the front and rear frames respectively The detection result of the mosaic diagram determines the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is static or moving.

其中,所述根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态的具体步骤包括:Wherein, the specific steps of determining the motion state of the isolating switch corresponding to all intermediate state isolating switch live moving images based on the detection results of the front and rear frame mosaic images respectively according to the on-site moving images of the isolating switch in the intermediate state include:

在所述已训练好的隔离开关状态检测模型进行目标检测时,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为空,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止;反之,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为有变化,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为正在运动。When the trained isolating switch state detection model is used for target detection, if the detected results of the on-site motion images of the isolating switch in a certain intermediate state based on the front and rear frame mosaic images are empty, then it is determined that the isolating switch corresponds to the currently detected The motion state of the disconnector in the intermediate state is stationary; on the contrary, if the live motion images of a disconnector in the intermediate state are detected based on the detection results of the front and rear frame mosaic images, it is determined that the disconnector corresponds to the current detection The motion state of the live motion image of the isolating switch in the intermediate state is in motion.

其中,所述方法进一步包括:Wherein, the method further includes:

在确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止之后,进行报警。After it is determined that the motion state of the isolating switch corresponding to the live moving image of the isolating switch in the currently detected intermediate state is static, an alarm is given.

本发明实施例还提供了一种摄像头,包括:The embodiment of the present invention also provides a camera, including:

视频采集及图像截取单元,用于采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;The video acquisition and image interception unit is used to collect the real-time video of the isolation switch during operation and perform image interception by frame to obtain live moving images of multiple isolation switches based on the frame sequence;

隔离开关运行检测单元,用于将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;The isolating switch operation detection unit is used to splice the on-site moving image of each isolating switch with the preset opening template map and the preset closing template map to obtain the corresponding opening splicing of all the isolating switch on-site moving images diagram and closing mosaic diagram, and combined with the trained isolation switch state detection model to carry out target detection, to obtain the detection results based on the opening mosaic diagram and closing mosaic diagram of all isolation switch scene motion images respectively; wherein, the detection Result includes result is changed and result is empty;

隔离开关运行状态识别单元,用于根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。The operating state recognition unit of the isolating switch is used to determine when the isolating switch corresponds to each live moving image of the isolating switch based on the frame sequence based on the two detection results obtained from the on-site moving image of the same isolating switch based on the opening mosaic diagram and the closing mosaic diagram. The operating state; wherein, the operating state is one of the opening state, the closing state and the intermediate state.

其中,所述隔离开关状态最终识别单元包括:Wherein, the final identification unit of the isolating switch state includes:

中间态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态;The intermediate state recognition module is used to detect the target when the trained isolating switch state detection model, if the result obtained by the detection of the opening mosaic diagram of the current isolating switch scene moving image and its closing mosaic diagram is the same, and the two If all the detection results show that the results have changed, then it is determined that the operating state of the current isolating switch scene moving image is an intermediate state;

分合闸态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;反之,待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。The opening and closing state recognition module is used for target detection when the trained isolating switch state detection model is used, if the opening mosaic diagram of the current isolating switch scene motion image is different from the result obtained by the detection of the closing mosaic diagram, And one test result is empty and the other test result is change, then after it is judged that the mosaic diagram corresponding to the detection result is empty is the opening mosaic diagram, determine the running status of the current isolating switch on-site moving image On the contrary, after it is judged that the mosaic diagram corresponding to the detection result is empty is the closing mosaic diagram, it is determined that the current operating state of the on-site moving image of the isolating switch is the closing state.

其中,还包括:Among them, also include:

中间态运动识别单元,用于在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;以及The intermediate state motion recognition unit is used to sequentially extract the on-site moving images of the isolating switches whose operating state is in the intermediate state from among the on-site moving images of the multiple isolating switches based on the frame sequence; and

将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合所述已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,所述运动状态为静止或正在运动。Splicing the extracted on-site moving image of each isolating switch in the intermediate state with the on-site moving image of the isolating switch in the next frame to obtain the corresponding front and rear frame mosaic images of the on-site moving images of the isolating switch in the intermediate state, and combining the The trained isolating switch state detection model performs target detection to obtain the detection results of all intermediate state isolating switch live moving images based on the front and rear frame mosaic images, and further according to the detection results of all intermediate state isolating switch live moving images based on the front and rear frames respectively The detection result of the mosaic diagram determines the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is static or moving.

其中,所述隔离开关状态检测模型为基于yolo v3算法的神经网络模型,其输入数据为拼接图,输出类别为有变化。Wherein, the isolating switch state detection model is a neural network model based on the yolo v3 algorithm, its input data is a mosaic graph, and its output category is changed.

实施本发明实施例,具有如下有益效果:Implementing the embodiment of the present invention has the following beneficial effects:

1、本发明中摄像头通过按帧方式将实时视频分解为基于帧序列的多个隔离开关现场运动图像,并将每一个隔离开关现场运动图像与预设的分闸模板图及合闸模板图进行拼接后,利用已训练好的隔离开关状态检测模型进行目标检测,得到任意一个隔离开关现场运动图像的两个拼接图的检测结果,且通过对比同一个隔离开关现场运动图像的两个拼接图的检测结果来快速确定隔离开关每一帧的运行状态,从而不用考虑摄像头所拍摄的视频角度是否发生变化,就可以进行隔离开关状态快速检测,从而实现基于摄像头任意角度采集的样本就可进行隔离开关状态快速检测,提高了检测精度,还减少了样本数量,降低了工作量;1. In the present invention, the camera decomposes the real-time video into a plurality of on-site moving images of isolating switches based on the frame sequence by frame, and compares the on-site moving images of each isolating switch with the preset opening template diagram and closing template diagram. After splicing, use the trained isolation switch state detection model to detect the target, and obtain the detection results of the two mosaic images of any live moving image of an isolating switch, and compare the two mosaic images of the same isolating switch live moving image. The detection results can be used to quickly determine the operating status of each frame of the isolation switch, so that the status of the isolation switch can be quickly detected without considering whether the angle of the video captured by the camera changes, so that the isolation switch can be performed based on samples collected from any angle of the camera. Rapid status detection improves detection accuracy, reduces the number of samples, and reduces workload;

2、本发明中摄像头还通过将中间态的每一个隔离开关现场运动图像均与其后一帧的运动图像进行拼接后,利用已训练好的隔离开关状态检测模型进行目标检测,快速对隔离开关在中间态的每一帧图像上的运动状态进行定位,以实现隔离开关的运动过程中的缺陷进行检测及报警。2. In the present invention, the camera also uses the trained isolating switch state detection model to perform target detection by splicing the on-site moving image of each isolating switch in the intermediate state with the moving image of the next frame, and quickly detects the isolating switch in the state. The motion state on each frame image of the intermediate state is positioned to realize the detection and alarm of defects in the motion process of the isolating switch.

附图说明Description of drawings

为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有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发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,根据这些附图获得其他的附图仍属于本发明的范畴。In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, obtaining other drawings based on these drawings still belongs to the scope of the present invention without any creative effort.

图1为本发明实施例提供的一种隔离开关状态检测方法的流程图;Fig. 1 is a flow chart of a method for detecting the state of an isolating switch provided by an embodiment of the present invention;

图2为本发明实施例提供的一种摄像头的系统结构示意图。FIG. 2 is a schematic structural diagram of a camera system provided by an embodiment of the present invention.

具体实施方式detailed description

为使本发明的目的、技术方案和优点更加清楚,下面将结合附图对本发明作进一步地详细描述。In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.

如图1所示,为本发明实施例中,提供的一种隔离开关状态检测方法,其在摄像头上实现,所述方法包括以下步骤:As shown in Figure 1, it is an isolating switch state detection method provided in an embodiment of the present invention, which is implemented on a camera, and the method includes the following steps:

步骤S1、所述摄像头采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;Step S1, the camera collects the real-time video of the isolating switch during operation and intercepts the image frame by frame, so as to obtain a plurality of live moving images of the isolating switch based on the frame sequence;

步骤S2、将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;Step S2. Splicing the on-site motion images of each isolating switch with the preset opening template diagram and the preset closing template diagram to obtain the corresponding opening mosaic diagrams and closing mosaic diagrams of all the on-site motion images of the isolating switches , and combined with the trained isolating switch state detection model to carry out target detection, to obtain the detection results based on the opening mosaic diagram and closing mosaic diagram of all the on-site motion images of the isolating switch respectively; wherein, the detection results include results that have change and result is empty;

步骤S3、根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。Step S3, according to the two detection results obtained based on the opening mosaic diagram and the closing mosaic diagram of the same isolating switch on-site moving image, determine the operating state of the isolating switch corresponding to each isolating switch on-site moving image based on the frame sequence; wherein, The operation state is one of an open state, a closed state and an intermediate state.

具体过程为,在步骤S1之前,摄像头中预存有隔离开关状态检测模型。该隔离开关状态检测模型为基于yolo v3算法的神经网络模型,其输入数据为拼接图,输出类别为有变化。当然,该隔离开关状态检测模型也可以采用其它算法,可根据实际需要进行设计,在此不再一一列举相应算法的种类。The specific process is that, before step S1, a detection model of the isolating switch state is pre-stored in the camera. The isolating switch state detection model is a neural network model based on the yolo v3 algorithm, and its input data is a mosaic graph, and the output category is a change. Of course, the isolating switch state detection model can also use other algorithms, which can be designed according to actual needs, and the types of corresponding algorithms will not be listed one by one here.

其次,摄像头中构建出隔离开关状态检测模型所需的样本数据。该样本数据的构建过程如下:Secondly, the sample data required for the state detection model of the isolation switch is constructed in the camera. The construction process of this sample data is as follows:

(1)截取隔离开关分闸到位的图像作为分闸模板图,以及截取隔离开关合闸到位的图像作为合闸模板图。(1) Capture the image of the isolating switch in place as the opening template diagram, and intercept the image of the isolating switch in place as the closing template diagram.

(2)截取隔离开关运动过程中不同位置的多个中间态图像,并将所有中间态图像均与分闸模板图进行拼接,以及将所有中间态图像均与合闸模板图进行拼接,且进一步将所有拼接图作为样本数据并进行标注。应当说明的是,由于隔离开关运动过程中,主要是开关臂在进行变化,所以标注开关臂分为分闸状态、中间态以及合闸状态这三类来生成标注文件。另外,除隔离开关外,还会增加其他类别图片,如同一场景下门未关闭和关闭的样本、桌面干净无杂物和桌面有垃圾的样本等等,同样进行拼接并标注。(2) Intercept a plurality of intermediate state images at different positions during the movement of the isolating switch, and splicing all intermediate state images with the opening template map, and splicing all intermediate state images with the closing template map, and further Take all mosaics as sample data and label them. It should be noted that during the movement of the isolating switch, the switch arm is mainly changing, so the marking switch arm is divided into three categories: the opening state, the intermediate state and the closing state to generate the marking file. In addition, in addition to the isolating switch, other types of pictures will be added, such as samples of the door not closed and closed in the same scene, samples of clean desktops and desktops with garbage, etc., which will also be stitched and marked.

(3)将样本数据制作样本数据集。(3) Make a sample data set from the sample data.

最后,对隔离开关状态检测模型进行训练及测试。将样本数据划分出训练样本和测试样本,利用训练样本对隔离开关状态检测模型进行训练,并通过测试样本进行验证,从而得到训练好的隔离开关状态检测模型。Finally, the state detection model of the isolation switch is trained and tested. Divide the sample data into training samples and test samples, use the training samples to train the isolation switch state detection model, and verify it through the test samples, so as to obtain the trained isolation switch state detection model.

在步骤S1中,摄像头采集隔离开关运行时的实时视频,并通过预存的视频分析软件对实时视频按帧进行图像截取,从而得到基于帧序列(如第1帧、第2帧、…、第n帧)的多个隔离开关现场运动图像。In step S1, the camera collects the real-time video during the operation of the isolating switch, and uses the pre-stored video analysis software to intercept the real-time video frame by frame, so as to obtain the frame-based sequence (such as the first frame, the second frame, ..., the nth frame) frame) live motion images of multiple isolating switches.

在步骤S2中,首先,摄像头将每一个隔离开关现场运动图像均与分闸模板图及合闸模板图进行拼接,得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图。In step S2, firstly, the camera splices the on-site moving images of each isolating switch with the opening template diagram and the closing template diagram to obtain the corresponding opening and closing mosaic diagrams of all the on-site moving images of the isolating switches.

其次,摄像头将每一个隔离开关现场运动图像的分闸拼接图导入已训练好的隔离开关状态检测模型进行目标检测,以及将每一个隔离开关现场运动图像的合闸拼接图导入已训练好的隔离开关状态检测模型进行目标检测,从而得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果,即同一个隔离开关现场运动图像有两个检测结果。Secondly, the camera imports the opening mosaic image of each isolating switch live moving image into the trained isolating switch state detection model for target detection, and imports the closing mosaic image of each isolating switch live moving image into the trained isolation switch The switch state detection model performs target detection, so as to obtain the detection results of all the on-site moving images of the isolating switch based on the opening mosaic diagram and the closing mosaic diagram, that is, there are two detection results for the same isolating switch on-site moving image.

此时,若检测到隔离开关现场运动图像与分闸拼接图或合闸拼接图无差别,则检测结果输出为空;反之,若检测到隔离开关现场运动图像与分闸拼接图或合闸拼接图有差别,则检测结果输出为有变化,该检测结果会显示在隔离开关现场运动图像中与分闸拼接图或合闸拼接图有差异的地方。At this time, if it is detected that there is no difference between the on-site motion image of the isolating switch and the opening mosaic diagram or closing mosaic diagram, the detection result output is empty; If there is a difference in the diagram, the output of the detection result is a change, and the detection result will be displayed in the place where there is a difference between the opening mosaic diagram or the closing mosaic diagram in the on-site moving image of the isolating switch.

在步骤S3中,隔离开关的运行状态有分闸状态、合闸状态及中间态这三种,且这三种状态可以通过对比每一个隔离开关现场运动图像的两个检测结果进行快速的定位,具体过程如下:In step S3, there are three operating states of the isolating switch: the opening state, the closing state and the intermediate state, and these three states can be quickly located by comparing the two detection results of the on-site motion images of each isolating switch, The specific process is as follows:

(1)中间态:在已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态,即隔离开关当前位置既不在合闸到位的位置,也不在分闸到位的位置。(1) Intermediate state: When the trained isolating switch state detection model is used for target detection, if the detection results of the opening mosaic diagram of the current isolating switch scene moving image and its closing mosaic diagram are the same, and the two detection results If the results are all changed, then it is determined that the current running state of the on-site moving image of the isolating switch is an intermediate state, that is, the current position of the isolating switch is neither in the closed position nor in the open position.

(2)分闸状态:在已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;(2) Opening state: When the trained isolating switch state detection model is used for target detection, if the detection results of the opening mosaic diagram of the current isolating switch scene moving image are different from those detected by the closing mosaic diagram, and there is a detection The result is that the result is empty and another detection result is that the result has changed. After the mosaic diagram corresponding to the detection result is judged to be empty, it is determined that the current running state of the on-site moving image of the isolating switch is the opening state. ;

(3)合闸状态:在已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。(3) Closing state: When the trained isolating switch state detection model is used for target detection, if the detection results of the opening mosaic diagram of the current isolating switch scene moving image are different from those detected by the closing mosaic diagram, and there is a detection The result is that the result is empty and the other detection result is that the result has changed. After the mosaic diagram corresponding to the detection result is judged to be empty is the closing mosaic diagram, determine that the current running state of the on-site moving image of the isolating switch is the closing state. .

在本发明实施例中,为了避免隔离开关处于未合到位或未分到位的安全隐患,需要实时监测隔离开关分合闸运动的中断情况。因此,摄像头在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;In the embodiment of the present invention, in order to avoid the potential safety hazard that the isolating switch is not in place or not in place, it is necessary to monitor the interruption of the opening and closing movement of the isolating switch in real time. Therefore, the camera sequentially extracts the on-site moving images of the isolating switches whose operating state is in the intermediate state from the live moving images of multiple isolating switches based on the frame sequence;

将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,运动状态为静止或正在运动。Each of the isolated switch live motion images extracted in the intermediate state are spliced with the subsequent frame of the disconnector live motion images to obtain the corresponding front and rear frame mosaic images of all the intermediate state disconnector live motion images, and combined with the trained A good isolating switch state detection model performs target detection to obtain the detection results of all intermediate-state isolating switch live moving images based on the front and rear frame mosaic images, and further according to all intermediate state isolating switch live moving images based on the front and rear frame mosaic images The detection results determine the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is stationary or in motion.

此时,在已训练好的隔离开关状态检测模型进行目标检测时,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为空,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止;反之,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为有变化,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为正在运动。即,检测到中间态的前后帧隔离开关现场运动图像间存在不同之处,若在后一帧图像上检测到缺陷,那么隔离开关处于正在运动状态;反之,未检测到中间态的前后帧隔离开关现场运动图像间存在不同之处,在后一帧图像上未检测缺陷,则隔离开关处于静止状态,那么隔离开关处于未合到位或未分到位,存在安全隐患,需要进一步的进行告警。At this time, when the trained isolating switch state detection model is used for target detection, if the live moving images of a certain intermediate state isolating switch based on the detection results of the front and rear frame mosaic images are empty, then it is determined that the isolating switch corresponds to the current detection The moving state of the isolating switch in the intermediate state is static; on the contrary, if the moving images of the isolating switch in the intermediate state are detected based on the results of the detection of the front and rear frame mosaic images, it is determined that the isolating switch corresponds to the current state. The motion state when the live motion image of the isolating switch in the detected intermediate state is in motion. That is to say, there are differences between the on-site moving images of the isolation switch before and after the intermediate state are detected. If a defect is detected on the next frame image, the isolation switch is in a moving state; otherwise, the isolation switch before and after the intermediate state is not detected. There are differences between the moving images of the switch scene. If no defect is detected on the next frame image, the isolating switch is in a static state. Then the isolating switch is not in place or not in place. There are potential safety hazards, and further alarms are required.

因此,在确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止之后,进行报警。Therefore, after it is determined that the motion state of the isolating switch corresponding to the live motion image of the isolating switch in the currently detected intermediate state is static, an alarm is given.

如图2所示,为本发明实施例中,提供的一种摄像头,包括:As shown in Figure 2, it is a camera provided in the embodiment of the present invention, including:

视频采集及图像截取单元110,用于采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;The video collection and image interception unit 110 is used to collect the real-time video of the disconnector during operation and perform image interception by frame, so as to obtain live moving images of multiple disconnectors based on the frame sequence;

隔离开关运行检测单元120,用于将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;The isolating switch operation detection unit 120 is used for splicing the on-site moving image of each isolating switch with the preset opening template diagram and the preset closing template diagram, so as to obtain the opening corresponding to the on-site moving images of all isolating switches Mosaic diagram and closing mosaic diagram, combined with the well-trained isolation switch state detection model to carry out target detection, to obtain the detection results based on the opening mosaic diagram and closing mosaic diagram of all isolating switch scene motion images respectively; Wherein, The detection result includes the result is changed and the result is empty;

隔离开关运行状态识别单元130,用于根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。The isolating switch operating state recognition unit 130 is used to determine the isolating switch corresponding to each isolating switch live moving image based on the frame sequence based on the two detection results obtained based on the same isolating switch live moving image based on the opening mosaic diagram and the closing mosaic diagram The operating state at the time; wherein, the operating state is one of the opening state, the closing state and the intermediate state.

其中,所述隔离开关状态最终识别单元130包括:Wherein, the final recognition unit 130 of the isolating switch state includes:

中间态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态;The intermediate state recognition module is used to detect the target when the trained isolating switch state detection model, if the result obtained by the detection of the opening mosaic diagram of the current isolating switch scene moving image and its closing mosaic diagram is the same, and the two If all the detection results show that the results have changed, then it is determined that the operating state of the current isolating switch scene moving image is an intermediate state;

分合闸态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;反之,待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。The opening and closing state recognition module is used for target detection when the trained isolating switch state detection model is used, if the opening mosaic diagram of the current isolating switch scene motion image is different from the result obtained by the detection of the closing mosaic diagram, And one test result is empty and the other test result is change, then after it is judged that the mosaic diagram corresponding to the detection result is empty is the opening mosaic diagram, determine the running status of the current isolating switch on-site moving image On the contrary, after it is judged that the mosaic diagram corresponding to the detection result is empty is the closing mosaic diagram, it is determined that the current operating state of the on-site moving image of the isolating switch is the closing state.

其中,还包括:Among them, also include:

中间态运动识别单元,用于在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;以及The intermediate state motion recognition unit is used to sequentially extract the on-site moving images of the isolating switches whose operating state is in the intermediate state from among the on-site moving images of the multiple isolating switches based on the frame sequence; and

将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合所述已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,所述运动状态为静止或正在运动。Splicing the extracted on-site moving image of each isolating switch in the intermediate state with the on-site moving image of the isolating switch in the next frame to obtain the corresponding front and rear frame mosaic images of the on-site moving images of the isolating switch in the intermediate state, and combining the The trained isolating switch state detection model performs target detection to obtain the detection results of all intermediate state isolating switch live moving images based on the front and rear frame mosaic images, and further according to the detection results of all intermediate state isolating switch live moving images based on the front and rear frames respectively The detection result of the mosaic diagram determines the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is static or moving.

其中,所述隔离开关状态检测模型为基于yolo v3算法的神经网络模型,其输入数据为拼接图,输出类别为有变化。Wherein, the isolating switch state detection model is a neural network model based on the yolo v3 algorithm, its input data is a mosaic graph, and its output category is changed.

实施本发明实施例,具有如下有益效果:Implementing the embodiment of the present invention has the following beneficial effects:

1、本发明中摄像头通过按帧方式将实时视频分解为基于帧序列的多个隔离开关现场运动图像,并将每一个隔离开关现场运动图像与预设的分闸模板图及合闸模板图进行拼接后,利用已训练好的隔离开关状态检测模型进行目标检测,得到任意一个隔离开关现场运动图像的两个拼接图的检测结果,且通过对比同一个隔离开关现场运动图像的两个拼接图的检测结果来快速确定隔离开关每一帧的运行状态,从而不用考虑摄像头所拍摄的视频角度是否发生变化,就可以进行隔离开关状态快速检测,从而实现基于摄像头任意角度采集的样本就可进行隔离开关状态快速检测,提高了检测精度,还减少了样本数量,降低了工作量;1. In the present invention, the camera decomposes the real-time video into a plurality of on-site moving images of isolating switches based on the frame sequence by frame, and compares the on-site moving images of each isolating switch with the preset opening template diagram and closing template diagram. After splicing, use the trained isolation switch state detection model to detect the target, and obtain the detection results of the two mosaic images of any live moving image of an isolating switch, and compare the two mosaic images of the same isolating switch live moving image. The detection results can be used to quickly determine the operating status of each frame of the isolation switch, so that the status of the isolation switch can be quickly detected without considering whether the angle of the video captured by the camera changes, so that the isolation switch can be performed based on samples collected from any angle of the camera. Rapid status detection improves detection accuracy, reduces the number of samples, and reduces workload;

2、本发明中摄像头还通过将中间态的每一个隔离开关现场运动图像均与其后一帧的运动图像进行拼接后,利用已训练好的隔离开关状态检测模型进行目标检测,快速对隔离开关在中间态的每一帧图像上的运动状态进行定位,以实现隔离开关的运动过程中的缺陷进行检测及报警。2. In the present invention, the camera also uses the trained isolating switch state detection model to perform target detection by splicing the on-site moving image of each isolating switch in the intermediate state with the moving image of the next frame, and quickly detects the isolating switch in the state. The motion state on each frame image of the intermediate state is positioned to realize the detection and alarm of defects in the motion process of the isolating switch.

值得注意的是,上述系统实施例中,所包括的各个系统单元只是按照功能逻辑进行划分的,但并不局限于上述的划分,只要能够实现相应的功能即可;另外,各功能单元的具体名称也只是为了便于相互区分,并不用于限制本发明的保护范围。It is worth noting that in the above system embodiments, the system units included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific functions of each functional unit The names are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the present invention.

本领域普通技术人员可以理解实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成,所述的程序可以存储于一计算机可读取存储介质中,所述的存储介质,如ROM/RAM、磁盘、光盘等。Those of ordinary skill in the art can understand that all or part of the steps in the method of the above-mentioned embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage Media such as ROM/RAM, magnetic disk, optical disk, etc.

以上所揭露的仅为本发明较佳实施例而已,当然不能以此来限定本发明之权利范围,因此依本发明权利要求所作的等同变化,仍属本发明所涵盖的范围。The above disclosures are only preferred embodiments of the present invention, and certainly cannot limit the scope of rights of the present invention. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope of the present invention.

Claims (9)

1.一种隔离开关状态检测方法,其特征在于,其在摄像头上实现,所述方法包括以下步骤:1. a disconnector state detection method, is characterized in that, it realizes on camera, and described method comprises the following steps: 所述摄像头采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;The camera collects the real-time video of the isolating switch in operation and intercepts the image frame by frame, so as to obtain a plurality of on-site moving images of the isolating switch based on the frame sequence; 将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;Splicing the on-site moving image of each isolating switch with the preset opening template diagram and the preset closing template diagram to obtain the corresponding opening mosaic diagram and closing mosaic diagram of all the on-site moving images of the isolating switch, and Combining the trained isolating switch state detection model for target detection, to obtain the detection results of all the on-site moving images of the isolating switches based on the opening mosaic diagram and the closing mosaic diagram; wherein, the detection results include the result being changed and the result Is empty; 根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。According to the two detection results obtained based on the opening mosaic diagram and the closing mosaic diagram of the same isolating switch on-site moving image, the operating state of the isolating switch corresponding to each isolating switch on-site moving image based on the frame sequence is determined; wherein the operation The state is one of the opening state, closing state and intermediate state. 2.如权利要求1所述的隔离开关状态检测方法,其特征在于,所述根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态的具体步骤包括:2. The isolating switch state detection method as claimed in claim 1, wherein said two detection results based on the same isolating switch on-site motion image based on the opening mosaic diagram and the closing mosaic diagram are used to determine whether the isolating switch corresponds to The specific steps of the running status of each isolating switch live moving image based on the frame sequence include: 在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态;When the trained isolating switch state detection model is used for target detection, if the results obtained by the detection of the mosaic diagram of the current isolating switch site moving image are the same as those detected by the mosaic diagram of the closing mosaic diagram, and both detection results are If there is a change, it is determined that the current operating state of the on-site moving image of the isolating switch is an intermediate state; 在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;反之,待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。When the trained isolating switch state detection model performs target detection, if the opening mosaic diagram of the current isolating switch scene moving image is different from the result detected by its closing mosaic diagram, and one of the detection results is that the result is empty And another detection result is that the result is changed, then after it is judged that the mosaic diagram corresponding to the detection result is empty is the mosaic diagram of opening, determine that the running state of the current moving image of the isolating switch is the opening state; otherwise, to be judged After the mosaic diagram corresponding to the detection result is empty is the closing mosaic diagram, it is determined that the current running state of the on-site moving image of the isolating switch is the closing state. 3.如权利要求2所述的隔离开关状态检测方法,其特征在于,所述方法进一步包括:3. The isolating switch state detection method as claimed in claim 2, wherein said method further comprises: 在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;From the live moving images of multiple isolating switches based on the frame sequence, the live moving images of the isolating switches whose operation status is in the intermediate state are sequentially extracted; 将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合所述已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,所述运动状态为静止或正在运动。Splicing the extracted on-site moving image of each isolating switch in the intermediate state with the on-site moving image of the isolating switch in the next frame to obtain the corresponding front and rear frame mosaic images of the on-site moving images of the isolating switch in the intermediate state, and combining the The trained isolating switch state detection model performs target detection to obtain the detection results of all intermediate state isolating switch live moving images based on the front and rear frame mosaic images, and further according to the detection results of all intermediate state isolating switch live moving images based on the front and rear frames respectively The detection result of the mosaic diagram determines the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is static or moving. 4.如权利要求3所述的隔离开关状态检测方法,其特征在于,所述根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态的具体步骤包括:4. the isolating switch state detection method as claimed in claim 3, is characterized in that, described according to the isolating switch spot moving image of all intermediate states respectively based on the detection result of front and back frame mosaic figure, determines the isolation of all intermediate states corresponding to isolating switch The specific steps of the motion state when switching live motion images include: 在所述已训练好的隔离开关状态检测模型进行目标检测时,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为空,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止;反之,若某一中间态的隔离开关现场运动图像各自基于前后帧拼接图所检测得到的结果为有变化,则确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为正在运动。When the trained isolating switch state detection model is used for target detection, if the detected results of the on-site motion images of the isolating switch in a certain intermediate state based on the front and rear frame mosaic images are empty, then it is determined that the isolating switch corresponds to the currently detected The motion state of the disconnector in the intermediate state is stationary; on the contrary, if the live motion images of a disconnector in the intermediate state are detected based on the detection results of the front and rear frame mosaic images, it is determined that the disconnector corresponds to the current detection The motion state of the live motion image of the isolating switch in the intermediate state is in motion. 5.如权利要求4所述的隔离开关状态检测方法,其特征在于,所述方法进一步包括:5. the isolating switch state detection method as claimed in claim 4, is characterized in that, described method further comprises: 在确定隔离开关对应当前检测的中间态的隔离开关现场运动图像时的运动状态为静止之后,进行报警。After it is determined that the motion state of the isolating switch corresponding to the live moving image of the isolating switch in the currently detected intermediate state is static, an alarm is given. 6.一种摄像头,其特征在于,包括:6. A camera, characterized in that, comprising: 视频采集及图像截取单元,用于采集隔离开关运行时的实时视频并按帧进行图像截取,得到基于帧序列的多个隔离开关现场运动图像;The video acquisition and image interception unit is used to collect the real-time video of the isolation switch during operation and perform image interception by frame to obtain live moving images of multiple isolation switches based on the frame sequence; 隔离开关运行检测单元,用于将每一个隔离开关现场运动图像均与预设的分闸模板图及预设的合闸模板图进行拼接,以得到所有隔离开关现场运动图像各自对应的分闸拼接图及合闸拼接图,并结合已训练好的隔离开关状态检测模型进行目标检测,以得到所有隔离开关现场运动图像各自基于分闸拼接图及合闸拼接图的检测结果;其中,所述检测结果包括结果为有变化和结果为空;The isolating switch operation detection unit is used to splice the on-site moving image of each isolating switch with the preset opening template map and the preset closing template map to obtain the corresponding opening splicing of all the isolating switch on-site moving images diagram and closing mosaic diagram, and combined with the trained isolation switch state detection model to carry out target detection, to obtain the detection results based on the opening mosaic diagram and closing mosaic diagram of all isolation switch scene motion images respectively; wherein, the detection Result includes result is changed and result is empty; 隔离开关运行状态识别单元,用于根据同一个隔离开关现场运动图像基于分闸拼接图及合闸拼接图所得的两个检测结果,确定隔离开关对应基于帧序列的每一个隔离开关现场运动图像时的运行状态;其中,所述运行状态为分闸状态、合闸状态及中间态之其中一种。The operating state recognition unit of the isolating switch is used to determine when the isolating switch corresponds to each live moving image of the isolating switch based on the frame sequence based on the two detection results obtained from the on-site moving image of the same isolating switch based on the opening mosaic diagram and the closing mosaic diagram. The operating state; wherein, the operating state is one of the opening state, the closing state and the intermediate state. 7.如权利要求6所述的摄像头,其特征在于,所述隔离开关状态最终识别单元包括:7. The camera according to claim 6, wherein the final recognition unit of the isolating switch state comprises: 中间态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果相同,且两个检测结果均是结果为有变化,则确定当前隔离开关现场运动图像的运行状态为中间态;The intermediate state recognition module is used to detect the target when the trained isolating switch state detection model, if the result obtained by the detection of the opening mosaic diagram of the current isolating switch scene moving image and its closing mosaic diagram is the same, and the two If all the detection results show that the results have changed, then it is determined that the operating state of the current isolating switch scene moving image is an intermediate state; 分合闸态识别模块,用于在所述已训练好的隔离开关状态检测模型进行目标检测时,若当前隔离开关现场运动图像的分闸拼接图与其合闸拼接图所检测得到的结果不同,且有一个检测结果是结果为空及另一个检测结果是结果为有变化,则待判定出结果为空所对应检测的拼接图为分闸拼接图后,确定当前隔离开关现场运动图像的运行状态为分闸状态;反之,待判定出结果为空所对应检测的拼接图为合闸拼接图后,确定当前隔离开关现场运动图像的运行状态为合闸状态。The opening and closing state recognition module is used for target detection when the trained isolating switch state detection model is used, if the opening mosaic diagram of the current isolating switch scene motion image is different from the result obtained by the detection of the closing mosaic diagram, And one test result is empty and the other test result is change, then after it is judged that the mosaic diagram corresponding to the detection result is empty is the opening mosaic diagram, determine the running status of the current isolating switch on-site moving image On the contrary, after it is judged that the mosaic diagram corresponding to the detection result is empty is the closing mosaic diagram, it is determined that the current operating state of the on-site moving image of the isolating switch is the closing state. 8.如权利要求7所述的摄像头,其特征在于,还包括:8. The camera according to claim 7, further comprising: 中间态运动识别单元,用于在基于帧序列的多个隔离开关现场运动图像中,依序提取出运行状态为中间态的隔离开关现场运动图像;以及The intermediate state motion recognition unit is used to sequentially extract the on-site moving images of the isolating switches whose operating state is in the intermediate state from among the on-site moving images of the multiple isolating switches based on the frame sequence; and 将提取的中间态的每一个隔离开关现场运动图像均与其后一帧的隔离开关现场运动图像进行拼接,以得到所有中间态的隔离开关现场运动图像各自对应的前后帧拼接图,并结合所述已训练好的隔离开关状态检测模型进行目标检测,以得到所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,且进一步根据所有中间态的隔离开关现场运动图像各自基于前后帧拼接图的检测结果,确定隔离开关对应所有中间态的隔离开关现场运动图像时的运动状态;其中,所述运动状态为静止或正在运动。Splicing the extracted on-site moving image of each isolating switch in the intermediate state with the on-site moving image of the isolating switch in the next frame to obtain the corresponding front and rear frame mosaic images of the on-site moving images of the isolating switch in the intermediate state, and combining the The trained isolating switch state detection model performs target detection to obtain the detection results of all intermediate state isolating switch live moving images based on the front and rear frame mosaic images, and further according to the detection results of all intermediate state isolating switch live moving images based on the front and rear frames respectively The detection result of the mosaic diagram determines the motion state of the isolating switch corresponding to all the live motion images of the isolating switch in the intermediate state; wherein, the motion state is static or moving. 9.如权利要求8所述的摄像头,其特征在于,所述隔离开关状态检测模型为基于yolov3算法的神经网络模型,其输入数据为拼接图,输出类别为有变化。9. The camera according to claim 8, wherein the isolating switch state detection model is a neural network model based on the yolov3 algorithm, its input data is a mosaic image, and its output category is changed.
CN202211225772.6A 2022-10-09 2022-10-09 A method for detecting the state of an isolating switch and a camera Pending CN115526871A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211225772.6A CN115526871A (en) 2022-10-09 2022-10-09 A method for detecting the state of an isolating switch and a camera

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211225772.6A CN115526871A (en) 2022-10-09 2022-10-09 A method for detecting the state of an isolating switch and a camera

Publications (1)

Publication Number Publication Date
CN115526871A true CN115526871A (en) 2022-12-27

Family

ID=84701974

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211225772.6A Pending CN115526871A (en) 2022-10-09 2022-10-09 A method for detecting the state of an isolating switch and a camera

Country Status (1)

Country Link
CN (1) CN115526871A (en)

Similar Documents

Publication Publication Date Title
CN109271872B (en) Device and method for judging on-off state and diagnosing fault of high-voltage isolating switch
CN110008877B (en) A Detection and Recognition Method of Substation Isolation Switch Based on Faster RCNN
CN109298785A (en) A man-machine joint control system and method for monitoring equipment
CN109409315B (en) Method and system for detecting remnants in panel area of ATM (automatic Teller machine)
CN101893686A (en) Device and method for online detection of circuit breaker action characteristics based on photographic digitization
Liu et al. Fault diagnosis method of substation equipment based on You Only Look Once algorithm and infrared imaging
CN108776452B (en) Special equipment field maintenance monitoring method and system
CN112083000A (en) Intelligent identification method and system for appearance defects of substation equipment
CN106339722A (en) Line knife switch state monitoring method and device
CN112710951A (en) GIS isolator state determination system
CN118552894A (en) Substation indoor state identification method and device
CN112906593A (en) Sluice image identification method based on fast RCNN
CN118887490A (en) A method, device, equipment and storage medium for identifying hidden dangers of distribution box
Wang et al. Visual defect detection for substation equipment based on joint inspection data of camera and robot
CN111428987A (en) Artificial intelligence-based image identification method and system for relay protection device
CN115376074A (en) Method and system for evaluating recognition effect of power transmission line monitoring device
Chen et al. Research on the state detection of the secondary panel of the switchgear based on the YOLOv5 network model
CN115526871A (en) A method for detecting the state of an isolating switch and a camera
CN115690659A (en) Substation safety detection method and device under occlusion conditions
CN117879155A (en) Auxiliary decision making system based on information acquisition of intelligent recorder
Yerunkar et al. Analytical review of machine learning techniques for transmission line insulator fault detection and classification
CN113850344B (en) A patrol robot equipment identification method and system
CN114549822A (en) Disconnecting switch closing state detection method based on preset position camera
CN114821309A (en) Indoor transformer substation switch and indicator lamp state identification method and system
CN116385739A (en) Switch state identification method, device, equipment and medium based on detection model

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