CN117575165B - Intelligent patrol management method and system for digital power distribution network - Google Patents
Intelligent patrol management method and system for digital power distribution network Download PDFInfo
- Publication number
- CN117575165B CN117575165B CN202311651704.0A CN202311651704A CN117575165B CN 117575165 B CN117575165 B CN 117575165B CN 202311651704 A CN202311651704 A CN 202311651704A CN 117575165 B CN117575165 B CN 117575165B
- Authority
- CN
- China
- Prior art keywords
- inspection
- target
- images
- sets
- management
- 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.)
- Active
Links
- 238000007726 management method Methods 0.000 title claims abstract description 132
- 238000007689 inspection Methods 0.000 claims abstract description 200
- 239000013598 vector Substances 0.000 claims abstract description 93
- 238000012545 processing Methods 0.000 claims abstract description 63
- 230000002159 abnormal effect Effects 0.000 claims abstract description 29
- 238000003066 decision tree Methods 0.000 claims abstract description 6
- 238000004458 analytical method Methods 0.000 claims description 28
- 238000012549 training Methods 0.000 claims description 25
- 238000010276 construction Methods 0.000 claims description 23
- 238000000034 method Methods 0.000 claims description 12
- 230000001360 synchronised effect Effects 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 10
- 238000013507 mapping Methods 0.000 description 6
- 238000012216 screening Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- 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
-
- 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
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- 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)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Operations Research (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an intelligent inspection management method and system for a digital power distribution network, which relate to the technical field of image processing, collect and preprocess infrared images of inspection nodes, combine inspection recognition channels constructed by normal image input, screen and gray scale a plurality of judgment results, calculate local vectors of a plurality of windows, further calculate target similarity to make node abnormal grade decisions and management scheme decisions, solve the technical problems that in the prior art, the inspection of the power distribution network cannot be carried out in a contactless manner, the operation of a power grid can be influenced during the inspection, the accuracy of part of automatic inspection means is lower, and the operation and maintenance requirements of the power distribution network cannot be met; and an independent processing channel is constructed by combining a decision tree, the decision of an abnormal grade and a management scheme is carried out, the non-contact inspection and accurate decision management of the operation of the power distribution network are realized, and the timely effectiveness of operation and maintenance is ensured.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an intelligent patrol management method and system for a digital power distribution network.
Background
The power supply is a necessary guarantee for life and production, timely discovery and effective treatment of the power distribution network fault problem are the basis for maintaining the integral safe and stable operation, and certain technical limitations exist in the construction aspect of the power distribution network nowadays. In the prior art, the power distribution network inspection cannot be conducted in a contactless manner, the operation of the power grid can be influenced during inspection, the accuracy of a part of automatic inspection means is low, and the operation and maintenance requirements of the power distribution network cannot be met.
Disclosure of Invention
The application provides an intelligent inspection management method and system for a digital power distribution network, which are used for solving the technical problems that in the prior art, the inspection of the power distribution network cannot be carried out in a contactless manner, the operation of the power distribution network can be influenced during the inspection, the accuracy of a part of automatic inspection means is low, and the operation and maintenance requirements of the power distribution network cannot be met.
In view of the above problems, the application provides an intelligent patrol management method and system for a digital power distribution network.
In a first aspect, the present application provides an intelligent patrol management method for a digital power distribution network, where the method includes:
collecting a plurality of infrared images of a plurality of inspection nodes in an inspection power distribution network;
constructing an inspection identification channel, wherein the inspection identification channel comprises a plurality of inspection identification paths corresponding to the plurality of inspection nodes, and the plurality of inspection identification paths are constructed based on a twin network;
Preprocessing the plurality of infrared images, combining a plurality of normal images of the plurality of inspection nodes in a normal running state, respectively inputting the plurality of inspection recognition paths to obtain a plurality of judgment results, wherein each judgment result comprises a judgment result of whether the preprocessed infrared images are consistent with the normal images;
Carrying out graying treatment on a plurality of target infrared images and a plurality of target normal images which are judged to be negative, and calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images to obtain a plurality of first local vector sets and a plurality of second local vector sets;
Calculating the similarity of the plurality of target infrared images and the plurality of target normal images according to the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarity;
And according to the multiple target similarities, carrying out node anomaly level decision and management scheme decision to obtain multiple target node anomalies and multiple target node management schemes, and carrying out inspection management on multiple target inspection nodes.
In a second aspect, the present application provides an intelligent patrol management system for a digital power distribution network, where the system includes:
the image acquisition module is used for acquiring a plurality of infrared images of a plurality of inspection nodes in the power distribution network for inspection;
The channel construction module is used for constructing a patrol identification channel, wherein the patrol identification channel comprises a plurality of patrol identification paths corresponding to the plurality of patrol nodes, and the plurality of patrol identification paths are constructed based on a twin network;
The image judging module is used for preprocessing the plurality of infrared images, combining a plurality of normal images of the plurality of inspection nodes in a normal running state, respectively inputting the images into the plurality of inspection recognition paths to obtain a plurality of judging results, wherein each judging result comprises a judging result of whether the preprocessed infrared images are consistent with the normal images or not;
The local vector acquisition module is used for carrying out graying treatment on a plurality of target infrared images and a plurality of target normal images which are judged to be negative, calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images, and obtaining a plurality of first local vector sets and a plurality of second local vector sets;
The similarity calculation module is used for calculating the similarity of the plurality of target infrared images and the plurality of target normal images according to the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarities;
And the decision management module is used for making node anomaly level decisions and management scheme decisions according to the plurality of target similarities, obtaining a plurality of target node anomalies and a plurality of target node management schemes and carrying out inspection management on a plurality of target inspection nodes.
One or more technical schemes provided by the application have at least the following technical effects or advantages:
According to the intelligent inspection management method for the digital power distribution network, provided by the embodiment of the application, a plurality of infrared images of a plurality of inspection nodes are collected in the power distribution network for inspection; constructing an inspection identification channel, wherein the inspection identification channel comprises a plurality of inspection identification paths corresponding to the plurality of inspection nodes, and the plurality of inspection identification paths are constructed based on a twin network; preprocessing the infrared images, combining the normal images of the inspection nodes in a normal running state, respectively inputting the normal images into the inspection recognition paths to obtain a plurality of judging results, screening the target infrared images and the target normal images with the judging results being negative, carrying out graying processing, calculating local vectors of windows in the target infrared images and the target normal images to obtain a plurality of first local vector sets and a plurality of second local vector sets, calculating and obtaining a plurality of target similarity, carrying out node abnormal grade decision and management scheme decision, and obtaining a plurality of target node abnormal degrees and a plurality of target node management scheme to carry out inspection management of the target inspection nodes.
The application solves the technical problems that in the prior art, the non-contact inspection cannot be realized for the inspection of the power distribution network, the operation of the power grid can be influenced during the inspection, the accuracy of part of automatic inspection means is lower, the operation and maintenance requirements of the power distribution network cannot be met, the infrared image and the similarity analysis are introduced, the infrared image is acquired for the inspection node, the modeling similarity analysis is carried out on the basis of a twin network to determine the stored abnormal image, an independent processing channel is constructed by combining a decision tree, the decision of an abnormal grade and a management scheme is carried out, the non-contact inspection and accurate decision management of the operation of the power distribution network is realized, and the timely effectiveness of the operation and maintenance is ensured.
Drawings
FIG. 1 is a schematic flow chart of an intelligent patrol management method for a digital power distribution network;
Fig. 2 is a schematic diagram of a patrol identification channel construction flow in an intelligent patrol management method for a digital power distribution network;
FIG. 3 is a schematic diagram of a process for obtaining anomaly degree of a plurality of target nodes and a plurality of target node management schemes in an intelligent patrol management method of a digital power distribution network;
fig. 4 is a schematic structural diagram of an intelligent patrol management system for a digital power distribution network.
Reference numerals illustrate: the system comprises an image acquisition module 11, a channel construction module 12, an image judgment module 13, a local vector acquisition module 14, a similarity calculation module 15 and a decision management module 16.
Detailed Description
The application provides an intelligent inspection management method and system for a digital power distribution network, which are used for acquiring infrared images of a plurality of inspection nodes, preprocessing the infrared images, combining an inspection identification channel constructed by normal image input, screening and graying a plurality of judgment results, calculating local vectors of a plurality of windows, and further calculating target similarity to carry out node abnormal grade decision and management scheme decision, so as to solve the technical problems that in the prior art, the inspection of the power distribution network cannot be carried out in a contactless manner, the operation of a power grid can be influenced during the inspection, and the accuracy of part of automatic inspection means is lower and cannot meet the operation and maintenance requirements of the power distribution network.
Example 1
As shown in fig. 1, the application provides an intelligent patrol management method for a digital power distribution network, which comprises the following steps:
step S100: collecting a plurality of infrared images of a plurality of inspection nodes in an inspection power distribution network;
in particular, at present, the improvement of life quality is accompanied by the improvement of the demand of the electric power market, the timely discovery and effective treatment of the fault problem of the power distribution network are the basis for maintaining the integral safe and stable operation, and certain technical limitations exist in the construction aspect of the power distribution network nowadays. According to the intelligent patrol management method for the digital power distribution network, provided by the application, infrared images are acquired aiming at patrol nodes, and modeling similarity analysis is performed on the basis of a twin network so as to determine different images; and an independent processing channel is constructed by combining a decision tree, the decision of an abnormal grade and a management scheme is carried out, the non-contact inspection and accurate decision management of the operation of the power distribution network are realized, and the timely effectiveness of operation and maintenance is ensured.
Specifically, a plurality of inspection nodes are set in the power distribution network to be inspected, and the inspection nodes are arranged in a self-defined mode based on inspection requirements, for example, the inspection nodes are uniformly distributed according to network distribution of the power distribution network. And further, remote infrared image acquisition is performed on the plurality of inspection nodes, and an unmanned aerial vehicle terminal device is introduced, and an infrared acquisition device is installed on the device, wherein the infrared acquisition device preferably has a high-precision holder, such as a thermal infrared imager, based on the installed infrared acquisition device, the infrared image acquisition of the power distribution network operation site is performed, based on the plurality of inspection nodes, the acquired images are matched and integrated, the plurality of infrared images are generated, and the plurality of infrared images are acquired source data for performing power distribution network operation abnormality judgment.
Step S200: constructing an inspection identification channel, wherein the inspection identification channel comprises a plurality of inspection identification paths corresponding to the plurality of inspection nodes, and the plurality of inspection identification paths are constructed based on a twin network;
further, as shown in fig. 2, the inspection identification channel is constructed, and step S200 of the present application further includes:
step S210: collecting historical inspection infrared image data of the plurality of inspection nodes to obtain the plurality of normal images and a plurality of historical infrared image sets;
Step S220: constructing a plurality of patrol recognition paths based on a twin network, wherein each patrol recognition path comprises two image convolution processing branches with the same network structure, the input data of one image convolution processing branch is a normal image, and the input data of the other image convolution processing branch is an infrared image;
Step S230: and adopting the normal images and the historical infrared image sets to perform synchronous supervision training on two image convolution processing branches in the patrol recognition paths to obtain the patrol recognition channel.
Further, the step S230 of the present application further includes performing synchronous supervision training on two image convolution processing branches in the plurality of inspection recognition paths by using the plurality of normal images and the plurality of historical infrared image sets:
step S231: constructing two image convolution processing branches in the plurality of patrol recognition paths based on a convolution neural network;
step S232: judging whether the historical infrared images are consistent with the corresponding normal images according to the normal images and the historical infrared image sets respectively to obtain a plurality of historical judgment result sets;
Step S233: adopting the plurality of normal images, the plurality of historical infrared image sets and the plurality of historical judgment result sets to respectively carry out synchronous supervision training on two image convolution processing branches in the plurality of patrol recognition paths until the two image convolution processing branches meet preset convergence conditions;
Step S234: and verifying and testing the plurality of inspection recognition paths, and obtaining the constructed plurality of inspection recognition paths when the model loss is smaller than the loss threshold value.
Specifically, the plurality of inspection nodes are respectively used as search targets, the power distribution network inspection records in a preset time interval are obtained, node information search is carried out, and a historical infrared image set corresponding to each inspection node is obtained; and meanwhile, calling the normal images of the plurality of inspection nodes, and extracting and determining based on the operation and maintenance standards of the power distribution network to acquire the plurality of normal images and the plurality of historical infrared image sets. Further constructing an inspection recognition path based on the twin network, wherein the inspection recognition path comprises two image convolution processing branches with the same network structure, and the system is respectively used for processing and analyzing the normal image and the infrared image, carrying out parallel arrangement of two channels, generating a patrol recognition path, and respectively constructing the patrol recognition path aiming at the plurality of patrol nodes.
Specifically, two image convolution processing branches in the plurality of inspection recognition paths are constructed based on a convolution neural network, and the specific network structure is the same and can comprise a convolution layer, a multistage pooling layer and a full connection layer. And mapping and corresponding the plurality of normal images and the plurality of historical infrared image sets, and judging the consistency of the images, wherein the plurality of historical judgment result sets can be generated by manually extracting and checking the image characteristics. And based on the plurality of inspection nodes, performing mapping association of the plurality of normal images, the plurality of infrared historical image sets and the plurality of historical judgment result sets by the inspection nodes to acquire a plurality of groups of training data.
Further extracting a group of training data, determining a training target based on the plurality of inspection recognition paths, taking a normal image and an infrared historical image set in the group of training data as input data of two image convolution processing branches in the inspection recognition path to be trained, taking a corresponding historical judgment result set as output data, performing synchronous supervision training of the two image convolution processing branches, and generating an inspection recognition path corresponding to the inspection node. And inputting the set of training data into the inspection recognition path again for inspection, checking the output result and the corresponding historical judgment result, judging whether the preset convergence condition is met, for example, the training precision meets the loss threshold value, if the checking result is not met, extracting the data which does not meet the loss threshold value from the set of training data, and carrying out supervision training and inspection again until the preset convergence condition is met, thereby obtaining the constructed inspection recognition path.
Further, the construction and verification test of the patrol identification paths are performed for the plurality of patrol nodes by respectively matching and calling the corresponding data in the plurality of normal images, the plurality of historical infrared image sets and the plurality of historical judgment result sets, wherein the construction methods of the patrol identification paths corresponding to the patrol nodes are the same, the specific construction data are different, and the loss threshold is met. Integrating the constructed multiple patrol identification paths and carrying out patrol node identification, and generating the constructed multiple patrol identification paths as the patrol identification channels. The inspection recognition paths are used for image analysis processing, have analysis pertinence and meet the processing precision requirement.
Step S300: preprocessing the plurality of infrared images, combining a plurality of normal images of the plurality of inspection nodes in a normal running state, respectively inputting the plurality of inspection recognition paths to obtain a plurality of judgment results, wherein each judgment result comprises a judgment result of whether the preprocessed infrared images are consistent with the normal images;
Specifically, as the infrared image acquisition process is in a moving state, noise and other factors exist in the acquired infrared image, so that the accuracy of an analysis result is influenced. The method includes the steps of performing special denoising processing on an unstable infrared image, for example, performing image segmentation processing on an acquired infrared image by taking the existence of caused noise as a standard, screening a noise-existing region for performing image denoising processing, and performing specific denoising mode without specific limitation, for example, performing denoising processing based on a wavelet denoising method, acquiring a plurality of preprocessed infrared images, combining the plurality of normal images of the plurality of inspection nodes in a normal operation state, wherein the plurality of normal images are the normal infrared images of the plurality of inspection nodes in the normal operation state, performing matching and input of a plurality of inspection recognition paths, so as to perform consistency mapping and checking of convolution characteristics of inter-channel image extraction, judging whether the preprocessed infrared images are consistent with corresponding normal images, acquiring a plurality of judgment results, and preferably, performing identification of judgment results based on different identification modes, so that the subsequent recognition analysis is performed, for example, the consistency judgment is positive, and the consistency judgment is negative.
Step S400: carrying out graying treatment on a plurality of target infrared images and a plurality of target normal images which are judged to be negative, and calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images to obtain a plurality of first local vector sets and a plurality of second local vector sets;
Further, the step S400 of the present application further includes calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images:
step S410: dividing the target infrared images and target normal images after graying treatment to obtain a plurality of first window sets and a plurality of second window sets, wherein each window comprises nine gray values;
Step S420: and in each window in the first window sets and the second window sets, calculating the difference value between the gray value of the edge point and the gray value of the middle point, and judging to obtain the first local vector sets and the second local vector sets.
Specifically, the plurality of judgment results are identified, and the plurality of target infrared images and the plurality of target normal images, the image group with abnormal operation possibly exists, are extracted if the judgment results are negative. When the local feature analysis is carried out, the identified local feature operator has gray scale invariance, and the gray scale image is needed to be used as a substrate for analysis. And carrying out graying processing on the extracted images, for example, configuring graying parameters, and carrying out processing by auxiliary operation software, such as matlab and the like, so as to obtain the target infrared images and the target normal images after the graying processing.
Further, the plurality of target infrared images and the plurality of target normal images after the graying processing are respectively divided by taking a central pixel and eight adjacent pixels as standards, the plurality of first window sets and the plurality of second window sets are determined, and each window contains nine gray values corresponding to each pixel. The difference between the current infrared image and the normal image in the normal running state is analyzed by introducing local features so as to judge whether the elements in the power distribution network are normal or not.
Further, traversing the first window sets and the second window sets, and calculating the difference value of gray values window by window, namely respectively calculating the difference value of the gray values of eight neighborhood pixels and the gray value of the center pixel by taking the gray value of the center pixel as a reference, and if the difference value is a positive number, marking as 1; if the difference is negative, the difference is marked as-1; if the difference value is 0, the difference value is recorded as 0, the gray difference value calculation results of the same window are sequentially arranged, an eight-bit binary number value is generated and used as a local vector, so that the refinement of the local vector is ensured, and the subsequent abnormality judgment accuracy is ensured. And calculating local vectors window by window, and carrying out integration normalization to generate the plurality of first local vector sets and the plurality of second local vector sets.
Step S500: calculating the similarity of the plurality of target infrared images and the plurality of target normal images according to the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarity;
further, according to the first local vector sets and the second local vector sets, the similarity between the target infrared images and the target normal images is calculated to obtain target similarities, and the step S500 further includes:
Step S510: according to the plurality of first local vector sets and the plurality of second local vector sets, carrying out division matching according to the corresponding first windows and second windows to obtain a plurality of local vector combinations;
step S520: and calculating the difference values of the first local vector and the second local vector in the plurality of local vector combinations to obtain a plurality of difference value sets, and summing the absolute values of the difference values in the plurality of difference value sets to obtain the plurality of target similarities.
Specifically, the plurality of first local vector sets and the plurality of second local vector sets are feature vectors for performing image differentiation determination. And mapping the plurality of first local vector sets and the plurality of second local feature vector sets based on the correspondence of the first local vector and the second local vector, wherein a group of local vectors is formed by mapping integration of local vectors in a group of images corresponding to the first window and the second window.
Further, traversing the plurality of local vector combinations, performing difference calculation on the first local vector and the second local vector mapped in each local vector combination, for example, the difference may be 2, 1, 0, -1, -2, obtaining the plurality of difference sets, determining the difference set in a group of images, performing addition calculation on the absolute value of the difference, and taking the calculation result as the similarity between the target infrared image and the target normal image in the group of images, wherein the similarity is smaller when the images are more similar, so as to improve the processing of image details and ensure the accuracy of obtaining the image similarity. The target infrared image and the target normal image correspond to a difference value set, and the difference value set comprises differences of local vectors in a plurality of corresponding first windows and second windows. And respectively calculating the plurality of difference value sets to acquire the plurality of target similarities. And carrying out abnormal judgment on the inspection nodes based on the multiple target similarities. According to the embodiment of the application, the window division and the local vector construction are carried out based on the LBP operator, so that the influence of illumination and the like on an infrared image can be eliminated, and the accuracy of similarity calculation with a normal image is improved.
Step S600: and according to the multiple target similarities, carrying out node anomaly level decision and management scheme decision to obtain multiple target node anomalies and multiple target node management schemes, and carrying out inspection management on multiple target inspection nodes.
Further, as shown in fig. 3, according to the multiple target similarities, a node anomaly level decision and a management scheme decision are performed, and step S600 of the present application further includes:
step S610: acquiring and processing historical inspection data of the power distribution network, acquiring a plurality of sample target similarities, and acquiring a plurality of sample node abnormal grades and a plurality of sample management schemes;
Step S620: constructing a patrol analysis branch and a management decision branch based on the similarity of the plurality of sample targets, the abnormal grades of the plurality of sample nodes and the plurality of sample management schemes, and integrating to obtain a patrol management channel;
Step S630: and inputting the plurality of target similarities into the inspection management channel to obtain the plurality of target node anomalies and a plurality of target node management schemes.
Further, based on the plurality of sample target similarities, the plurality of sample node anomaly levels, and the plurality of sample management schemes, a patrol analysis branch and a management decision branch are constructed, and step S620 of the present application further includes:
Step S621: taking the target similarity as a decision feature, and constructing a multi-layer inspection decision node and a multi-layer management decision node according to a decision tree based on the plurality of sample target similarities;
Step S622: adopting the abnormal grades of the plurality of sample nodes and a plurality of sample management schemes as decision results of the multi-layer inspection decision nodes and the multi-layer management decision nodes to obtain inspection analysis branches and management decision branches;
step S623: and merging decision nodes in the inspection analysis branch and the management decision branch to obtain the inspection management channel.
Specifically, historical inspection data in a preset time period of the power distribution network is collected, wherein the preset time period is a historical time interval bordering the current time point. And carrying out abnormal data identification on the historical inspection data, acquiring an abnormal image and carrying out similarity analysis to acquire the target similarity of the plurality of samples, wherein the specific similarity processing mode is the same as the earlier processing mode of the embodiment of the application, and the abnormal grades of the plurality of sample nodes and the plurality of sample management schemes can be directly identified and extracted. Illustratively, the smaller the sample target similarity, the greater the sample node anomaly level, the more urgent the sample management scheme, e.g., shut down power distribution for overhaul, etc.
Further, mapping and correlating the plurality of sample target similarities, the plurality of sample node anomaly levels and the plurality of sample management schemes to determine sample construction data. And constructing the patrol management channel based on the sample construction data. Specifically, taking the target similarity as a decision feature, randomly extracting a term based on the plurality of sample target similarities, respectively serving as a layer of inspection decision nodes and a layer of management decision nodes, and performing classification of the plurality of sample target similarities; further randomly extracting a term based on the similarity of the plurality of sample targets again, respectively serving as a two-layer inspection decision node and a two-layer management decision node, and dividing again according to the two-layer classification result; repeating construction of decision nodes for a plurality of times until convergence conditions are met, for example, the maximum construction layer number is reached, acquiring N layers of inspection decision nodes and N layers of management decision nodes, and connecting the one layer of inspection decision nodes and the two layers of inspection decision nodes to the layers of inspection decision nodes until the N layers of inspection decision nodes to generate the multi-layer inspection decision nodes; and carrying out hierarchical connection on the one-layer management decision node and the two-layer management decision node until the N-layer management decision node to generate the multi-layer management decision node.
Further, traversing the multi-layer routing inspection decision node, matching the abnormal grades of the plurality of sample nodes, and generating the routing inspection analysis branch by using the abnormal grades as a hierarchical decision result and marking the abnormal grades; and traversing the multi-layer and multi-layer management decision node, matching the plurality of sample management schemes, serving as a hierarchical decision result, identifying, and generating the management decision branch to perform targeted independent decision processing. And further merging the patrol analysis branch with the management decision branch, and merging patrol decision nodes with the same sample target similarity with the management decision nodes to serve as the patrol management channel. Based on the inspection management channel, based on the input target similarity, abnormal decisions and management decisions of inspection nodes are accurately and efficiently carried out.
Further, the target similarities are input into the inspection analysis branch and the management decision branch of the inspection management channel, hierarchical matching decisions are respectively carried out, identification data corresponding to matching results are identified, the identification data are used as the anomaly of the target nodes and the management schemes of the target nodes to be output, the anomaly of the target nodes and the management schemes of the target nodes are in one-to-one correspondence, and the anomaly of the target nodes and the management schemes of the target nodes are used for carrying out fault management and control of corresponding inspection nodes.
The intelligent patrol management method for the digital power distribution network provided by the embodiment of the application has the following technical effects:
1. The remote infrared image acquisition is carried out to realize non-contact inspection, a plurality of inspection recognition paths corresponding to each inspection node are constructed based on a twin network, the consistency analysis and judgment of the images are carried out by taking the infrared images and the normal images as inputs, the targeted construction of the inspection recognition paths is carried out, the degree of fit with the corresponding inspection nodes is ensured, the processing accuracy is improved, and the efficient and accurate screening of the abnormal infrared images is carried out.
2. Taking eight pixels of a center pixel and a neighborhood as image window dividing standards, taking a gray value as measurement data, calculating the gray value difference between the center and the neighborhood to generate a local vector and calculate the similarity, constructing a patrol management channel to perform targeted independent decision analysis of an abnormal grade and a management scheme, performing refined analysis processing on an abnormal infrared image, performing targeted management decision on the basis of ensuring analysis completeness, ensuring timely effectiveness of operation and maintenance,
Example two
Based on the same inventive concept as the intelligent patrol management method of the digital power distribution network in the foregoing embodiment, as shown in fig. 4, the present application provides an intelligent patrol management system of the digital power distribution network, where the system includes:
the image acquisition module 11 is used for acquiring a plurality of infrared images of a plurality of inspection nodes in the power distribution network for inspection;
The channel construction module 12 is configured to construct a patrol identification channel, where the patrol identification channel includes a plurality of patrol identification paths corresponding to the plurality of patrol nodes, and the plurality of patrol identification paths are constructed based on a twin network;
The image judging module 13 is configured to preprocess the plurality of infrared images, and respectively input the plurality of normal images of the plurality of inspection nodes in a normal running state into the plurality of inspection recognition paths to obtain a plurality of judging results, where each judging result includes a judging result of whether the preprocessed infrared images are consistent with the normal images;
The local vector acquisition module 14 is configured to perform graying processing on the target infrared images and the target normal images with no judgment result, and calculate local vectors of a plurality of windows in the target infrared images and the target normal images, so as to obtain a plurality of first local vector sets and a plurality of second local vector sets;
the similarity calculating module 15 is configured to calculate, according to the first local vector sets and the second local vector sets, similarities between the target infrared images and the target normal images, and obtain target similarities;
And the decision management module 16 is used for making a node anomaly level decision and a management scheme decision according to the plurality of target similarities, obtaining a plurality of target node anomalies and a plurality of target node management schemes, and carrying out inspection management on a plurality of target inspection nodes.
Further, the channel construction module 12 further includes:
The infrared image acquisition module is used for acquiring historical inspection infrared image data of the inspection nodes and acquiring the normal images and the historical infrared image sets;
The inspection recognition path construction module is used for constructing the plurality of inspection recognition paths based on a twin network, wherein each inspection recognition path comprises two image convolution processing branches with the same network structure, the input data of one image convolution processing branch is a normal image, and the input data of the other image convolution processing branch is an infrared image;
And the channel training module is used for carrying out synchronous supervision training on two image convolution processing branches in the inspection recognition paths by adopting the normal images and the historical infrared image sets to obtain the inspection recognition channel.
Further, the channel training module further includes:
the image convolution processing branch construction module is used for constructing two image convolution processing branches in the plurality of inspection recognition paths based on a convolution neural network;
the consistency judging module is used for judging whether the historical infrared images are consistent with the corresponding normal images according to the normal images and the historical infrared image sets respectively to obtain a plurality of historical judging result sets;
The monitoring training module is used for adopting the plurality of normal images, the plurality of historical infrared image sets and the plurality of historical judgment result sets to respectively carry out synchronous monitoring training on two image convolution processing branches in the plurality of inspection recognition paths until the two image convolution processing branches meet preset convergence conditions;
And the verification test module is used for verifying and testing the plurality of inspection recognition paths, and obtaining the constructed plurality of inspection recognition paths when the model loss is smaller than the loss threshold value.
Further, the local vector acquisition module 14 further includes:
The image window dividing module is used for dividing the target infrared images and the target normal images after the graying treatment to obtain a plurality of first window sets and a plurality of second window sets, and each window comprises nine gray values;
The local vector judging module is used for calculating the difference value between the gray value of the edge point and the gray value of the middle point in each window in the first window sets and the second window sets, and judging and obtaining the first local vector sets and the second local vector sets.
Further, the similarity calculating module 15 further includes:
The local vector combination acquisition module is used for carrying out division matching according to the first local vector sets and the second local vector sets and corresponding first windows and second windows to obtain a plurality of local vector combinations;
the target similarity acquisition module is used for calculating the difference values of the first local vector and the second local vector in the plurality of local vector combinations to obtain a plurality of difference value sets, and summing the absolute values of the difference values in the plurality of difference value sets to obtain the plurality of target similarities.
Further, the decision management module 16 further includes:
The system comprises a sample acquisition module, a sample management module and a sample analysis module, wherein the sample acquisition module is used for acquiring and processing historical inspection data of a power distribution network, acquiring a plurality of sample target similarities, and acquiring a plurality of sample node abnormal grades and a plurality of sample management schemes;
The inspection management channel acquisition module is used for constructing an inspection analysis branch and a management decision branch based on the similarity of the plurality of sample targets, the abnormal grades of the plurality of sample nodes and the plurality of sample management schemes, and integrating to obtain an inspection management channel;
And the management information acquisition module is used for inputting the plurality of target similarities into the inspection management channel to obtain the plurality of target node anomalies and a plurality of target node management schemes.
Further, the patrol management channel acquisition module further includes:
the decision node construction module is used for constructing a multi-layer inspection decision node and a multi-layer management decision node according to the decision tree based on the plurality of sample target similarities by taking the target similarities as decision characteristics;
the branch acquisition module is used for acquiring the inspection analysis branch and the management decision branch by adopting the abnormal grades of the plurality of sample nodes and the plurality of sample management schemes as decision results of the multi-layer inspection decision nodes and the multi-layer management decision nodes;
and the decision node merging module is used for merging the decision nodes in the inspection analysis branch and the management decision branch to obtain the inspection management channel.
Through the foregoing detailed description of an intelligent patrol management method for a digital power distribution network, those skilled in the art can clearly know an intelligent patrol management method and system for a digital power distribution network in this embodiment, and for the device disclosed in the embodiment, the description is relatively simple because it corresponds to the method disclosed in the embodiment, and relevant places refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (6)
1. An intelligent patrol management method for a digital power distribution network is characterized by comprising the following steps:
collecting a plurality of infrared images of a plurality of inspection nodes in an inspection power distribution network;
constructing an inspection identification channel, wherein the inspection identification channel comprises a plurality of inspection identification paths corresponding to the plurality of inspection nodes, and the plurality of inspection identification paths are constructed based on a twin network;
Preprocessing the plurality of infrared images, combining a plurality of normal images of the plurality of inspection nodes in a normal running state, respectively inputting the plurality of inspection recognition paths to obtain a plurality of judgment results, wherein each judgment result comprises a judgment result of whether the preprocessed infrared images are consistent with the normal images;
Carrying out graying treatment on a plurality of target infrared images and a plurality of target normal images which are judged to be negative, and calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images to obtain a plurality of first local vector sets and a plurality of second local vector sets;
Calculating the similarity of the plurality of target infrared images and the plurality of target normal images according to the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarity;
according to the target similarity, making a node anomaly level decision and a management scheme decision to obtain a plurality of target node anomalies and a plurality of target node management schemes, and carrying out inspection management on a plurality of target inspection nodes;
wherein, establish the inspection discernment passageway, include:
Collecting historical inspection infrared image data of the plurality of inspection nodes to obtain the plurality of normal images and a plurality of historical infrared image sets;
Constructing a plurality of patrol recognition paths based on a twin network, wherein each patrol recognition path comprises two image convolution processing branches with the same network structure, the input data of one image convolution processing branch is a normal image, and the input data of the other image convolution processing branch is an infrared image;
Adopting the normal images and the historical infrared image sets to carry out synchronous supervision training on two image convolution processing branches in the inspection recognition paths to obtain the inspection recognition channel;
the method for performing synchronous supervision training on two image convolution processing branches in the inspection recognition paths by adopting the normal images and the historical infrared image sets comprises the following steps:
constructing two image convolution processing branches in the plurality of patrol recognition paths based on a convolution neural network;
judging whether the historical infrared images are consistent with the corresponding normal images according to the normal images and the historical infrared image sets respectively to obtain a plurality of historical judgment result sets;
Adopting the plurality of normal images, the plurality of historical infrared image sets and the plurality of historical judgment result sets to respectively carry out synchronous supervision training on two image convolution processing branches in the plurality of patrol recognition paths until the two image convolution processing branches meet preset convergence conditions;
And verifying and testing the plurality of inspection recognition paths, and obtaining the constructed plurality of inspection recognition paths when the model loss is smaller than the loss threshold value.
2. The method of claim 1, wherein computing local vectors for a plurality of windows within the plurality of target infrared images and the plurality of target normal images comprises:
Dividing the target infrared images and target normal images after graying treatment to obtain a plurality of first window sets and a plurality of second window sets, wherein each window comprises nine gray values;
and in each window in the first window sets and the second window sets, calculating the difference value between the gray value of the edge point and the gray value of the middle point, and judging to obtain the first local vector sets and the second local vector sets.
3. The method of claim 2, wherein computing the similarity of the plurality of target infrared images and the plurality of target normal images from the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarities comprises:
According to the plurality of first local vector sets and the plurality of second local vector sets, carrying out division matching according to the corresponding first windows and second windows to obtain a plurality of local vector combinations;
And calculating the difference values of the first local vector and the second local vector in the plurality of local vector combinations to obtain a plurality of difference value sets, and summing the absolute values of the difference values in the plurality of difference value sets to obtain the plurality of target similarities.
4. The method of claim 1, wherein making node anomaly level decisions and management scheme decisions based on the plurality of target similarities comprises:
Acquiring and processing historical inspection data of the power distribution network, acquiring a plurality of sample target similarities, and acquiring a plurality of sample node abnormal grades and a plurality of sample management schemes;
Constructing a patrol analysis branch and a management decision branch based on the similarity of the plurality of sample targets, the abnormal grades of the plurality of sample nodes and the plurality of sample management schemes, and integrating to obtain a patrol management channel;
And inputting the plurality of target similarities into the inspection management channel to obtain the plurality of target node anomalies and a plurality of target node management schemes.
5. The method of claim 4, wherein constructing a patrol analysis branch and a management decision branch based on the plurality of sample target similarities, a plurality of sample node anomaly levels, and a plurality of sample management schemes comprises:
Taking the target similarity as a decision feature, and constructing a multi-layer inspection decision node and a multi-layer management decision node according to a decision tree based on the plurality of sample target similarities;
adopting the abnormal grades of the plurality of sample nodes and a plurality of sample management schemes as decision results of the multi-layer inspection decision nodes and the multi-layer management decision nodes to obtain inspection analysis branches and management decision branches;
And merging decision nodes in the inspection analysis branch and the management decision branch to obtain the inspection management channel.
6. An intelligent patrol management system for a digital power distribution network, the system comprising:
the image acquisition module is used for acquiring a plurality of infrared images of a plurality of inspection nodes in the power distribution network for inspection;
The channel construction module is used for constructing a patrol identification channel, wherein the patrol identification channel comprises a plurality of patrol identification paths corresponding to the plurality of patrol nodes, and the plurality of patrol identification paths are constructed based on a twin network;
The image judging module is used for preprocessing the plurality of infrared images, combining a plurality of normal images of the plurality of inspection nodes in a normal running state, respectively inputting the images into the plurality of inspection recognition paths to obtain a plurality of judging results, wherein each judging result comprises a judging result of whether the preprocessed infrared images are consistent with the normal images or not;
The local vector acquisition module is used for carrying out graying treatment on a plurality of target infrared images and a plurality of target normal images which are judged to be negative, calculating local vectors of a plurality of windows in the plurality of target infrared images and the plurality of target normal images, and obtaining a plurality of first local vector sets and a plurality of second local vector sets;
The similarity calculation module is used for calculating the similarity of the plurality of target infrared images and the plurality of target normal images according to the plurality of first local vector sets and the plurality of second local vector sets to obtain a plurality of target similarities;
The decision management module is used for making node anomaly level decisions and management scheme decisions according to the plurality of target similarities, obtaining a plurality of target node anomalies and a plurality of target node management schemes and carrying out inspection management on a plurality of target inspection nodes;
wherein, the channel construction module further comprises:
The infrared image acquisition module is used for acquiring historical inspection infrared image data of the inspection nodes and acquiring the normal images and the historical infrared image sets;
The inspection recognition path construction module is used for constructing the plurality of inspection recognition paths based on a twin network, wherein each inspection recognition path comprises two image convolution processing branches with the same network structure, the input data of one image convolution processing branch is a normal image, and the input data of the other image convolution processing branch is an infrared image;
The channel training module is used for performing synchronous supervision training on two image convolution processing branches in the inspection recognition paths by adopting the normal images and the historical infrared image sets to obtain the inspection recognition channel;
wherein, the channel training module further comprises:
the image convolution processing branch construction module is used for constructing two image convolution processing branches in the plurality of inspection recognition paths based on a convolution neural network;
the consistency judging module is used for judging whether the historical infrared images are consistent with the corresponding normal images according to the normal images and the historical infrared image sets respectively to obtain a plurality of historical judging result sets;
The monitoring training module is used for adopting the plurality of normal images, the plurality of historical infrared image sets and the plurality of historical judgment result sets to respectively carry out synchronous monitoring training on two image convolution processing branches in the plurality of inspection recognition paths until the two image convolution processing branches meet preset convergence conditions;
And the verification test module is used for verifying and testing the plurality of inspection recognition paths, and obtaining the constructed plurality of inspection recognition paths when the model loss is smaller than the loss threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311651704.0A CN117575165B (en) | 2023-12-05 | 2023-12-05 | Intelligent patrol management method and system for digital power distribution network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311651704.0A CN117575165B (en) | 2023-12-05 | 2023-12-05 | Intelligent patrol management method and system for digital power distribution network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117575165A CN117575165A (en) | 2024-02-20 |
| CN117575165B true CN117575165B (en) | 2024-05-07 |
Family
ID=89895403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311651704.0A Active CN117575165B (en) | 2023-12-05 | 2023-12-05 | Intelligent patrol management method and system for digital power distribution network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117575165B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114694050A (en) * | 2022-03-09 | 2022-07-01 | 重庆邮电大学 | Power equipment running state detection method based on infrared image |
| CN116203970A (en) * | 2023-04-27 | 2023-06-02 | 泰坦(天津)能源技术有限公司 | Intelligent obstacle avoidance method and system for inspection robot |
| CN116345685A (en) * | 2023-03-02 | 2023-06-27 | 国网浙江省电力有限公司杭州供电公司 | Power distribution network digital inspection method based on 3D virtual reality technology |
| CN116797977A (en) * | 2023-06-28 | 2023-09-22 | 南京天创电子技术有限公司 | Method and device for identifying dynamic target of inspection robot and measuring temperature and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114219374B (en) * | 2022-02-21 | 2022-05-06 | 济南法诺商贸有限公司 | Big data analysis decision system and method based on block chain |
-
2023
- 2023-12-05 CN CN202311651704.0A patent/CN117575165B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114694050A (en) * | 2022-03-09 | 2022-07-01 | 重庆邮电大学 | Power equipment running state detection method based on infrared image |
| CN116345685A (en) * | 2023-03-02 | 2023-06-27 | 国网浙江省电力有限公司杭州供电公司 | Power distribution network digital inspection method based on 3D virtual reality technology |
| CN116203970A (en) * | 2023-04-27 | 2023-06-02 | 泰坦(天津)能源技术有限公司 | Intelligent obstacle avoidance method and system for inspection robot |
| CN116797977A (en) * | 2023-06-28 | 2023-09-22 | 南京天创电子技术有限公司 | Method and device for identifying dynamic target of inspection robot and measuring temperature and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117575165A (en) | 2024-02-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110598736B (en) | Power equipment infrared image fault positioning, identifying and predicting method | |
| Eisenbach et al. | How to get pavement distress detection ready for deep learning? A systematic approach | |
| CN113205176B (en) | Method, device and equipment for training defect classification detection model and storage medium | |
| CN108680833B (en) | Composite insulator defect detection system based on unmanned aerial vehicle | |
| CN108711148B (en) | Tire defect intelligent detection method based on deep learning | |
| CN112949715A (en) | SVM (support vector machine) -based rail transit fault diagnosis method | |
| CN110728300A (en) | Method and system for identifying fault type based on turnout action current curve | |
| CN112036464A (en) | Insulator infrared image fault detection method based on YOLOv3-tiny algorithm | |
| CN111831430A (en) | Electrical equipment defect identification system based on edge calculation | |
| CN115719475B (en) | Three-stage trackside equipment fault automatic detection method based on deep learning | |
| CN112115770A (en) | Method and system for identifying autonomous inspection defects of unmanned aerial vehicle of overhead line | |
| CN112508911A (en) | Rail joint touch net suspension support component crack detection system based on inspection robot and detection method thereof | |
| CN116704733A (en) | Aging early warning method and system for aluminum alloy cable | |
| CN115619592A (en) | Photovoltaic string loss electric quantity calculation method and system | |
| CN117575165B (en) | Intelligent patrol management method and system for digital power distribution network | |
| CN113837994A (en) | Photovoltaic panel defect diagnosis method based on edge detection convolutional neural network | |
| CN112508946B (en) | Cable tunnel anomaly detection method based on antagonistic neural network | |
| CN116956189A (en) | Current abnormality detection system, method, electronic equipment and medium | |
| Shin et al. | Visualization for explanation of deep learning-based defect detection model using class activation map | |
| CN114863543B (en) | Feature updating face recognition method and system | |
| CN112598666B (en) | Cable tunnel anomaly detection method based on convolutional neural network | |
| CN116596411B (en) | Production safety evaluation method and system combining two-ticket detection | |
| CN117152149B (en) | Edge smoothness detection method and system for wafer dicing | |
| CN113469995B (en) | Transformer substation equipment thermal fault diagnosis method and system | |
| CN118038230A (en) | Power distribution network defect identification system and method based on multi-algorithm fusion |
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 |