CN115063390A - Scissor type knife switch closing in-place recognition device and judgment method - Google Patents
Scissor type knife switch closing in-place recognition device and judgment method Download PDFInfo
- Publication number
- CN115063390A CN115063390A CN202210780834.3A CN202210780834A CN115063390A CN 115063390 A CN115063390 A CN 115063390A CN 202210780834 A CN202210780834 A CN 202210780834A CN 115063390 A CN115063390 A CN 115063390A
- Authority
- CN
- China
- Prior art keywords
- knife switch
- type knife
- scissor
- switch
- scissor type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a scissor-type knife switch in-place switching identification device and a scissor-type knife switch in-place switching identification method, which are used for identifying whether a scissor-type knife switch of a transformer substation is accurately switched in place or not. The device includes: the device comprises a camera, a scissor type switch identification front-end device, a scissor type switch identification background learning training device and a data and control bus. Aiming at the problem of serious accidents such as heating and explosion caused by the fact that the scissor type knife switch is not in place due to the fact that the scissor type knife switch is abnormal in structure caused by long-term use, a cloud-edge fusion framework is adopted, semantic segmentation is conducted by utilizing a U-net network to achieve target identification, then a Hough transform algorithm is used for searching straight line segments from a knife switch double-arm area to conduct edge line segment extraction, and a method for detecting that the scissor type knife switch is not in place is provided on the basis, so that accurate identification of the scissor type knife switch state is achieved, manual inspection can be replaced, manpower is saved, safety is improved, and further the method can be used for monitoring other equipment of a transformer substation.
Description
Technical Field
The invention relates to a detection technology of a state of an electric knife switch, in particular to accurate detection of the state of a scissor-type knife switch, and specifically relates to a device and a method for identifying the switching-on position of the scissor-type knife switch.
Background
The invention relates to a knife switch, also called a high-voltage isolating switch, which is a main device of a transformer substation.
In long-term operation, the scissor-type knife switch may not be closed in place for various reasons. The existing switching-on in-place judgment mainly depends on auxiliary contacts of a scissor-type disconnecting link, the scissor-type disconnecting link basically works outdoors and has to face the problems of sun and rain, metal corrosion damage and abrasion, the auxiliary contacts or a transmission part thereof are abnormal to cause the misjudgment of the switching-on and switching-off positions, wrong signals are uploaded, the safety and the life cycle of equipment are threatened, even large electric power accidents are caused, and the serious casualties and property loss are brought.
The traditional method adopts a polling person to judge the fault of the improper switching-on by naked eyes. Through sending the personnel of patrolling and examining to the scene with the naked eye observation, nevertheless because the transformer substation distributes in more in the more remote place, the quantity of every transformer substation's scissors formula switch is also more, and it is not only time-consuming but also hard to arrive the on-the-spot observation. In addition, as society develops, the number of the transformer substations and the number of the scissor type disconnecting links used in the transformer substations increase continuously, and more human resources are consumed compared with the prior art in order to detect the closing conditions of a plurality of scissor type disconnecting links at the same time. The manual double detection method consumes a large amount of manpower, is poor in real-time performance, is a short plate for realizing full automation of a transformer substation, has certain danger, and can possibly generate discharge sparks and even explosion when an inspection worker needs to observe the short plate in a close range and a high-voltage isolating switch is heated when the switch is not closed in place.
The development of a double-check method for automatically identifying the fault that the switch-on is not in place is necessary, and related notifications are issued recently to national power grids and southern power grids. The existing method for identifying the opening and closing state is only simple binary logic judgment, the feature that the opening is not in place is not obvious, the difficulty of identifying the opening and closing in place is much higher than that of identifying the opening and closing, an accurate identification device and method are needed, and the existing image acquisition or video monitoring method can only judge whether the opening and closing is in place or not and cannot accurately judge whether the opening and closing is in place or not. Other various methods are researched, such as an infrared method, an attitude sensor and the like, the heating abnormity of the scissor-type knife switch can be judged only by electrifying for a period of time after the infrared method is switched on, and a certain economic loss is caused if the power is cut off, so that the normal operation of a power grid is influenced, and the method belongs to a method for repairing the power grid afterwards. Attitude sensor needs in addition to install equipment on scissors formula switch, can install when the electric wire netting cuts off the power supply, and its maintenance is also in the outage of electric wire netting, influences the electric wire netting operation. These methods are not expected to be effective, and development of better double check methods is urgently needed.
The invention researches the identification of the switching-on in-place state of the scissor type disconnecting link.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a scissor-type knife switch closing in-place recognition device and a scissor-type knife switch closing in-place recognition method, so as to solve the problems.
The invention provides the following technical scheme:
a scissors formula knife switch combined floodgate recognition device that targets in place, it includes:
the system comprises a camera, a scissor type switch identification front-end device, a scissor type switch identification background learning and training device and a data and control bus;
the cameras are used for shooting images of the scissor type disconnecting links, the images are output to the scissor type disconnecting link identification front-end device through a data and control bus, 3 cameras are used for shooting a three-phase scissor type disconnecting link, and each camera is used for shooting a single-phase scissor type disconnecting link;
the scissors type knife switch identification front-end device sends a control signal to control the shooting action of the camera and can also control the holder of the camera to pitch and rotate so as to adjust the shooting direction, the scissors type knife switch identification front-end device identifies the scissors type knife switch image and detects the closing angles of the two scissors arms, judges whether the closing is in place or not, sends a signal indicating whether the closing is in place or not to the scissors type knife switch identification background learning training device, and simultaneously forwards the image of the scissors type knife switch to the scissors type knife switch identification background learning training device, and the identification parameters of the scissors type knife switch identification front-end device are provided and refreshed by the scissors type knife switch identification background learning training device;
the scissor type switch identification background learning training device carries out deep learning training according to the received scissor type switch image, the well-learned parameters are sent to the scissor type switch identification front-end device, meanwhile, a signal of whether the switch is in place or not is forwarded to a dispatching center, and a plurality of scissor type switch identification front-end devices can share one scissor type switch identification background learning training device;
the data and control bus is used for transmitting data and control signals among the camera, the scissors type switch identification front-end device and the scissors type switch identification background learning and training device;
the scissors type knife switch is a single-phase scissors type knife switch, and one three-phase scissors type knife switch comprises three linked single-phase scissors type knife switches.
Preferably, the arrangement of the cameras is as follows: bases of 3 left, middle and right cameras are arranged in a straight line in parallel, the left and right cameras with fixed shooting angles are arranged in the middle, the camera with a cloud platform is arranged in the middle and is used for shooting each phase of scissors type knife switch of a three-phase scissors type knife switch, and the center of each camera is aligned with the intersection point of two scissors arms of one scissors type knife switch; besides shooting the middle disconnecting link, the middle camera can be used for shooting other electric equipment.
In order to overcome the defects in the prior art, a second object of the present invention is to provide a method for determining a closing state of a scissor-type knife switch closing in-place recognition device, comprising:
acquiring and preprocessing a scissor type knife switch image;
scissors type knife switch target identification: finding out a needed scissor type knife switch in an image containing background information, filtering out the background, and extracting the scissor type knife switch from a shot picture;
extracting edge line segments: extracting the contour lines of the two scissor arms of the scissor type knife switch in a segmented manner so as to accurately detect the closing state by adopting a geometric method;
detecting a closing included angle: and calculating an included angle of contour line segments of two scissor arms of the scissor type knife switch, if the error of the included angle and the correct switching-on included angle exceeds a given threshold value, sending a switching-on-not-in-place signal, otherwise, sending a switching-on-success signal, if the switching-on of any one scissor type knife switch of the three-phase scissor type knife switch is not in place, and then, the whole three-phase scissor type knife switch is classified as the switching-on-not-in-place.
Preferably, the scissor-type knife gate image acquisition and preprocessing comprises:
data acquisition: sending a shooting instruction, and reading a scissor type knife switch image shot by a camera;
data clipping: different camera resolutions are different, for adapting to different cameras, reduce the unnecessary amount of computation simultaneously, obtain the picture back from the camera, and the unified picture of cutting out into fixed resolution.
Preferably, the scissors type knife switch target recognition is realized by performing semantic segmentation on the image by adopting a U-net network.
Preferably, the edge line segment extraction includes:
image smoothing: smoothing the sawtooth of the knife switch image by a median filtering method;
and searching straight line segments from the double-arm area of the disconnecting link by adopting Hough algorithm (Hough) to extract edge line segments.
Preferably, the detection of the closing included angle includes:
taking pixel points: B. d is the bottom end point of the contour line segment at the lower part of the left and right scissor arms and the pixel coordinate (x) b ,y b )、(x d ,y d ) A, C represents the top point (x) of the lower contour line of the left and right scissor arms a ,y a )、(x c ,y c ) (ii) a Calculating a closing included angle: the included angle of the inner contour line segment at the lower part of the crossed arm of the scissors is set as alpha, which is calculated by the following formula,
and (3) error calculation: if the correct angle when the scissors type knife switch is switched on in place is beta, the error of the switching-on angle is as follows: and alpha-beta, if the error is smaller than a given threshold value, judging that the scissor type knife switch is completely closed, otherwise, judging that the scissor type knife switch is not closed in place.
Preferably, the cloud deck control method of the scissor-type knife switch closing in-place recognition device adopts a fuzzy control algorithm, and comprises the following steps:
fuzzy control is carried out by recursion regulation rules according to a formula method;
carrying out local continuity on a formula method;
and optimizing a locally continuous formula method by adopting a global optimization method of spherical gap migration.
Compared with the prior art, the scissor type knife switch closing in-place recognition device and the determination method provided by the invention have the following advantages and effects:
1. whether the switching-on of the scissor type knife switch is in place can be accurately judged;
2. the safety is good, and non-contact optical video detection is realized;
3. the installation and maintenance are not required to be carried out in a power failure mode, and the normal operation of a power system is not influenced;
4. the multifunctional monitoring device has multiple purposes, and can monitor other electric equipment besides the state of the disconnecting link.
Drawings
FIG. 1 is a schematic diagram of a module structure and a principle of a scissor-type knife switch closing in-place recognition device according to an embodiment of the invention
Fig. 2 is a schematic flow chart of a switching-on state determination method according to an embodiment of the present invention;
FIG. 3 is an image of a scissor-type knife switch according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the U-net network implementing target recognition by semantic segmentation according to the embodiment of the present invention
FIG. 5 is a schematic diagram of the scissors-type knife gate after the contour line smoothing process according to the embodiment of the present invention;
fig. 6 is a schematic diagram of the scissor-type knife gate edge line segment extraction process according to the embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
Examples
Referring to fig. 1 to 6, a scissor type knife switch closing in-place recognition device provided by an embodiment of the present invention includes: the system comprises a camera, a scissor type switch identification front-end device, a scissor type switch identification background learning and training device, a data and control bus and a router;
assuming that a certain transformer substation has n scissor type disconnecting links, 3 cameras on the left, middle and right sides for shooting the scissor type disconnecting link 1 and 3 cameras on the left, middle and right sides for shooting the scissor type disconnecting link n are drawn in fig. 1, and a holder head is arranged in the middle of the cameras and can be used for pitching and rotating for shooting. For simplicity, fig. 1 shows in dashed lines that there are other scissor-type knife switches. The bases of the left, middle and right 3 cameras are mounted side by side in a straight line to photograph each phase of a three-phase scissor-type switch, the center of each camera is aligned with the connecting shafts of the two telescopic arms of a scissor-type switch of each phase, the image of a scissor-type switch of a certain phase of a three-phase scissor-type switch is shown in fig. 3, and the three-phase scissor-type switch has two scissor-shaped arms, the centers of which are cross points, and three scissor-type switches.
In fig. 1, a three-phase scissor-type switch is equipped with a scissor-type switch identification front-end device, when the scissor-type switch identification front-end device sends out a shooting signal, scissor-type switch images shot by 3 cameras on the left, the middle and the right are output to the scissor-type switch identification front-end device through a data and control bus; the scissor type disconnecting link identification front-end device can also control the holder of the middle camera to pitch and rotate so as to adjust the shooting direction.
The scissor type switch identification front-end device identifies a scissor type switch image by adopting a deep learning algorithm, calculates an included angle of a scissor cross arm outline line segment, judges whether switching-on is in place or not, sends a signal indicating whether switching-on is in place or not to the scissor type switch identification background learning training device, and simultaneously forwards the scissor type switch image to the scissor type switch identification background learning training device, wherein deep learning network parameters of the scissor type switch identification front-end device are provided and refreshed by the scissor type switch identification background learning training device; in fig. 1, n three-phase scissor type switches are arranged, but only one scissor type switch identification background learning and training device is needed, because the deep learning network parameters of the scissor type switch identification front-end devices are the same, further, a plurality of transformer substations can share one scissor type switch identification background learning and training device, so that the scissor type switch identification front-end device in fig. 1 is connected to a data and control bus through a router, and the cloud edge fusion framework is advanced at present.
The scissors type knife switch identification background learning training device carries out deep learning training according to the received images of the scissors type knife switch, the parameters which are well learned and trained are sent to the scissors type knife switch identification front-end device, and meanwhile, a signal indicating whether the switch-on is in place or not is forwarded to a dispatching center.
The data and control bus in fig. 1 can adopt a universal wired ethernet or a wireless wifi, the cameras are all network cameras, the scissors type knife switch identification front-end device and the scissors type knife switch identification background learning and training device are all provided with network interfaces, and any device can transmit data and control signals through a network. The dispatching center can also directly acquire the scissor type knife switch image to monitor the scissor type knife switch without the scissor type knife switch identification front end device.
FIG. 2 is a schematic flow chart of a method for determining the closing status of the scissor-type knife switch closing in-place recognition device,
the switching-on state judgment method adopting the scissor type knife switch switching-on in-place recognition device provided by the embodiment of the invention specifically comprises the following steps:
s1, the scissor type disconnecting link identification front-end device sends shooting instructions to the left camera, the middle camera and the right camera, the middle camera returns to shoot a certain closed three-phase scissor type disconnecting link together, and the cameras send images of the 3 scissor type single-phase disconnecting links to the scissor type disconnecting link identification front-end device through data and control buses;
s2, data clipping: different cameras have different resolutions, and in order to adapt to different cameras and reduce unnecessary calculation amount, pictures are uniformly cut into pictures with fixed resolutions after the pictures are obtained from the cameras;
s3, scissors type knife switch target identification: adopting a U-net network to carry out semantic segmentation on the image to realize that a required scissor type knife switch is found in an image containing background information, filtering the background, and extracting the scissor type knife switch from a shot image; as shown in fig. 3; the development of deep learning greatly accelerates the pace of the computer vision field and is applied to various tasks in the field; the task of image recognition is to solve what problem, that is, what the landing point is in the image, the task of target detection is to solve where problem, that is, the landing point is the position of the target in the image, and semantic segmentation is to answer the above two problems from the pixel level, and identify the content and the position existing in the image by searching all pixels belonging to the target;
s4, image smoothing: smoothing the sawtooth of the knife switch image by a median filtering method; as shown in fig. 4;
s5, edge line segment extraction: and searching straight line segments from the double-arm area of the disconnecting link by adopting Hough algorithm (Hough) to extract edge line segments. Contour lines of two scissor arms of the scissor type knife switch are extracted in a segmented mode so that the closing state can be accurately detected by adopting a geometric method; as shown in fig. 5;
s6, taking pixel points: B. d is the bottom end point of the contour line segment at the lower part of the left and right scissor arms and the pixel coordinate (x) b ,y b )、(x d ,y d ) A, C is the top point (x) of the lower contour line of the left and right scissor arms a ,y a )、(x c ,y c ) (ii) a As shown in fig. 6;
s7, calculating a closing included angle: the included angle of the inner contour line segment at the lower part of the crossed arm of the scissors is set as alpha, which is calculated by the following formula,
s8, error calculation: if the correct angle when the scissors type knife switch is switched on in place is beta, the error of the switching-on angle is as follows: and alpha-beta, if the error is smaller than a given threshold value, judging that the scissor type knife switch is completely closed, otherwise, judging that the scissor type knife switch is not closed in place.
And S9, if any one of the three-phase scissor type knife switches is not switched on in place, the whole three-phase scissor type knife switch is switched on in place.
And S10, collecting switching-on judgment signals of all scissor type knife switch recognition front-end devices by the scissor type knife switch recognition background learning training device, presenting the switching-on judgment signals through an interface, and forwarding the switching-on judgment signals to a dispatching center.
And S11, ending.
The above flow is an extension of fig. 2, and a module of fig. 2 actually includes a plurality of related steps.
The invention provides a cloud deck control method of a scissor type knife switch closing in-place recognition device, which adopts a fuzzy control algorithm and comprises the following steps:
fuzzy control is carried out by recursion regulation rules according to a formula method;
carrying out local continuity on a formula method;
and optimizing a locally continuous formula method by adopting a global optimization method of spherical gap migration.
The camera can adopt cloud platform integration spherical camera in two in figure 1, and the benefit of this scheme is as follows:
1) because the gun-shaped camera is not provided with a pan-tilt and a control mechanism, the price of the gun-shaped camera is far lower than that of a dome camera when the optical multiples are the same;
2) the gun-shaped camera is fixed in position and angle, does not move or rotate, and is high in identification precision and small in error;
3) because the ball-type camera will rotate and luffing motion, its positioning accuracy is not high relatively, but can 360 degrees control surrounding area, except shooting the disconnecting link, can also shoot all other equipment of transformer substations such as generating lines, through ball-type camera of installation in every intermediate position, can reach the purpose of monitoring other equipment of transformer substation with less ball-type camera, therefore can partially replace the manual work to patrol and examine the robot, the camera shelters from or takes place other troubles by bird droppings simultaneously about, middle ball-type camera can also replace temporarily.
The scheme not only realizes a plurality of functions, but also saves cost. The key point is the control algorithm of the dome camera, the dome camera is a combination device of the camera and the pan-tilt, and the dome camera can horizontally rotate and vertically tilt, so that the dome camera can shoot a large range of surrounding scenery, the visual field of the dome camera is far larger than that of a gun type camera with a fixed angle, but the dome camera also has 2 defects:
the rotation angle of the device is regulated in a step-by-step manner, and stepless positioning cannot be realized, so that the device is difficult to accurately position to a certain angle; the cradle head is mechanically driven by a motor gear, has a gap, usually has a certain virtual position error, and can accurately return to the position of the shooting disconnecting link after shooting surrounding equipment.
However, the identification of the closing state of the knife switch must reach a high-precision positioning level, otherwise, it is difficult to accurately identify and judge whether the closing is in place, and for this reason, the present invention is implemented by using a local continuous surface high-precision fuzzy control algorithm, see reference 1.
In order to realize accurate and rapid control of the dome camera, according to the mainstream automatic control theory at present, a differential equation of the dome camera needs to be written out, and a relevant mathematical model is established. The local continuous surface high-precision fuzzy control algorithm does not need to establish a model of a controlled object, carries out local control surface continuity on a table look-up method and a formula method, keeps the characteristic of high speed, has precision even exceeding a rule reasoning method, and is not complex.
The local continuous surface high-precision fuzzy control algorithm has a plurality of parameters, which greatly affect the control precision and the return speed, and the invention adopts the ball gap migration algorithm to optimize the control parameters, which is specifically referred to in document 2.
Reference documents:
1) hujinsong, Zheng Qilun, Wujie, time-varying correction factor two-stage self-optimizing fuzzy controller, journal of Automation in 2002, pp.1006-1011.
2) Hujinsong, Zhengenhui, global optimization is realized by a sphere-gap migration algorithm, reported by a computer, volume 35, No. 2, month 2012, pp.193-201.
Aiming at the problem that serious accidents such as heating and explosion are caused due to the fact that the scissor type knife switch is not in place due to the fact that the scissor type knife switch is abnormal in structure after being used for a long time, the embodiment of the invention adopts a cloud-edge fusion framework, utilizes a U-net network to carry out semantic segmentation to realize target identification, then uses a Hough transform algorithm to search straight-line segments from a knife switch double-arm area to carry out edge line segment extraction, and provides a method for detecting that the scissor type knife switch is not in place on the basis, so that the scissor type knife switch state is accurately identified, manual inspection can be replaced, manpower is saved, safety is improved, and furthermore, the scissor type knife switch can be used for monitoring other equipment of a transformer substation.
The above description is only exemplary of the present invention and should not be taken as limiting the invention, and any modifications, equivalents, improvements and the like that are made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (8)
1. The utility model provides a scissors formula switch combined floodgate recognition device that targets in place which characterized in that, it includes:
the system comprises a camera, a scissor type switch identification front-end device, a scissor type switch identification background learning and training device and a data and control bus;
the cameras are used for shooting images of the scissor type disconnecting links, the images are output to the scissor type disconnecting link identification front-end device through a data and control bus, 3 cameras are used for shooting a three-phase scissor type disconnecting link, and each camera is used for shooting a single-phase scissor type disconnecting link;
the scissors type knife switch identification front-end device sends a control signal to control the shooting action of the camera and can also control the holder of the camera to pitch and rotate so as to adjust the shooting direction, the scissors type knife switch identification front-end device identifies the scissors type knife switch image and detects the closing angles of the two scissors arms, judges whether the closing is in place or not, sends a signal indicating whether the closing is in place or not to the scissors type knife switch identification background learning training device, and simultaneously forwards the image of the scissors type knife switch to the scissors type knife switch identification background learning training device, and the identification parameters of the scissors type knife switch identification front-end device are provided and refreshed by the scissors type knife switch identification background learning training device;
the scissor type switch identification background learning training device carries out deep learning training according to the received scissor type switch image, the well-learned parameters are sent to the scissor type switch identification front-end device, meanwhile, a signal of whether the switch is in place or not is forwarded to a dispatching center, and a plurality of scissor type switch identification front-end devices can share one scissor type switch identification background learning training device;
the data and control bus is used for transmitting data and control signals among the camera, the scissors type switch identification front-end device and the scissors type switch identification background learning and training device;
the scissors type knife switch is a single-phase scissors type knife switch, and one three-phase scissors type knife switch comprises three linked single-phase scissors type knife switches.
2. The scissor type knife switch closing in-place recognition device of claim 1, wherein the arrangement of the cameras is as follows:
bases of the left camera, the middle camera and the right camera are arranged in a straight line in parallel, the left camera and the right camera are cameras with fixed shooting angles, the middle camera is a camera with a cloud platform and is used for shooting each phase of scissors type knife switch of a three-phase scissors type knife switch, and the center of each camera is aligned with the intersection point of two scissors arms of one scissors type knife switch; besides shooting the middle disconnecting link, the middle camera can be used for shooting other electric equipment.
3. A switching-on state judgment method using the scissor type knife switch switching-on in-place recognition device of claim 1 or 2, characterized by comprising the following steps:
acquiring and preprocessing a scissor type knife switch image;
scissors type knife switch target identification: finding out a needed scissor type knife switch in an image containing background information, filtering out the background, and extracting the scissor type knife switch from a shot picture;
extracting edge line segments: contour lines of two scissor arms of the scissor type knife switch are extracted in a segmented mode so that the closing state can be accurately detected by adopting a geometric method;
detecting a closing included angle: and calculating an included angle of contour line segments of two scissor arms of the scissor type knife switch, if the error of the included angle and the correct closing included angle exceeds a given threshold value, sending a signal that the closing is not in place, otherwise, sending a signal that the closing is successful, if any one scissor type knife switch of the three-phase scissor type knife switch is not in place, and then the whole three-phase scissor type knife switch is classified as the closing is not in place.
4. The closing state determination method according to claim 3, wherein the scissors-type knife switch image acquisition and preprocessing comprises the following steps:
data acquisition: sending a shooting instruction, and reading a scissor type knife switch image shot by a camera;
data clipping: different camera resolutions are different, for adapting to different cameras, reduce the unnecessary amount of computation simultaneously, obtain the picture back from the camera, and the unified picture of cutting out into fixed resolution.
5. The method for determining the closing state according to claim 3, wherein the scissors-type knife switch target recognition is implemented by performing semantic segmentation on an image by using a U-net network.
6. The closing state judgment method according to claim 3, wherein the edge line segment extraction comprises the following steps:
image smoothing: smoothing the sawtooth of the knife switch image by a median filtering method;
and searching straight line segments from the double-arm area of the disconnecting link by adopting Hough algorithm (Hough) to extract edge line segments.
7. The method for judging the closing state according to claim 3, wherein the detection of the included closing angle comprises the following steps:
taking pixel points: B. d is the bottom end point of the contour line segment at the lower part of the left and right scissor arms and the pixel coordinate (x) b ,y b )、(x d ,y d ) A, C is the top point (x) of the lower contour line of the left and right scissor arms a ,y a )、(x c ,y c )；
Calculating a closing included angle: the included angle of the inner contour line segment at the lower part of the crossed arm of the scissors is set as alpha, which is calculated by the following formula,
and (3) error calculation: if the correct angle when the scissors type knife switch is switched on in place is beta, the error of the switching-on angle is as follows: and alpha-beta, if the error is smaller than a given threshold value, judging that the scissor type knife switch is completely closed, otherwise, judging that the scissor type knife switch is not closed in place.
8. The method for judging the closing state according to claim 3, wherein the cradle head control method of the scissor type knife switch closing in-place recognition device adopts a fuzzy control algorithm and comprises the following steps:
fuzzy control is carried out by recursion regulation rules according to a formula method;
carrying out local continuity on a formula method;
and optimizing a locally continuous formula method by adopting a global optimization method of spherical gap migration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210780834.3A CN115063390A (en) | 2022-07-04 | 2022-07-04 | Scissor type knife switch closing in-place recognition device and judgment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210780834.3A CN115063390A (en) | 2022-07-04 | 2022-07-04 | Scissor type knife switch closing in-place recognition device and judgment method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115063390A true CN115063390A (en) | 2022-09-16 |
Family
ID=83204534
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210780834.3A Pending CN115063390A (en) | 2022-07-04 | 2022-07-04 | Scissor type knife switch closing in-place recognition device and judgment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115063390A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117351499A (en) * | 2023-12-04 | 2024-01-05 | 深圳市铁越电气有限公司 | Split-combined indication state identification method, system, computer equipment and medium |
-
2022
- 2022-07-04 CN CN202210780834.3A patent/CN115063390A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117351499A (en) * | 2023-12-04 | 2024-01-05 | 深圳市铁越电气有限公司 | Split-combined indication state identification method, system, computer equipment and medium |
| CN117351499B (en) * | 2023-12-04 | 2024-02-02 | 深圳市铁越电气有限公司 | Split-combined indication state identification method, system, computer equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111311597B (en) | Unmanned aerial vehicle inspection method and system for defective insulator | |
| CN102722166B (en) | Intelligent vision detection system and state detection method of transformer substation device | |
| CN206023052U (en) | A kind of HV Transmission Line Routing Inspection robot | |
| CN108957240A (en) | Electric network fault is remotely located method and system | |
| CN108489425B (en) | All-weather automatic video monitoring system and method for opening and closing states of isolating switch | |
| CN110889383A (en) | On-line monitoring system for state of isolating switch based on AI technology | |
| CN102609730B (en) | Method based on contact proportion of dynamic contact and static contact and used for distinguishing switching-on and switching-off reliability of disconnecting link | |
| CN115063390A (en) | Scissor type knife switch closing in-place recognition device and judgment method | |
| CN110008937B (en) | Switch cabinet running state management monitoring system and method and computing equipment | |
| CN109494617B (en) | Obstacle clearing inspection robot and obstacle clearing method thereof | |
| CN113569914A (en) | Power transmission line inspection method and system fusing point cloud data | |
| CN115761427A (en) | Device and method for accurately identifying state of high-voltage isolating switch | |
| CN114879679A (en) | Transformer substation inspection method and system | |
| CN113900436B (en) | Inspection control method, inspection control device, inspection control equipment and storage medium | |
| CN112543272B (en) | Transformer substation inspection camera device with light regulation function and method | |
| CN114973108A (en) | Device for recognizing closing in place of vertical telescopic disconnecting link and method for judging closing state | |
| CN111426428A (en) | Transformer substation indoor water leakage detection system and method based on image acquisition | |
| CN113533947A (en) | On-line monitoring system for state of isolating switch based on AI technology | |
| CN115063392A (en) | Horizontal rotary type knife switch closing in-place recognition device and closing state judgment method | |
| CN114697569A (en) | Information picture scaling method, terminal and system applied to power information system | |
| CN114841932A (en) | Foreign matter detection method, system, equipment and medium for photovoltaic panel of photovoltaic power station | |
| CN115167412A (en) | Inspection robot and inspection robot-based power distribution room equipment control method | |
| CN114330477A (en) | Power equipment defect detection system and method based on mixed reality equipment | |
| CN114666550A (en) | Contact point parameter detection device and method of close contact checker based on image processing | |
| CN115063391A (en) | Accurate identification device for state of horizontal rotary type disconnecting link and switching-on state judgment method |
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 |