CN113705372A - AI identification system for join in marriage net job site violating regulations - Google Patents
AI identification system for join in marriage net job site violating regulations Download PDFInfo
- Publication number
- CN113705372A CN113705372A CN202110914549.1A CN202110914549A CN113705372A CN 113705372 A CN113705372 A CN 113705372A CN 202110914549 A CN202110914549 A CN 202110914549A CN 113705372 A CN113705372 A CN 113705372A
- Authority
- CN
- China
- Prior art keywords
- module
- image
- area
- noise reduction
- edge computing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000005540 biological transmission Effects 0.000 claims abstract description 66
- 238000009826 distribution Methods 0.000 claims abstract description 35
- 238000004364 calculation method Methods 0.000 claims abstract description 34
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 230000009467 reduction Effects 0.000 claims description 64
- 230000002159 abnormal effect Effects 0.000 claims description 54
- 238000007726 management method Methods 0.000 claims description 41
- 206010000117 Abnormal behaviour Diseases 0.000 claims description 30
- 238000012545 processing Methods 0.000 claims description 22
- 238000010191 image analysis Methods 0.000 claims description 19
- 230000009471 action Effects 0.000 claims description 13
- 230000005856 abnormality Effects 0.000 claims description 8
- 230000015572 biosynthetic process Effects 0.000 claims description 8
- 230000001681 protective effect Effects 0.000 claims description 8
- 238000013468 resource allocation Methods 0.000 claims description 8
- 238000003786 synthesis reaction Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 abstract description 26
- 238000004519 manufacturing process Methods 0.000 abstract description 11
- 230000006399 behavior Effects 0.000 description 27
- 238000013473 artificial intelligence Methods 0.000 description 22
- 238000012544 monitoring process Methods 0.000 description 20
- 238000005516 engineering process Methods 0.000 description 15
- 230000009286 beneficial effect Effects 0.000 description 11
- 238000000034 method Methods 0.000 description 9
- 238000001514 detection method Methods 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000003860 storage Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000001174 ascending effect Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012946 outsourcing Methods 0.000 description 3
- 238000013439 planning Methods 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 238000013145 classification model Methods 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000007786 learning performance Effects 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 238000009430 construction management Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000004080 punching Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Water Supply & Treatment (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an AI identification system applied to distribution network operation site violation, comprising: the video image acquisition module is used for acquiring regional images of different regions of the distribution network operation site; the first transmission module is respectively connected with the video image acquisition module and the edge calculation device; the edge computing device is used for receiving the area image transmitted by the first transmission module, and carrying out intelligent identification analysis on the area image to obtain an identification result; the second transmission module is respectively connected with the edge computing device and the background server; and the background server is used for receiving the identification result transmitted by the second transmission module, generating warning information according to the identification result and transmitting the warning information to the field personnel terminal. The real-time management and control of the field operation state, the personnel operation behavior and the safe and civilized construction can be realized, and the safe production management and control efficiency and the lean management level of the construction operation can be improved in all directions.
Description
Technical Field
The invention relates to the technical field of distribution network operation site detection, in particular to an AI (artificial intelligence) identification system applied to distribution network operation site violation.
Background
The Suzhou power supply company implements gridding planning as early as 2013, divides a city distribution network into 1412 grids, takes the grids as basic units, implements the planning of reserve projects, and promotes the construction target of an 'first-class' power grid. In the process of promoting the construction of the target grid frame and the implementation of the project, the implementation environment of the project construction is generally complex, the scope is wide, the number of points is large, the construction period is short, the participation units are large, the qualification of the construction units is different from the qualification of the construction team, and higher requirements are put forward for standardization, specialization and comprehension of the project construction and safety management.
From the aspect of project construction safety management, the most critical and active factor of the safety production is the most important factor influencing the safety production. Constructors have high mobility and lack safety responsibility consciousness, and the method has the defects of poor understanding and execution of safety work standards and safety measures and poor supervision of violation behaviors, and becomes a working key point for improving field safety management.
From the company safety operation supervision and management system, the manual on-site safety operation inspection is still the most traditional, most direct and most effective safety supervision and management means. Although the video monitoring system of the operation site is brought into the security production management and control system of the provincial company, the mobile video monitoring device is gradually applied to the outdoor operation site, all levels of supervision personnel need to check against the regulations by watching the returned video in real time or calling the stored historical video in the later period, and although some regulations can be found, a large amount of manpower, energy and material resources are consumed, and possible careless situations can be artificially judged. In addition, with the increase of the number of monitoring cameras in the operation site, the requirement of transmitting real-time video streams on network transmission bandwidth is higher and higher, the network bandwidth requirement is greatly increased, and meanwhile, the pressure of a rear-end server is increased.
The intelligent and flat management and control of the operation site safety is realized, the urgent problem of improving the quality and the efficiency of the safety management of the current operation site is solved, and therefore an AI identification system applied to the violation of the distribution network operation site is urgently needed.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the art described above. Therefore, the invention aims to provide an AI identification system applied to distribution network operation site violation, which integrates the existing PMS production business system, business outsourcing safety management system and safety production operation risk management and control system according to the requirements of 'construction standardization, management standardization, team specialization and cooperative team', and realizes real-time management and control of site operation state, personnel operation behavior and safety civilization construction by applying the latest artificial intelligent image identification technology, visual perception edge calculation technology and monocular visual ranging technology on the basis of the existing operation site video monitoring system, thereby comprehensively improving the safety production management and control efficiency and the lean management level of construction operation.
In order to achieve the above object, an embodiment of the present invention provides an AI identification system applied to distribution network operation site violation, including:
the video image acquisition module is used for acquiring regional images of different regions of the distribution network operation site;
the first transmission module is respectively connected with the video image acquisition module and the edge calculation device;
the edge computing device is used for receiving the area image transmitted by the first transmission module, and carrying out intelligent identification analysis on the area image to obtain an identification result;
the second transmission module is respectively connected with the edge computing device and the background server;
and the background server is used for receiving the identification result transmitted by the second transmission module, generating warning information according to the identification result and transmitting the warning information to the field personnel terminal.
According to some embodiments of the invention, the edge computing device comprises:
the first acquisition module is used for receiving the area image transmitted by the first transmission module and acquiring the attribute information of the area image;
the first determining module is used for determining the priority order of image analysis on the area images according to the attribute information;
the second acquisition module is used for acquiring authority level information and calculation information of the edge calculation nodes;
the resource allocation management module is used for determining the control relation among a plurality of edge computing nodes according to the authority level information, and the edge computing nodes with high authority levels carry out resource allocation management on the edge computing nodes with low authority levels according to the priority order and the computing information;
the edge computing node carries out image analysis on the distributed area image to obtain an image analysis result; and judging whether the target area corresponding to the area image has area abnormality according to the image analysis result, and marking the target area with the area abnormality to obtain an identification result.
According to some embodiments of the invention, the video image acquisition module comprises at least one of an intelligent safety helmet, an intelligent deployment and control ball, and an intelligent unmanned aerial vehicle.
According to some embodiments of the invention, the edge computing device comprises:
the extraction module is used for extracting the characteristics of the region image to obtain a human body image;
the first positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain head key points and arm key points; respectively positioning a head area and an arm area according to the head key points and the arm key points;
the first recognition module is used for inputting the head area into a pre-trained head recognition model, recognizing the wearing condition of the safety helmet and obtaining a first recognition result;
and the second recognition module is used for inputting the arm area into a pre-trained skin color model, recognizing the short sleeve wearing condition and obtaining a second recognition result.
According to some embodiments of the invention, the edge computing device further comprises:
the second positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain trunk key points, and positioning a trunk area according to the trunk key points;
and the third recognition module is used for inputting the trunk area into a pre-trained trunk recognition model, recognizing the safety belt condition and obtaining a third recognition result.
According to some embodiments of the present invention, the system further includes an encryption module, connected to the edge computing device, configured to encrypt the identification result obtained by the edge computing device to obtain encrypted data, and transmit the encrypted data to the background server based on the second transmission module.
According to some embodiments of the invention, the edge computing device further comprises:
the second determination module is used for determining the abnormal object in the marking area, and identifying the action posture of the abnormal object to acquire the abnormal behavior characteristic of the abnormal object;
the third determining module is used for inquiring an abnormal behavior feature-abnormal score table according to the abnormal behavior feature and determining an abnormal score corresponding to the abnormal behavior feature;
the fourth determining module is used for determining the total abnormal score according to the abnormal score corresponding to the abnormal behavior characteristics and the number of the abnormal behavior characteristics;
and the first alarm module is used for determining the abnormal grade of the abnormal object according to the abnormal total score and sending out first alarm prompts of different grades according to different abnormal grades.
According to some embodiments of the invention, the second determining module comprises:
the determining submodule is used for determining an action posture contour map of the abnormal object;
the marking submodule is used for marking the positions of all limbs of the abnormal object on the action posture contour map according to the human body proportion model;
and the obtaining submodule is used for respectively identifying whether each limb of the abnormal object wears a protective tool or not based on a preset standard, and obtaining the abnormal behavior characteristics when the protective tool is not worn on the limb.
According to some embodiments of the invention, the edge computing device further comprises:
a fifth determining module, configured to determine a target image in the mark region;
the characteristic acquisition module is used for extracting the characteristics of the target image to acquire hue characteristics, saturation characteristics and brightness characteristics;
the comparison module is used for comparing the hue characteristics with preset hues respectively, comparing the saturation characteristics with preset saturations respectively and comparing the brightness characteristics with preset brightness respectively, and when the hue characteristics, the saturations and the brightness are determined to be consistent with each other, indicating that a fire disaster occurs in a marked area;
and the second alarm module is used for determining the range and the severity of the fire when the fire in the marked area is determined, and further sending out second alarm prompts in different grades.
In one embodiment, the edge computing device includes:
the synthesis module is used for carrying out image synthesis on the plurality of area images to obtain a high dynamic range image;
the query module is used for acquiring the environment brightness corresponding to the plurality of area images, calculating to obtain average environment brightness, and querying a preset data table according to the average environment brightness to obtain a noise reduction model;
the first noise reduction processing module is used for carrying out first noise reduction processing on the high dynamic range image according to the noise reduction model to obtain a first noise reduction image;
the conversion module is used for converting the first noise reduction image into a Bayer image based on a Bayer color filter and determining a noise reduction point and texture intensity corresponding to the noise reduction point on the Bayer image according to a preset rule;
the second noise reduction processing module is used for inquiring a preset texture intensity-noise reduction coefficient table according to the texture intensity, determining a corresponding noise reduction coefficient, and performing second noise reduction processing on the Bayer image according to the noise reduction coefficient to obtain a second noise reduction image;
and the output module is used for inputting the second noise reduction image into a pre-trained intelligent recognition model for recognition and outputting a recognition result.
Has the advantages that:
1. the edge computing device realizes violation identification, and an artificial intelligent edge computing module is added in the traditional fixed camera, the mobile distribution control ball and the intelligent safety helmet equipment to form a fixed visual analysis device and a mobile visual analysis device, so that the identification and judgment of the on-site violation behaviors can be directly carried out, and the on-site violation behaviors are actively reported to a background server. The edge calculation has the characteristics of low time delay, safety, strong flexibility and independence on cloud and network. The artificial intelligence AI recognition edge calculation is completed at the video monitoring terminal, and the problems of untimely cloud calculation response and high power consumption are solved. The most important change of edge calculation and video monitoring is to change passive monitoring into active analysis and early warning, so that the problem that massive monitoring data needs to be manually processed is solved. The edge calculation removes redundant information by preprocessing a video image, so that part or all of video analysis is migrated to the edge, thereby reducing the demands on cloud center calculation, storage and network bandwidth and improving the video analysis speed. Meanwhile, the invalid video storage is reduced, the storage space is reduced, the occupation of a large amount of high-definition video network transmission bandwidth is greatly reduced, the 'in-the-fact' evidence video data is stored to the maximum extent, the reliability of evidence information is enhanced, and the utilization rate of the video data storage space is improved.
2. The convolutional network model based on the attention mechanism realizes standard dressing identification, and a convolutional neural model based on the attention mechanism is constructed, so that local significant characteristics of a human body can be captured quickly, and the learning performance of the arm skin color characteristics is excellent; therefore, the skin color is used as foreground information, the clothing is used as background information, an area candidate Network (RPN) is introduced into an attention mechanism, and the skin color of the arm of the person is subjected to example segmentation, so that high-precision identification of standard clothing is realized.
3. After violation is identified, fast pushing rectification and real-time intervention of the background server are combined, the result of identification of the violation behaviors on the site is calculated by the edge, rectification information is pushed to the site personnel terminal in the background server, real-time intervention is achieved in operation safety construction of the power grid industry by the Internet of things technology, and early discovery and early processing are achieved within 1 minute of corresponding time.
4. The large-area automatic operation and violation identification are realized by utilizing the AI edge calculation of the unmanned aerial vehicle, so that the application of the intelligent unmanned aerial vehicle technology in the field of violation identification of operation safety construction in the power grid industry is further expanded, and the detection efficiency is improved for multi-angle remote interference-free violation detection on a large-area construction operation site.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a block diagram of an AI identification system for distribution network job site violation according to one embodiment of the present invention;
FIG. 2 is a block diagram of an edge computing device according to one embodiment of the invention;
FIG. 3 is a block diagram of an edge computing device according to yet another embodiment of the invention;
FIG. 4 is a schematic diagram of an AI identification system for distribution network job site violation according to one embodiment of the present invention;
FIG. 5 is a flow diagram of performing violation identification according to one embodiment of the present invention;
fig. 6 is a flow chart for performing violation identification according to yet another embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
As shown in fig. 1, an embodiment of the present invention provides an AI identification system applied to distribution network job site violation, including:
the video image acquisition module is used for acquiring regional images of different regions of the distribution network operation site;
the first transmission module is respectively connected with the video image acquisition module and the edge calculation device;
the edge computing device is used for receiving the area image transmitted by the first transmission module, and carrying out intelligent identification analysis on the area image to obtain an identification result;
the second transmission module is respectively connected with the edge computing device and the background server;
and the background server is used for receiving the identification result transmitted by the second transmission module, generating warning information according to the identification result and transmitting the warning information to the field personnel terminal.
The working principle of the technical scheme is as follows: the video image acquisition module is used for acquiring regional images of different regions of the distribution network operation site; the first transmission module is respectively connected with the video image acquisition module and the edge calculation device; the first transmission module is a wireless transmission module; the edge computing device is used for receiving the area image transmitted by the first transmission module, and carrying out intelligent identification analysis on the area image to obtain an identification result; the second transmission module is respectively connected with the edge computing device and the background server; the second transmission module may be a 4Gapn module. And the background server is used for receiving the identification result transmitted by the second transmission module, generating warning information according to the identification result and transmitting the warning information to the field personnel terminal. The video image acquisition module comprises at least one of an intelligent safety helmet, an intelligent distribution control ball and an intelligent unmanned aerial vehicle.
The beneficial effects of the above technical scheme are that: the edge computing system for identifying the violation behaviors in the operation field deploys the deep learning intelligent identification algorithm to the operation field end by marginalizing the computing platform. When the front-end visual analysis device identifies the violation behaviors, the identification result is transmitted to the background server, the conversion from passive monitoring to active early warning is realized, and the network bandwidth and the pressure of the background server are greatly reduced. Meanwhile, aiming at the problems of high power consumption, insufficient stock and the like of a video monitoring camera, the edge calculation module is deployed in a conventional ball distribution and control system, an intelligent safety helmet and the like, so that violation identification is realized on the basis of recycling stock monitoring equipment, and the waste of equipment resources is avoided. The result of the identification of the site violation behaviors is calculated by the edge, and the rectification information is pushed to the site personnel terminal in the background server, so that the technology of the Internet of things is interfered in real time in the operation safety construction of the power grid industry, and the corresponding time in the period only needs 1 minute, thereby realizing early discovery and early processing. And large-area automatic operation and violation identification are realized by using the AI edge calculation of the unmanned aerial vehicle. The unmanned aerial vehicle automatic cruise monitoring technology is used for developing and testing a flight mode of an unmanned aerial vehicle executing program design, so that the unmanned aerial vehicle generates a flight path through existing data and automatically executes a patrol monitoring plan; and transmitting the power grid operation site video acquired by the unmanned aerial vehicle to the edge computing device in real time by utilizing an automatic unmanned aerial vehicle and a real-time streaming media transmission protocol.
As shown in fig. 2, the edge calculation apparatus includes:
the first acquisition module is used for receiving the area image transmitted by the first transmission module and acquiring the attribute information of the area image;
the first determining module is used for determining the priority order of image analysis on the area images according to the attribute information;
the second acquisition module is used for acquiring authority level information and calculation information of the edge calculation nodes;
the resource allocation management module is used for determining the control relation among a plurality of edge computing nodes according to the authority level information, and the edge computing nodes with high authority levels carry out resource allocation management on the edge computing nodes with low authority levels according to the priority order and the computing information;
the edge computing node carries out image analysis on the distributed area image to obtain an image analysis result; and judging whether the target area corresponding to the area image has area abnormality according to the image analysis result, and marking the target area with the area abnormality to obtain an identification result.
The working principle of the technical scheme is as follows: the edge computing means comprises a number of computing edge nodes. The first acquisition module is used for receiving the area image transmitted by the first transmission module and acquiring the attribute information of the area image; the attribute information comprises the resolution ratio of the regional image, the size of occupied memory and the like; the first determining module is used for determining the priority order of image analysis on the area images according to the attribute information; in an example, the larger the memory occupied by the area image is, the more priority is given to image analysis. The resource allocation management module is used for determining the control relation among a plurality of edge computing nodes according to the authority level information, and the edge computing nodes with high authority levels carry out resource allocation management on the edge computing nodes with low authority levels according to the priority order and the computing information; the calculation information includes the calculation load capacity, calculation cost and the like of the edge calculation node. The edge computing node carries out image analysis on the distributed area image to obtain an image analysis result; and judging whether the target area corresponding to the area image has area abnormality according to the image analysis result, and marking the target area with the area abnormality to obtain an identification result.
The beneficial effects of the above technical scheme are that: the edge computing nodes are mutually communicated, resources are reasonably distributed according to the priority order, the computing capacity and the computing cost of the edge computing nodes, the load balance of the edge computing nodes is ensured, the total computing cost is reduced, the accurate analysis of the area image is realized, the area abnormity is more accurately monitored, the distribution of the edge computing nodes is more reasonable, the effective scheduling of the computing resources of the edge computing nodes is realized to realize the maximization of the computing profit, the computing speed is further improved, the quick response is realized on the edge side based on the edge computing nodes according to the area image acquired by the video image acquisition module, the requirement of intelligent monitoring is met, the marking is carried out when the area abnormity occurs, the accurate identification result is convenient to obtain, and the background server can conveniently give timely and accurate alarm, and reminding the personnel in the corresponding area to take corresponding measures in time.
According to some embodiments of the invention, the video image acquisition module comprises at least one of an intelligent safety helmet, an intelligent deployment and control ball, and an intelligent unmanned aerial vehicle.
The working principle and the beneficial effects of the technical scheme are as follows: the intelligent safety helmet is provided with an integrated 4G individual safety helmet device with an artificial intelligent edge computing module, on-site video image acquisition can be realized, and through analysis and processing technologies such as spatial positioning sensing and artificial intelligence, on-site unsafe factors and violation behaviors are intelligently identified and fed back to the background server in real time, so that risks are prompted in time. The intelligent ball distributing and controlling device is provided with an integrated ball distributing and controlling device with an artificial intelligent edge computing module, can realize field video image acquisition, intelligently identifies field unsafe factors and violation behaviors through analysis and processing technologies such as spatial positioning sensing and artificial intelligence, and feeds back the unsafe factors and the violation behaviors to a background server in real time so as to prompt risks in time. Intelligent unmanned aerial vehicle possesses the unmanned aerial vehicle of artificial intelligence edge calculation (ground satellite station) module, can realize remote video safety supervision, the discernment violating the regulations in job site to real-time feedback uploads backstage server, in time points out the risk.
As shown in fig. 4, in an embodiment, the edge computing device is used as a core processor for connecting an engineering construction site and a violation management and control platform, and the obtained site monitoring picture/video data of stock video monitoring equipment (stock guns, stock ball machines, stock distribution control balls), video safety helmets, unmanned aerial vehicles and the like are transmitted to the edge computing device for processing, so that a set of front-end video analysis system is established; and transmitting the front-end analysis result to a background server through the APN/wired public network/wireless public network, realizing refined test and autonomous early warning, and finally uploading the important early warning data to the background server or a violation management and control platform. And the management mode is optimized, and the management and control efficiency is improved. The AI identification system for distribution network operation site violation integrates the management functions of the existing PMS production service system, the service outsourcing safety management system, the safety production operation risk management system and the operation site video monitoring system, changes the previous chain type management state that the project management department and the project management department supervise the operation points layer by layer and guide the operation points by one level into the linear management of the company and the project management department on the same operation site at the same time and different places, and greatly improves the management efficiency.
The intelligent identification personnel, the safety of entering the field is highlighted. The AI identification system combines intelligent image analysis and face recognition technology, collects the field information of the operating personnel through a field video device, realizes the face recognition identity authentication of the operating personnel entering and leaving the field, increases the identification of the safety helmet and the dressing standard of the personnel before entering the field, and automatically sends out an alarm in real time for the personnel not in a list library, the safety helmet and the dressing not meeting the standard requirements. In addition, based on the video image intelligent analysis technology, a virtual electronic fence is established in a video picture, so that the situation that workers mistakenly enter a surrounding electrified dangerous area and mistakenly touch electrified equipment is avoided, and external personnel approach a construction operation area is avoided.
The front end identifies violation and increases the management and control site. The AI identification system adopts the fusion application of the video distribution control ball and the video safety helmet, so that the monitoring range of an operation site is greatly enlarged, and the newly-added artificial intelligent AI edge calculation module is used for carrying out front-end real-time identification judgment and real-time alarm on typical violation behaviors such as safety helmets, personnel dressing, safety belts and safety measures, so that the violation identification timeliness is improved. The front-end identification can effectively reduce redundant video transmission and storage space, greatly reduce the requirements on cloud center calculation, storage and network bandwidth, and improve the video analysis speed.
And (4) intelligently and comprehensively evaluating and standardizing operation behaviors. The intelligent evaluation system carried by the project records the discovered violation behaviors in real time, and identifies and hooks the violation behaviors, constructors and construction units by comparing the violation behaviors with data of a business management system such as production management, operation planning, risk management and control, outsourcing personnel management and the like, so that the intelligent evaluation system provides a powerful basis for companies and construction management departments to master and know the safety condition of field operation in time and comprehensively evaluate and assess the construction units.
Finally, the result of the identification of the site violation behaviors is calculated by the edge, and the rectification information is pushed to the site personnel terminal in the background server, so that the technology of the Internet of things is interfered in real time in the operation safety construction of the power grid industry.
As shown in fig. 3 and 5, the edge calculating apparatus includes:
the extraction module is used for extracting the characteristics of the region image to obtain a human body image;
the first positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain head key points and arm key points; respectively positioning a head area and an arm area according to the head key points and the arm key points;
the first recognition module is used for inputting the head area into a pre-trained head recognition model, recognizing the wearing condition of the safety helmet and obtaining a first recognition result;
and the second recognition module is used for inputting the arm area into a pre-trained skin color model, recognizing the short sleeve wearing condition and obtaining a second recognition result.
The working principle of the technical scheme is as follows: the edge calculation device comprises an extraction module, a feature extraction module and a feature extraction module, wherein the extraction module is used for extracting the features of the region image to obtain a human body image; the first positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain head key points and arm key points; respectively positioning a head area and an arm area according to the head key points and the arm key points; the first recognition module is used for inputting the head area into a pre-trained head recognition model, recognizing the wearing condition of the safety helmet and obtaining a first recognition result; and the second recognition module is used for inputting the arm area into a pre-trained skin color model, recognizing the short sleeve wearing condition and obtaining a second recognition result. The head recognition model is obtained by obtaining a head sample picture and performing ResNet training.
The beneficial effects of the above technical scheme are that: the key identification of the illegal behaviors of the distribution network operation site, including safety helmet and dressing conditions, is realized by adopting a convolution neural model based on an attention mechanism aiming at constructors in a monitoring video, so that local remarkable characteristics of a human body can be rapidly captured, and particularly, the learning performance of the arm skin color characteristics is excellent; therefore, the skin color is used as foreground information, the clothing is used as background information, an area candidate Network (RPN) is introduced into an attention mechanism, and the skin color of the arm of the person is subjected to example segmentation, so that high-precision identification of standard clothing is realized. First, a sample image of the head of a constructor is acquired, including three types: (1) the safety helmet is not worn; (2) the safety helmet is not normally worn; (3) the dress is correctly put on. The project constructs a depth residual error network to identify and classify the three images, and trains on the collected samples to obtain a classification model. For the non-standard clothing (short sleeves), the arm area of the constructor is obtained through an attitude estimation algorithm, and then a skin color detection model is adopted to judge whether the arm is exposed. The skin color is one of the obvious characteristics of the human body surface, although the skin color of the human body presents different colors due to different races, the skin color tone is basically consistent after the influence of brightness, visual environment and the like on the skin color is eliminated, and theoretical basis is provided for the skin color segmentation by utilizing color information. The safety helmet and the wearing condition can be conveniently and accurately identified, and whether the violation behaviors exist or not can be accurately judged. The convolutional network model based on the attention mechanism realizes standard dressing identification, and the workload of project implementation is mainly focused on applying a deep learning convolutional neural network self-learning technology and implementing a violation behavior detection module. The action recognition technology based on deep learning overcomes the visual angle of operators and the dynamic change of clothes, and accurately positions the human body behavior of the power grid operation video key frame.
As shown in fig. 6, according to some embodiments of the invention, the edge computing device further comprises:
the second positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain trunk key points, and positioning a trunk area according to the trunk key points;
and the third recognition module is used for inputting the trunk area into a pre-trained trunk recognition model, recognizing the safety belt condition and obtaining a third recognition result.
The working principle of the technical scheme is as follows: the edge calculation means further includes: the second positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain trunk key points, and positioning a trunk area according to the trunk key points; and the third recognition module is used for inputting the trunk area into a pre-trained trunk recognition model, recognizing the safety belt condition and obtaining a third recognition result. Utilize monocular vision principle to judge the operation personnel height during the operation of ascending a height, the analysis personnel of ascending a height safety belt wearing condition whether have during the use safety belt hang high with the phenomenon, whether someone's support when using the ladder, whether use the metal ladder that does not conform to safety regulation, whether the operation personnel stand more than the safety line of ladder highest point, whether have high altitude parabolic phenomenon etc..
The beneficial effects of the above technical scheme are that: to joining in marriage key discernment of net job site violation of regulations, including the construction safety belt wearing condition of ascending a height, at first, gather constructor truck sample image, contain two kinds: (1) no safety belt is used; (2) the harness is used normally. The method is characterized in that a depth residual error network is adopted to identify and classify two images as a safety helmet identification network, and a classification model is obtained by training an acquired sample. The wearing condition of the safety belt for the climbing construction is conveniently and accurately identified, and then whether the violation behaviors exist or not is accurately judged. And the first table is a statistical table of identification results of the violation behaviors.
Watch 1
In one embodiment, intelligent image analysis and face recognition technology are further utilized, face recognition identity authentication of personnel entering and exiting a production operation field and safety helmet and wearing violation identification are achieved, face recognition is added during entering the field, personnel with unmatched face recognition are pushed and alarmed at the backstage, field face pictures are collected, post-affair investigation is facilitated, common cheating behaviors such as card punching, imposition and the like are avoided, and quality safety management of construction projects is further enhanced. The voice alarm is given to the irregular dressing personnel who do not correctly wear the safety helmet and the like.
According to some embodiments of the present invention, the system further includes an encryption module, connected to the edge computing device, configured to encrypt the identification result obtained by the edge computing device to obtain encrypted data, and transmit the encrypted data to the background server based on the second transmission module.
The beneficial effects of the above technical scheme are that: the safety and the reliability of data transmission are ensured.
In one embodiment, for the stock ball distribution and control equipment, the ball distribution and control equipment deployed in the construction and maintenance site is connected to the edge computing device through wifi, the edge computing device carries out intelligent identification analysis on the acquired image, and the image with violation alarm in the site is sent to the background server through a Nanrui or Puhua encryption chip and through 4G. For the guns or the ball machines deployed in the transformer substation, the intelligent edge computing device is accessed through the Ethernet through a transformer substation local area network cabinet, the acquired images are subjected to intelligent identification and analysis, and the pictures with the warning of the violation of the badge on the site are sent to the background server through the intranet.
According to some embodiments of the invention, the edge computing device further comprises:
the second determination module is used for determining the abnormal object in the marking area, and identifying the action posture of the abnormal object to acquire the abnormal behavior characteristic of the abnormal object;
the third determining module is used for inquiring an abnormal behavior feature-abnormal score table according to the abnormal behavior feature and determining an abnormal score corresponding to the abnormal behavior feature;
the fourth determining module is used for determining the total abnormal score according to the abnormal score corresponding to the abnormal behavior characteristics and the number of the abnormal behavior characteristics;
and the first alarm module is used for determining the abnormal grade of the abnormal object according to the abnormal total score and sending out first alarm prompts of different grades according to different abnormal grades.
The working principle of the technical scheme is as follows: the second determination module is used for determining the abnormal object in the marking area, and identifying the action posture of the abnormal object to acquire the abnormal behavior characteristic of the abnormal object; the third determining module is used for inquiring an abnormal behavior feature-abnormal score table according to the abnormal behavior feature and determining an abnormal score corresponding to the abnormal behavior feature; the fourth determining module is used for determining the total abnormal score according to the abnormal score corresponding to the abnormal behavior characteristics and the number of the abnormal behavior characteristics; and the first alarm module is used for determining the abnormal grade of the abnormal object according to the abnormal total score and sending out first alarm prompts of different grades according to different abnormal grades.
The beneficial effects of the above technical scheme are that: the user can accurately acquire the abnormal grade of the marked area, and then corresponding countermeasures are taken.
According to some embodiments of the invention, the second determining module comprises:
the determining submodule is used for determining an action posture contour map of the abnormal object;
the marking submodule is used for marking the positions of all limbs of the abnormal object on the action posture contour map according to the human body proportion model;
and the obtaining submodule is used for respectively identifying whether each limb of the abnormal object wears a protective tool or not based on a preset standard, and obtaining the abnormal behavior characteristics when the protective tool is not worn on the limb.
The working principle of the technical scheme is as follows: the determining submodule is used for determining an action posture contour map of the abnormal object; the marking submodule is used for marking the positions of all limbs of the abnormal object on the action posture contour map according to the human body proportion model; and the obtaining submodule is used for respectively identifying whether each limb of the abnormal object wears a protective tool or not based on a preset standard, and obtaining the abnormal behavior characteristics when the protective tool is not worn on the limb. For example, when a non-local staff man-hour exists in a detected area, a safety helmet is preset to be worn when the non-local staff enters the area, whether the safety helmet is worn on the head of the non-local staff is judged, and abnormal behavior characteristics are obtained.
The beneficial effects of the above technical scheme are that: the method and the device realize effective monitoring of the abnormal objects in the marked area and accurately acquire the abnormal behavior characteristics.
According to some embodiments of the invention, the edge computing device further comprises:
a fifth determining module, configured to determine a target image in the mark region;
the characteristic acquisition module is used for extracting the characteristics of the target image to acquire hue characteristics, saturation characteristics and brightness characteristics;
the comparison module is used for comparing the hue characteristics with preset hues respectively, comparing the saturation characteristics with preset saturations respectively and comparing the brightness characteristics with preset brightness respectively, and when the hue characteristics, the saturations and the brightness are determined to be consistent with each other, indicating that a fire disaster occurs in a marked area;
and the second alarm module is used for determining the range and the severity of the fire when the fire in the marked area is determined, and further sending out second alarm prompts in different grades.
The beneficial effects of the above technical scheme are that: whether the fire disaster is found in the area is intelligently achieved, second alarm prompts of different levels are achieved, a user is reminded to timely process the fire disaster, loss is reduced, and meanwhile accuracy of area abnormity monitoring is guaranteed.
Aiming at the violation identification behaviors of the power grid operation field, edge calculation and artificial intelligence are used as core devices. The traditional fixed or mobile camera can be accessed, artificial intelligent edge calculation is realized at the front end of the site, unsafe factors and violation behaviors in the site are intelligently identified, and real-time feedback and uploading to a back-end server and automatic early warning are carried out; the device such as edge computing device, video safety helmet can realize scene video image acquisition, through analysis and processing techniques such as spatial localization perception, artificial intelligence, intelligent recognition scene insecurity factor and the act of breaking rules and regulations and real-time feedback upload back-end server, in time indicate the risk. Based on the image deep learning algorithm, the functions of automatic snapshot, comparison, alarm and pushing of the illegal behaviors such as no person in the picture, no safety helmet wearing correctly, no safety belt wearing correctly (short sleeve wearing), no safety belt fastening, low hanging height of the safety belt and the like are realized through edge calculation. Intelligent unmanned aerial vehicle. The unmanned aerial vehicle with the artificial intelligent edge computing (ground station) module can realize remote video safety supervision and violation identification on an operation site, feed back and upload the information to a rear-end server in real time, and prompt risks in time.
In one embodiment, an edge computing device, comprising:
the synthesis module is used for carrying out image synthesis on the plurality of area images to obtain a high dynamic range image;
the query module is used for acquiring the environment brightness corresponding to the plurality of area images, calculating to obtain average environment brightness, and querying a preset data table according to the average environment brightness to obtain a noise reduction model;
the first noise reduction processing module is used for carrying out first noise reduction processing on the high dynamic range image according to the noise reduction model to obtain a first noise reduction image;
the conversion module is used for converting the first noise reduction image into a Bayer image based on a Bayer color filter and determining a noise reduction point and texture intensity corresponding to the noise reduction point on the Bayer image according to a preset rule;
the second noise reduction processing module is used for inquiring a preset texture intensity-noise reduction coefficient table according to the texture intensity, determining a corresponding noise reduction coefficient, and performing second noise reduction processing on the Bayer image according to the noise reduction coefficient to obtain a second noise reduction image;
and the output module is used for inputting the second noise reduction image into a pre-trained intelligent recognition model for recognition and outputting a recognition result.
The working principle and the beneficial effects of the technical scheme are as follows: the synthesis module is used for carrying out image synthesis on the plurality of area images to obtain a high dynamic range image; the query module is used for acquiring the environment brightness corresponding to the plurality of area images, calculating to obtain average environment brightness, and querying a preset data table according to the average environment brightness to obtain a noise reduction model; the preset data table is an average ambient brightness-noise reduction model corresponding table. The method comprises a plurality of noise reduction models which are obtained by training sample images under different environment brightness. The first noise reduction processing module is used for carrying out first noise reduction processing on the high dynamic range image according to the noise reduction model to obtain a first noise reduction image; the method can effectively avoid overlapping shadows caused by alignment errors between images when the images are synthesized, simultaneously eliminates noise caused by different brightness, improves the image purity to a certain extent and keeps more detailed characteristics of the images. The conversion module is used for converting the first noise reduction image into a Bayer image based on a Bayer color filter and determining a noise reduction point and texture intensity corresponding to the noise reduction point on the Bayer image according to a preset rule; the Bayer image is subjected to image segmentation and is divided into a plurality of data blocks, and each data block comprises two green pixel units, one blue pixel unit and one red pixel unit. The preset rule is that the fixed position based on the data block is set as a noise reduction point row by row and column by column. The second noise reduction processing module is used for inquiring a preset texture intensity-noise reduction coefficient table according to the texture intensity, determining a corresponding noise reduction coefficient, and performing second noise reduction processing on the Bayer image according to the noise reduction coefficient to obtain a second noise reduction image; and performing noise reduction based on the similarity of the noise reduction point and the surrounding pixel units of the same color channel. The noise caused by different colors can be effectively eliminated. According to the method and the device, the luminance noise and the chrominance noise in the image are denoised respectively, the denoising effect is improved, and the recognition accuracy of inputting the denoised second denoising image into a pre-trained intelligent recognition model for recognition is further improved. The intelligent recognition model is obtained based on sample image training.
In an embodiment, the apparatus further includes a channel transmission rate detection module, configured to detect a channel transmission rate of the second transmission module, determine whether the channel transmission rate is less than a preset channel transmission rate, and send an alarm prompt when it is determined that the channel transmission rate is less than the preset channel transmission rate;
detecting a channel transmission rate of a second transmission module, comprising:
determining a data packet queue sent by the edge computing device to the second transmission module, and computing the waiting time delay M of the data packet of each threshold value i based on the length of the data packet queuei:
Wherein m isiThe number of effective data packets which are successfully sent at a sampling time interval t corresponding to a threshold value i for a transmission node; x is the number ofijTransmitting a jth data packet corresponding to a threshold i in a sampling time interval t for the transmission node; t (x)ij) For data packet xijDequeue time in a data packet queue; p (x)ij) For data packet xijAn enqueue time in a data packet queue; [ T (x)ij)-P(xij)]For data packet xijLatency in the data packet queue; l is the number of the corresponding threshold values of the transmission nodes;
calculating the average value of the waiting time delay of the data packet of each threshold value i
Calculating the channel transmission rate V of the second transmission module:
wherein H is the channel bandwidth; epsilon is the channel gain; w is transmission power; n is noise power spectral density; e is a natural constant; m0Is a preset time delay.
The working principle and the beneficial effects of the technical scheme are as follows: the channel transmission rate detection module is used for detecting the channel transmission rate of the second transmission module, judging whether the channel transmission rate is smaller than a preset channel transmission rate or not, and sending an alarm prompt when the channel transmission rate is determined to be smaller than the preset channel transmission rate; the channel transmission rate of the second transmission module is convenient to guarantee, and further the data transmission rate between the edge computing device and the background server is guaranteed, the timeliness of system response is convenient to guarantee, and user experience is improved. In the data packet queue, the end-to-end time delays corresponding to different thresholds are different, and the selection of the thresholds greatly influences the network time delay performance, and further influences the channel transmission rate. The predetermined delay is a delay that is satisfactory to most customers based on multiple experimental determinations. Based on the formula, the channel transmission rate of the second transmission module is accurately calculated, and the accuracy of judging the channel transmission rate and the preset channel transmission rate is improved.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. The utility model provides an AI identification system for join in marriage net job site violating regulations which characterized in that includes:
the video image acquisition module is used for acquiring regional images of different regions of the distribution network operation site;
the first transmission module is respectively connected with the video image acquisition module and the edge calculation device;
the edge computing device is used for receiving the area image transmitted by the first transmission module, and carrying out intelligent identification analysis on the area image to obtain an identification result;
the second transmission module is respectively connected with the edge computing device and the background server;
and the background server is used for receiving the identification result transmitted by the second transmission module, generating warning information according to the identification result and transmitting the warning information to the field personnel terminal.
2. The AI identification system for distribution network job site violations of claim 1 wherein said edge computing device includes:
the first acquisition module is used for receiving the area image transmitted by the first transmission module and acquiring the attribute information of the area image;
the first determining module is used for determining the priority order of image analysis on the area images according to the attribute information;
the second acquisition module is used for acquiring authority level information and calculation information of the edge calculation nodes;
the resource allocation management module is used for determining the control relation among a plurality of edge computing nodes according to the authority level information, and the edge computing nodes with high authority levels carry out resource allocation management on the edge computing nodes with low authority levels according to the priority order and the computing information;
the edge computing node carries out image analysis on the distributed area image to obtain an image analysis result; and judging whether the target area corresponding to the area image has area abnormality according to the image analysis result, and marking the target area with the area abnormality to obtain an identification result.
3. The AI identification system for distribution network job site violation application of claim 1, wherein said video image capture module comprises at least one of an intelligent safety helmet, an intelligent distribution control ball, and an intelligent drone.
4. The AI identification system for distribution network job site violations of claim 1 wherein said edge computing device includes:
the extraction module is used for extracting the characteristics of the region image to obtain a human body image;
the first positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain head key points and arm key points; respectively positioning a head area and an arm area according to the head key points and the arm key points;
the first recognition module is used for inputting the head area into a pre-trained head recognition model, recognizing the wearing condition of the safety helmet and obtaining a first recognition result;
and the second recognition module is used for inputting the arm area into a pre-trained skin color model, recognizing the short sleeve wearing condition and obtaining a second recognition result.
5. The AI identification system for distribution network job site violations of claim 4 wherein the edge computing device further comprises:
the second positioning module is used for detecting key points of the human body image through a posture estimation algorithm to obtain trunk key points, and positioning a trunk area according to the trunk key points;
and the third recognition module is used for inputting the trunk area into a pre-trained trunk recognition model, recognizing the safety belt condition and obtaining a third recognition result.
6. The AI identification system for distribution network job site violation according to claim 1 further comprising an encryption module coupled to said edge computing device for encrypting said identification result obtained by said edge computing device to obtain encrypted data and transmitting said encrypted data to a background server based on a second transmission module.
7. The AI identification system for distribution network job site violations of claim 2 wherein the edge computing device further comprises:
the second determination module is used for determining the abnormal object in the marking area, and identifying the action posture of the abnormal object to acquire the abnormal behavior characteristic of the abnormal object;
the third determining module is used for inquiring an abnormal behavior feature-abnormal score table according to the abnormal behavior feature and determining an abnormal score corresponding to the abnormal behavior feature;
the fourth determining module is used for determining the total abnormal score according to the abnormal score corresponding to the abnormal behavior characteristics and the number of the abnormal behavior characteristics;
and the first alarm module is used for determining the abnormal grade of the abnormal object according to the abnormal total score and sending out first alarm prompts of different grades according to different abnormal grades.
8. The AI identification system for distribution network job site violations of claim 7 wherein said second determination module includes:
the determining submodule is used for determining an action posture contour map of the abnormal object;
the marking submodule is used for marking the positions of all limbs of the abnormal object on the action posture contour map according to the human body proportion model;
and the obtaining submodule is used for respectively identifying whether each limb of the abnormal object wears a protective tool or not based on a preset standard, and obtaining the abnormal behavior characteristics when the protective tool is not worn on the limb.
9. The AI identification system for distribution network job site violations of claim 2 wherein the edge computing device further comprises:
a fifth determining module, configured to determine a target image in the mark region;
the characteristic acquisition module is used for extracting the characteristics of the target image to acquire hue characteristics, saturation characteristics and brightness characteristics;
the comparison module is used for comparing the hue characteristics with preset hues respectively, comparing the saturation characteristics with preset saturations respectively and comparing the brightness characteristics with preset brightness respectively, and when the hue characteristics, the saturations and the brightness are determined to be consistent with each other, indicating that a fire disaster occurs in a marked area;
and the second alarm module is used for determining the range and the severity of the fire when the fire in the marked area is determined, and further sending out second alarm prompts in different grades.
10. The AI identification system for distribution network job site violations of claim 1 wherein said edge computing device includes:
the synthesis module is used for carrying out image synthesis on the plurality of area images to obtain a high dynamic range image;
the query module is used for acquiring the environment brightness corresponding to the plurality of area images, calculating to obtain average environment brightness, and querying a preset data table according to the average environment brightness to obtain a noise reduction model;
the first noise reduction processing module is used for carrying out first noise reduction processing on the high dynamic range image according to the noise reduction model to obtain a first noise reduction image;
the conversion module is used for converting the first noise reduction image into a Bayer image based on a Bayer color filter and determining a noise reduction point and texture intensity corresponding to the noise reduction point on the Bayer image according to a preset rule;
the second noise reduction processing module is used for inquiring a preset texture intensity-noise reduction coefficient table according to the texture intensity, determining a corresponding noise reduction coefficient, and performing second noise reduction processing on the Bayer image according to the noise reduction coefficient to obtain a second noise reduction image;
and the output module is used for inputting the second noise reduction image into a pre-trained intelligent recognition model for recognition and outputting a recognition result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110914549.1A CN113705372B (en) | 2021-08-10 | 2021-08-10 | AI identification system for join in marriage net job site violating regulations |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110914549.1A CN113705372B (en) | 2021-08-10 | 2021-08-10 | AI identification system for join in marriage net job site violating regulations |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113705372A true CN113705372A (en) | 2021-11-26 |
CN113705372B CN113705372B (en) | 2023-04-07 |
Family
ID=78652135
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110914549.1A Active CN113705372B (en) | 2021-08-10 | 2021-08-10 | AI identification system for join in marriage net job site violating regulations |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113705372B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114070839A (en) * | 2022-01-17 | 2022-02-18 | 国网瑞嘉(天津)智能机器人有限公司 | Intelligent wearable device interaction system, method, electronic device and storage medium |
CN114531619A (en) * | 2022-02-15 | 2022-05-24 | 广州伏羲智能科技有限公司 | Construction site monitoring method and system based on edge calculation |
CN115100604A (en) * | 2022-07-11 | 2022-09-23 | 国网江苏省电力有限公司南通市通州区供电分公司 | Feedback method and system based on intelligent safety helmet |
CN115797850A (en) * | 2023-02-06 | 2023-03-14 | 中国石油大学(华东) | Oil field production safety early warning analysis system based on video stream |
CN115934318A (en) * | 2022-11-16 | 2023-04-07 | 鹏橙网络技术(深圳)有限公司 | Employee file management method, system and device |
CN117115755A (en) * | 2023-10-23 | 2023-11-24 | 科曼智能科技有限公司 | Power operation site violation monitoring alarm recognition system based on image recognition |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106533941A (en) * | 2016-11-25 | 2017-03-22 | 中国电子科技集团公司第五十四研究所 | Compressed sensing theory based low-jitter coding awareness routing method |
CN108174165A (en) * | 2018-01-17 | 2018-06-15 | 重庆览辉信息技术有限公司 | Electric power safety operation and O&M intelligent monitoring system and method |
CN108242159A (en) * | 2018-03-09 | 2018-07-03 | 连云港杰瑞电子有限公司 | A kind of urban traffic area coordinated control system based on edge calculations node |
CN108271172A (en) * | 2018-01-11 | 2018-07-10 | 重庆邮电大学 | A kind of honeycomb D2D communication joint clusters and content disposition method |
CN110119656A (en) * | 2018-02-07 | 2019-08-13 | 中国石油化工股份有限公司 | Intelligent monitor system and the scene monitoring method violating the regulations of operation field personnel violating the regulations |
CN112272288A (en) * | 2020-10-23 | 2021-01-26 | 岭东核电有限公司 | Nuclear power operation monitoring method and device, computer equipment and storage medium |
CN112927464A (en) * | 2020-11-23 | 2021-06-08 | 北京宏链科技有限公司 | Job site macro-chain safety management and control system based on edge intelligence |
CN113158752A (en) * | 2021-02-05 | 2021-07-23 | 国网河南省电力公司鹤壁供电公司 | Intelligent safety management and control system for electric power staff approach operation |
-
2021
- 2021-08-10 CN CN202110914549.1A patent/CN113705372B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106533941A (en) * | 2016-11-25 | 2017-03-22 | 中国电子科技集团公司第五十四研究所 | Compressed sensing theory based low-jitter coding awareness routing method |
CN108271172A (en) * | 2018-01-11 | 2018-07-10 | 重庆邮电大学 | A kind of honeycomb D2D communication joint clusters and content disposition method |
CN108174165A (en) * | 2018-01-17 | 2018-06-15 | 重庆览辉信息技术有限公司 | Electric power safety operation and O&M intelligent monitoring system and method |
CN110119656A (en) * | 2018-02-07 | 2019-08-13 | 中国石油化工股份有限公司 | Intelligent monitor system and the scene monitoring method violating the regulations of operation field personnel violating the regulations |
CN108242159A (en) * | 2018-03-09 | 2018-07-03 | 连云港杰瑞电子有限公司 | A kind of urban traffic area coordinated control system based on edge calculations node |
CN112272288A (en) * | 2020-10-23 | 2021-01-26 | 岭东核电有限公司 | Nuclear power operation monitoring method and device, computer equipment and storage medium |
CN112927464A (en) * | 2020-11-23 | 2021-06-08 | 北京宏链科技有限公司 | Job site macro-chain safety management and control system based on edge intelligence |
CN113158752A (en) * | 2021-02-05 | 2021-07-23 | 国网河南省电力公司鹤壁供电公司 | Intelligent safety management and control system for electric power staff approach operation |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114070839A (en) * | 2022-01-17 | 2022-02-18 | 国网瑞嘉(天津)智能机器人有限公司 | Intelligent wearable device interaction system, method, electronic device and storage medium |
CN114531619A (en) * | 2022-02-15 | 2022-05-24 | 广州伏羲智能科技有限公司 | Construction site monitoring method and system based on edge calculation |
CN115100604A (en) * | 2022-07-11 | 2022-09-23 | 国网江苏省电力有限公司南通市通州区供电分公司 | Feedback method and system based on intelligent safety helmet |
CN115934318A (en) * | 2022-11-16 | 2023-04-07 | 鹏橙网络技术(深圳)有限公司 | Employee file management method, system and device |
CN115934318B (en) * | 2022-11-16 | 2023-09-19 | 鹏橙网络技术(深圳)有限公司 | Staff file management method, system and device |
CN115797850A (en) * | 2023-02-06 | 2023-03-14 | 中国石油大学(华东) | Oil field production safety early warning analysis system based on video stream |
CN117115755A (en) * | 2023-10-23 | 2023-11-24 | 科曼智能科技有限公司 | Power operation site violation monitoring alarm recognition system based on image recognition |
CN117115755B (en) * | 2023-10-23 | 2024-01-23 | 科曼智能科技有限公司 | Power operation site violation monitoring alarm recognition system based on image recognition |
Also Published As
Publication number | Publication date |
---|---|
CN113705372B (en) | 2023-04-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113705372B (en) | AI identification system for join in marriage net job site violating regulations | |
CN111508097B (en) | Site inspection method, device, equipment and storage medium | |
CN109377703A (en) | A kind of forest fireproofing early warning system and its method based on machine vision | |
CN112396658B (en) | Indoor personnel positioning method and system based on video | |
CN111275923B (en) | Man-machine collision early warning method and system for construction site | |
CN112804498A (en) | System and method for monitoring wearing condition of safety protection harness | |
CN112804302A (en) | Power video image analysis system and method based on cloud edge cooperation | |
CN114665608B (en) | Intelligent sensing inspection system and method for transformer substation | |
CN113610011A (en) | AI identification system for distribution network operation site personnel management and control | |
CN109935080A (en) | The monitoring system and method that a kind of vehicle flowrate on traffic route calculates in real time | |
CN114863489B (en) | Virtual reality-based movable intelligent auxiliary inspection method and system for construction site | |
CN111920129A (en) | Intelligent safety helmet system | |
CN112184773A (en) | Helmet wearing detection method and system based on deep learning | |
CN115035088A (en) | Helmet wearing detection method based on yolov5 and posture estimation | |
CN213128247U (en) | Intelligent safety helmet system | |
CN113239760A (en) | Power grid operation field violation identification system | |
CN115797850A (en) | Oil field production safety early warning analysis system based on video stream | |
CN113988573A (en) | Risk judgment method, system and medium for routing inspection unmanned aerial vehicle based on power system | |
CN115761591A (en) | Multi-operation-site violation identification method | |
CN115880631A (en) | Power distribution station fault identification system, method and medium | |
Jiao et al. | Forest fire detection with color features and wavelet analysis based on aerial imagery | |
CN112532927A (en) | Intelligent safety management and control system for construction site | |
CN115690687A (en) | Safe wearing standard detection system based on deep learning technology | |
CN115829324A (en) | Personnel safety risk silent monitoring method | |
CN115641607A (en) | Method, device, equipment and storage medium for detecting wearing behavior of power construction site operator |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |