CN115220479A - Dynamic and static cooperative power transmission line refined inspection method and system - Google Patents
Dynamic and static cooperative power transmission line refined inspection method and system Download PDFInfo
- Publication number
- CN115220479A CN115220479A CN202211140267.1A CN202211140267A CN115220479A CN 115220479 A CN115220479 A CN 115220479A CN 202211140267 A CN202211140267 A CN 202211140267A CN 115220479 A CN115220479 A CN 115220479A
- Authority
- CN
- China
- Prior art keywords
- fault
- inspection
- static
- information
- aerial vehicle
- 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
- 238000007689 inspection Methods 0.000 title claims abstract description 154
- 230000003068 static effect Effects 0.000 title claims abstract description 137
- 230000005540 biological transmission Effects 0.000 title claims abstract description 82
- 238000000034 method Methods 0.000 title claims abstract description 69
- 230000000007 visual effect Effects 0.000 claims abstract description 44
- 238000004891 communication Methods 0.000 claims abstract description 32
- 238000013145 classification model Methods 0.000 claims abstract description 19
- 238000009826 distribution Methods 0.000 claims description 60
- 230000006870 function Effects 0.000 claims description 39
- 238000012549 training Methods 0.000 claims description 35
- 230000007246 mechanism Effects 0.000 claims description 34
- 230000008569 process Effects 0.000 claims description 29
- 238000001514 detection method Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 12
- 239000000779 smoke Substances 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000007613 environmental effect Effects 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 230000004888 barrier function Effects 0.000 claims description 4
- 235000005770 birds nest Nutrition 0.000 claims description 4
- 244000240635 birds nest Species 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 239000012212 insulator Substances 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 235000005765 wild carrot Nutrition 0.000 claims description 4
- 230000006872 improvement Effects 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 238000011835 investigation Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000013024 troubleshooting Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000013500 data storage Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention belongs to the technical field of routing inspection, and provides a dynamic and static cooperative power transmission line refined routing inspection method and system. Which comprises the following steps: the fixed acquisition equipment acquires the returned static information; the multi-layer sensing machine integrates multi-class static information to evaluate the fault degree; patrol and examine unmanned aerial vehicle and carry out the meticulous inspection and upload multi-angle image information to transmission line: if the non-emergency fault occurs, calling an inspection unmanned aerial vehicle to perform fine inspection, acquiring inspection information of the inspection unmanned aerial vehicle, and transmitting the inspection information to a target data control center through a communication module; and fusing a fault classification model of the image information of the unmanned aerial vehicle with multiple visual angles and a fixed visual angle.
Description
Technical Field
The invention relates to a power transmission line routing inspection method and a power transmission line routing inspection system, in particular to a power transmission line fine routing inspection method and a power transmission line fine routing inspection system with dynamic and static coordination.
Background
The inspection of the power transmission line is a basic work for effectively ensuring the safety of the power transmission line and equipment thereof, the running condition of the line and the change of the surrounding environment are mastered through the inspection of the power transmission and distribution line, and the equipment defects and potential safety hazards are found in time, so that the safety of the power transmission and distribution line and the stability of a power system are ensured. Along with the development of the automatic patrol of the unmanned aerial vehicle of the power transmission line, the automatic patrol technology of the unmanned aerial vehicle is more mature, and in the development of the automatic patrol process, the patrol efficiency is greatly improved by the method for patrolling by the unmanned aerial vehicle. Unmanned aerial vehicle carries on all kinds of visible light, infrared or laser equipment and carries out and patrols and examines the task, can detect comprehensively and master the safe condition of electric power tower, patrols and examines through unmanned aerial vehicle, and reduce cost and intensity of labour can be reduced by a wide margin, guarantee transmission line's safe and reliable's operation.
Because transmission line is very huge, spreads all over, consequently, transmission line often adopts sectional type management, sets up unmanned aerial vehicle respectively for each section. At present, the unmanned aerial vehicle inspection mode that relatively extensively adopts is that fixed airline patrols and examines, and unmanned aerial vehicle patrols and examines along preset airline in each district section promptly, and the advantage of this kind of mode of patrolling and examining is patrolled and examined efficiently, flight factor of safety is high, nevertheless to some component failures that exist in the transmission line, for example insulator trouble, equalizer ring trouble, transmission line damage etc. are difficult to realize the fault detection that becomes more meticulous.
Disclosure of Invention
The invention aims to provide a power transmission line refined inspection method and system with dynamic and static cooperation, which realize the refined inspection of a power system in a large-range coverage through the cooperative cooperation of dynamic and static detection, improve the detection refinement by adopting the sectional detection of static and dynamic data, and save unnecessary manpower and calculation resources.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a fine routing inspection method for a power transmission line with dynamic and static coordination comprises the following steps:
s1: the fixed acquisition equipment acquires the returned static information: receiving static information acquired by fixed acquisition equipment of a power transmission line of a target inspection section of a power system, and uploading the static information so as to perform preliminary state judgment in the next step;
s2: the multilayer perceptron integrates multi-class static information to evaluate the fault degree: according to static information acquired by fixed acquisition equipment, the current equipment state is evaluated by fusing multi-class data through a multilayer perceptron, and whether the routing inspection unmanned aerial vehicle is called for fine routing inspection is determined according to an evaluation result;
s3: patrol and examine unmanned aerial vehicle and carry out the meticulous inspection and upload multi-angle image information to transmission line: if non-emergency faults occur, calling the inspection unmanned aerial vehicle to perform fine inspection, acquiring inspection information of the inspection unmanned aerial vehicle, and transmitting the inspection information to a target data control center through a communication module;
s4, fusing fault classification models of multi-view and fixed-view image information of the unmanned aerial vehicle: the method includes the steps of fusing static image information of a fixed camera and multi-angle image information of fine routing inspection of an unmanned aerial vehicle, using a multi-head attention mechanism model based on image features to classify image faults, and introducing a confidence gating mechanism and information divergence constraint to achieve multi-view learning.
Preferably, the process of the S1 step further includes:
s11: the method comprises the steps that static information under a power transmission and transformation scene is collected by fixed collection equipment located on a power transmission line; the control center of the machine nest is communicated with the fixed acquisition equipment through a communication module, and receives and stores the static information acquired by the fixed acquisition equipment;
s12: the control center of the cell is used as the core part of the system to control the transmission of information and the interaction among all modules; the collector communication module provides an information channel for data transmission of the fixed collection equipment and the control center of the machine nest, and ensures that the control center of the machine nest can receive various types of static information collected by the fixed collection equipment;
s13: the multi-type static information of the site is collected through the fixed collecting equipment of the power transmission line, and the multi-type static information is uploaded to the control center of the machine nest through the communication module, so that the running state of the power transmission line is preliminarily judged.
Preferably, the process of the S2 step further includes:
s21: the software part included in the step comprises a nest control center and a static judgment module, wherein the nest static judgment module based on static information is carried on the control center of the nest, receives the static information acquired by fixed acquisition equipment, preliminarily judges the state of target routing inspection area equipment through a multilayer sensor model according to the static information, determines whether to call a routing inspection unmanned aerial vehicle for fine routing inspection according to an evaluation result, and generates trigger information if a non-emergency fault occurs; the nest control center receives the trigger information generated by the static judgment module and further sends a target instruction to the nest control module to schedule the inspection unmanned aerial vehicle for fine inspection;
s22: the machine nest static judging module adopts a multilayer perceptron model to synthesize multi-class static information to judge the current equipment state, and the method comprises the following processes:
the method comprises the steps of constructing a multi-class information equipment state judgment data set through static information acquired by fixed acquisition equipment and fault class historical records corresponding to the static information, constructing data set data through the static information acquired by the fixed acquisition equipment and integrating corresponding equipment state information, marking corresponding fault classes by adopting three classes of no fault, emergent fault and non-emergent fault, namely equipment state labels corresponding to the multi-class static information at the same time, training a multilayer perceptron model through the data set, integrating four kinds of static information including temperature, humidity, smoke concentration and ambient gas concentration acquired by the fixed acquisition equipment to serve as input of the multilayer perceptron model, judging faults through the multilayer perceptron model, outputting three fault classes of no fault, emergent fault and non-emergent fault, and defining loss function
The training process is constrained by the constraints of the training process,
the formula is defined as follows:
wherein the content of the first and second substances,
a label that indicates the level of the fault,
a failure level indicative of the predicted output is displayed,
represents the total number of samples;
training the multilayer sensor model based on the training step, deploying the trained multilayer sensor model in a nest static judgment module, and performing primary judgment on the equipment state;
s23: and respectively taking measures to deal with the fault levels output by the multilayer perceptron model deployed in the nest static judgment module: aiming at the emergency fault, as the emergency fault needs to be processed and maintained in time, related workers need to be sent to the site for further investigation and processing; aiming at non-emergency faults, the inspection unmanned aerial vehicle is adopted for further fine inspection, and trigger information is generated by the nest static judgment module and is sent to the nest control center, so that the inspection unmanned aerial vehicle is further dispatched for fine inspection;
s24: judging whether a fault occurs according to a nest static judging module under a nest control center, and if the fault occurs, dispatching related workers to the site in time for further troubleshooting; if non-emergency faults occur, the trigger information is generated by the nest static judging module and is transmitted back to the nest control center so as to call the inspection unmanned aerial vehicle for fine inspection.
Preferably, the process of the S3 step further includes:
s31: the hardware part included in the step comprises an inspection unmanned aerial vehicle, and multi-angle and high-precision image acquisition of all equipment and parts around the power transmission line and the tower is realized;
s32: the software part included in the step comprises a nest control module, a nest control center, an unmanned aerial vehicle communication module and a data storage center, wherein the nest control module receives trigger information issued by the nest control center and schedules the inspection unmanned aerial vehicle to carry out fine inspection; the unmanned aerial vehicle communication module is used for receiving the instruction from the nest control center and returning the patrol inspection information to the nest control center; the data storage center is built based on a database technology, receives all fine routing inspection information of the unmanned aerial vehicle from the nest control center, and stores the generated data file and the generated flight log into a database.
Preferably, the process of step S4 further includes:
s41: a multi-view image fault classification data set is constructed based on a database obtained by a fine routing inspection result of an unmanned aerial vehicle, multi-angle image acquisition is performed on each part around a power transmission line and a tower through a close distance of the routing inspection unmanned aerial vehicle, then a method of artificial comprehensive judgment is adopted based on large-scale static image information returned by a high-definition camera carried by the routing inspection unmanned aerial vehicle to mark fault types of each part, and the known fault types are defined as follows: the method comprises the following steps that an insulator fault, a grading ring fault, a vibration damper fault, a tower structure fault, a power transmission line damage, a power transmission line approaching barrier and bird nesting are carried out, a sample pair of a data set consists of multi-view images and fault type marking results, specifically, the multi-view images comprise N unmanned aerial vehicle inspection view images, and the multi-view images are obtained by carrying out multi-angle image acquisition on the power transmission line or the tower and other components; the multi-view image also comprises large-scale static image information collected by a camera fixed on the tower;
s42: image fault classification is carried out by using a multi-head attention mechanism model based on image characteristics, and a confidence gating mechanism and information divergence constraint are introduced for improvement, so that multi-view learning is realized, and the fault classification accuracy is improved;
s43: the method comprises the steps of using a loss function to constrain the training process of a fault classification model, wherein the loss function comprises two parts, the first part is to carry out information divergence-based constraint on probability distribution obtained by carrying out pre-training model feature extraction, multi-head attention mechanism model coding and multi-layer perceptron model mapping on N unmanned aerial vehicle inspection visual angle images and fixed visual angle images and screening through a confidence gating mechanism, so that N +1 probability distributions of different visual angles are kept consistent as much as possible, compatible complementation between image information of different visual angles is realized, and the loss of the part is definedA loss function of
The formula is shown as (3),
wherein N represents the number of the unmanned aerial vehicle inspection visual angles, x and y represent two different visual angles respectively,
and
respectively representing the probability distribution of the corresponding visual angles, wherein the value of the loss function is equal to the sum of information divergence between every two visual angles of the N +1 visual angles; the second part is that the probability distribution of the N +1 visual angles is voted by adopting the idea of integrated learning, the result of each visual angle prediction has the same weight, the prediction category with the highest vote number is taken as the integral fault classification result and is subjected to loss calculation with the result of artificial comprehensive research and judgment marking, and the loss function of the part is defined as
The formula is shown as (4),
wherein
Representing the probability distribution corresponding to the fault manual labeling category,
and representing the class probability distribution of the fault obtained through prediction of a classification model, and calculating the geometric distance between the two as a value of a loss function. Combining the loss functions of the two parts as a function of the resultThe total loss function formula of the training loss function of the barrier classification model is shown as (5),
wherein
For balancing the ratio of the two partial loss functions,
the value range of (1) is between 0 and 1, the reasonable value can prevent the model from excessively fitting the optimization of a single loss function, and the training of the model should take into account the optimization of the two loss functions.
Preferably, the specific process is as follows:
s421, calling a ResNet18 residual network pre-training model pre-trained on a large-scale public data set, keeping parameters of ResNet18 unchanged during training, taking the ResNet18 as a universal image feature extractor, performing feature extraction on a multi-view image, outputting N +1 feature maps with the size of 512 × 7, and expanding the feature maps into one-dimensional feature vectors;
s422, respectively encoding the one-dimensional feature vectors corresponding to the view images by using a multi-head attention mechanism model to obtain N +1 one-dimensional high-grade feature vectors with the length of 1024;
s423, respectively inputting the N +1 one-dimensional high-level feature vectors obtained after the coding in the S422 into a multi-layer perceptron classifier, and outputting the predicted fault type probability distribution corresponding to the N +1 view images;
s424, introducing a confidence gating mechanism to the N +1 probability distributions in S423, calculating a maximum probability category corresponding to the probability distributions by using an argmax function, knowing that the probability value of each category is between 0 and 1, and if the probability value corresponding to the maximum probability category of the distribution is smaller than a gating threshold β, discarding a probability distribution result predicted by the view image, where the specific calculation result is shown in formula (2):
where p represents the probability distribution of a certain view prediction, x is the class corresponding to the maximum probability in the probability distribution,
a probability value for the category; for example, if the gating threshold β is set to 0.8, and the probability value of the xth class in the probability distribution of a certain view prediction is 0.75 at most, the requirement of the confidence gating mechanism is not satisfied, and the probability distribution result of the view image prediction is discarded.
S425, calculating information divergence between every two of the plurality of probability distributions subjected to the confidence gating mechanism in the step S424, and constraining distribution consistency of the probability distributions so as to enable the prediction probability distributions of the images of all the view angles to be similar as much as possible, wherein the prediction fault types of the model are determined by the image information of the plurality of view angles together;
s426, voting is carried out on the fault categories obtained by predicting the images of the plurality of visual angles by using the idea of ensemble learning, the final result of the prediction of the fault classification model is obtained by comprehensive study and judgment, the fault categories belong to the known fault types defined in the S41, the multi-head attention mechanism model receives image characteristics extracted by a pre-training model ResNet18 from different visual angles, the image characteristics comprise N unmanned aerial vehicle inspection visual angles and visual angles of fixed cameras, multiple coding is carried out, finally, the fused characteristics are mapped to a lower dimensionality by using a multilayer sensing machine, and the probability distribution of the fault categories corresponding to the images is output.
A system for realizing the fine routing inspection method of the power transmission line with the dynamic and static coordination is characterized in that: comprises a fixed acquisition device, a dynamic detection module, a processing module and a judgment module,
the system comprises a fixed acquisition device, a control center and a control center, wherein the fixed acquisition device comprises a temperature sensor, a humidity sensor, a smoke concentration sensor, an environmental gas sensor and a fixed camera, the sensors and the fixed camera are arranged on corresponding monitoring nodes according to the monitoring requirements of a power transmission line, and the fixed acquisition device transmits acquired routing inspection information to the control center through a communication module;
the dynamic detection module comprises an inspection unmanned aerial vehicle inspection module, and the unmanned aerial vehicle receives a control center signal through an unmanned aerial vehicle control module;
the processing module comprises a control center of the machine nest, a static judging module and a communication module, the communication module is connected with the control center of the machine nest, the static judging module is communicated with the control center, the communication module is connected with the fixed acquisition equipment and the dynamic detection module, the static judging module is a software module used for judging faults and calling the unmanned aerial vehicle and subsequent data processing, and the software module is carried on the control center of the machine nest.
The invention has the advantages that:
according to the method, comprehensive cooperation is carried out on multi-class static information of fixed acquisition equipment in the power transmission line and multi-view dynamic image information of fine routing inspection of the unmanned aerial vehicle, the multi-class static information of the fixed acquisition equipment guides the routing inspection unmanned aerial vehicle to carry out fine routing inspection, fault class classification is carried out on the multi-view dynamic image information acquired by the routing inspection unmanned aerial vehicle and large-scale static image information returned by the fixed camera, practice shows that the fine routing inspection method of the power transmission line with dynamic and static cooperation greatly improves the fine degree of routing inspection of the power system, excavation is carried out aiming at equipment state information contained in the static information acquired by the fixed acquisition equipment, the possibly occurring fault class is effectively judged, and a key reference is provided for the power system to efficiently process faults of various classes; in addition, fault classification carried out by integrating multi-view dynamic image information and large-scale static image information comprehensively judges faults from multiple views more accurately, and helps an electric power system to know fault degree in a fine-grained manner, so that more reasonable and accurate processing is carried out. Meanwhile, the power transmission line refined inspection system based on the dynamic and static cooperation constructed by the method has great reference and instructive significance for fault detection and fault processing work in daily fault maintenance of the power transmission line, and plays an extremely important role in maintaining the stability and the high efficiency of the power transmission line.
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.
FIG. 1 is a flow chart of a method of an embodiment of the present invention
FIG. 2 is a block diagram of a system according to an embodiment of the present invention
FIG. 3 is a diagram of a static information fault classification model according to an embodiment of the present invention
Fig. 4 is a model diagram of multi-view information fault classification according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
Detailed description of the steps:
firstly, static information under a power transmission and transformation scene is collected by fixed collection equipment positioned on a power transmission line and is uploaded to a cell control center through a communication module; the control center forwards the static information data to a static judgment module, and the static judgment module evaluates the current equipment state by fusing multi-class static information through a multi-layer perceptron and outputs the current equipment state; if the static judgment module judges that the non-emergency fault occurs, generating unmanned aerial vehicle polling trigger information and sending the information to a nest control center; the nest control center receives the unmanned aerial vehicle polling trigger information and forwards the information to the control module, and the control module further schedules the unmanned aerial vehicle for fine polling; the inspection unmanned aerial vehicle performs fine inspection according to a set line, acquires equipment images from multiple angles, and uploads the image information to the nest control center through the communication module; the nest control center uploads the multi-angle image information of the inspection unmanned aerial vehicle and the static image information of the fixed camera to the data storage module for storage, and the data storage module forwards the multi-angle image information and the static image information to the fault classification model for further fault detection; the fault classification model fuses static image information of a fixed camera and multi-angle image information of fine routing inspection of the unmanned aerial vehicle, image fault classification is carried out by using a multi-head attention mechanism model which introduces a confidence gating mechanism and information divergence constraint and is based on image characteristics, and finally detected image fault categories are output.
The technical problems to be solved in practice by the invention are as follows:
although the traditional fixed route routing inspection has high inspection speed and long inspection distance, on the premise that the fault type is not determined in advance, a corresponding sensor cannot be carried for the specific fault type to carry out refined inspection, the detection capability of multiple types of faults is limited, the single inspection efficiency is low, and the high-precision and high-efficiency detection on various types of faults possibly occurring on equipment along the route cannot be realized, so that a novel refined unmanned aerial vehicle routing inspection mode needs to be provided;
the monitoring camera of fixed position has that visual angle is little, image resolution ratio low grade shortcoming, and some fault types are difficult to discover through single angle moreover to be competent in multiple fault type's detection task, consequently need to synthesize the large-scale static image information who patrols and examines the multi-view dynamic image information that unmanned aerial vehicle gathered and fixed camera passback and classify the fault classification.
Referring to fig. 1, a fine routing inspection method for a power transmission line with dynamic and static coordination includes the following steps:
s1: the fixed acquisition equipment acquires the returned static information: receiving static information acquired by fixed acquisition equipment of a power transmission line in a target inspection section of a power system, and uploading the static information so as to perform preliminary state judgment in the next step;
s2: the multi-layer perception machine integrates multi-class static information to evaluate the fault degree: according to static information acquired by fixed acquisition equipment, the current equipment state is evaluated by fusing multi-class data through a multilayer perceptron, and whether the routing inspection unmanned aerial vehicle is called for fine routing inspection is determined according to an evaluation result;
s3: patrol and examine unmanned aerial vehicle and carry out the meticulous patrol and examine and upload multi-angle image information to transmission line: if non-emergency faults occur, calling the inspection unmanned aerial vehicle to perform fine inspection, acquiring inspection information of the inspection unmanned aerial vehicle, and transmitting the inspection information to a target data control center through a communication module;
s4, fusing fault classification models of multi-view and fixed-view image information of the unmanned aerial vehicle: the static image information of a fixed camera and the multi-angle image information of the fine routing inspection of the unmanned aerial vehicle are fused, a multi-head attention mechanism model based on image characteristics is used for carrying out image fault classification, and a confidence gating mechanism and information divergence constraint are introduced to realize multi-view learning.
In this embodiment, the process of step S1 further includes:
s11: the static information under the power transmission and transformation scene is collected by fixed collection equipment positioned on a power transmission line, wherein the fixed collection equipment comprises a temperature sensor, a humidity sensor, a smoke concentration sensor, an environmental gas sensor and the like; the control center of the machine nest is communicated with the fixed acquisition equipment through a communication module, and receives and stores the static information acquired by the fixed acquisition equipment;
s12: the control center of the machine nest is used as the core part of the system, and controls the transmission of information and the interaction among all modules; the collector communication module provides an information channel for data transmission of the fixed collection equipment and the control center of the machine nest, and ensures that the control center of the machine nest can receive various types of static information collected by the fixed collection equipment;
s13: the method comprises the steps of collecting on-site multi-type static information through fixed collection equipment of the power transmission line, and uploading the multi-type static information to a control center of a machine nest through a communication module to preliminarily judge the running state of the power transmission line.
In this embodiment, the process of the step S2 further includes:
s21: the software part included in the step comprises a nest control center and a static judgment module, wherein the nest static judgment module based on static information is carried on the control center of the nest, receives static information such as temperature, humidity, smoke concentration, environmental gas concentration and the like collected by fixed collection equipment, preliminarily judges the state of equipment in a target routing inspection area according to the static information through a multilayer sensor model, determines whether to call a routing inspection unmanned aerial vehicle for fine routing inspection according to an evaluation result, and generates trigger information if a non-emergency fault occurs; the airframe control center receives the trigger information generated by the static judgment module, and further issues a target instruction to the airframe control module to schedule the inspection unmanned aerial vehicle for fine inspection;
s22: the machine nest static state judgment module adopts a multilayer perceptron model to synthesize multi-class static information to judge the current equipment state, and comprises the following processes:
constructing a multi-class information equipment state judgment data set through static information acquired by fixed acquisition equipment and corresponding fault class historical records thereof, constructing data set data through the static information acquired by the fixed acquisition equipment and integrating the corresponding equipment state information thereof, wherein the corresponding fault class is marked by three classes of no fault, emergent fault and non-emergent fault, namely equipment state labels corresponding to the multi-class static information at the same time, training a multilayer perceptron model by using the data set, wherein four kinds of static information of temperature, humidity, smoke concentration and environmental gas concentration acquired by the fixed acquisition equipment are integrated to be used as the input of the multilayer perceptron model, fault judgment is carried out through the multilayer perceptron, three fault classes of no fault, emergent fault and non-emergent fault are output, an integral model diagram is shown in figure 3, and a loss function is defined
The training process is constrained by the constraints of the training process,
the formula is defined as follows:
wherein the content of the first and second substances,
a label that indicates the level of the fault,
a failure level indicating the predicted output is indicated,
representing the total number of samples;
training the multilayer sensor model based on the training step, deploying the trained multilayer sensor model in a nest static judgment module, and performing primary judgment on the equipment state;
s23: and respectively taking measures to deal with the fault levels output by the multilayer perceptron model deployed in the nest static judgment module: for an emergency fault, due to the fact that the emergency fault needs to be processed and maintained in time, relevant workers need to be sent to the site immediately for further investigation and processing, for example, faults such as power transmission line fire and the like; aiming at non-emergency faults, the inspection unmanned aerial vehicle is adopted for further fine inspection, and trigger information is generated by the nest static judgment module and is sent to the nest control center, so that the inspection unmanned aerial vehicle is further dispatched for fine inspection;
s24: judging whether a fault occurs according to a nest static judging module under a nest control center, and if the fault occurs, dispatching related workers to the site in time for further troubleshooting; if non-emergency faults occur, the trigger information is generated by the machine nest static judging module and is transmitted back to the machine nest control center so as to call the inspection unmanned aerial vehicle for fine inspection.
In this embodiment, the process of step S3 further includes:
s31: the hardware part included in the step comprises an inspection unmanned aerial vehicle, wherein the inspection unmanned aerial vehicle adopts a multi-rotor unmanned aerial vehicle, and carries a high-definition camera to realize multi-angle and high-precision image acquisition of various devices and parts around the power transmission line and the tower;
s32: the software part included in the step comprises a nest control module, a nest control center, an unmanned aerial vehicle communication module and a data storage center, wherein the nest control module receives trigger information issued by the nest control center and schedules the inspection unmanned aerial vehicle to carry out fine inspection; the unmanned aerial vehicle communication module is used for receiving the instruction from the nest control center and returning the routing inspection information to the nest control center; the data storage center is built based on a database technology, receives all the fine routing inspection information of the unmanned aerial vehicle from the nest control center, and stores the generated data file and the generated flight log into a database.
In this embodiment, the process of step S4 further includes:
s41: the method comprises the following steps of constructing a multi-view image fault classification data set based on a database obtained by a fine inspection result of an unmanned aerial vehicle, carrying out multi-angle image acquisition on each part around a power transmission line and a tower in a close range through the inspection unmanned aerial vehicle, and marking the fault type of each part by adopting a manual comprehensive judgment method based on large-scale static image information returned by a high-definition camera carried by the inspection unmanned aerial vehicle, wherein the defined known fault types are as follows: the method comprises the steps that insulator faults, grading ring faults, vibration damper faults, tower structure faults, transmission line damage, the fact that a transmission line is close to a barrier, birds nest and the like are achieved, a sample pair of a data set consists of multi-view images and fault type marking results, specifically, the multi-view images comprise N unmanned aerial vehicle inspection view images, and the multi-view images are obtained by conducting multi-angle image acquisition on the transmission line or the tower and other components, for example, aiming at the birds nest fault on the tower, vision shielding may exist when the single unmanned aerial vehicle view angle conducts image acquisition, and the fault information that birds nest exists can be completely captured through the multi-view image acquisition; the multi-view image also comprises large-scale static image information collected by a camera fixed on a tower, the camera adopts a fixed camera, has the advantages of wide view angle, large image scale and the like, can provide background information support for fault classification in multi-view image feature fusion, for example, an unmanned aerial vehicle collecting images at a short distance can not find the fault in time aiming at the fault condition that the power transmission line is too close to an obstacle, and the fixed camera can well detect that the power transmission line is too close to the obstacle on a large scale;
s42: using a multi-head attention mechanism model based on image features to classify image faults, and introducing a confidence gating mechanism and information divergence constraint to improve, thereby realizing multi-view learning and improving fault classification accuracy, as shown in fig. 4, the specific process is as follows:
s421, calling a ResNet18 residual network pre-training model pre-trained on a large-scale public data set, fixing parameters of ResNet18 during training, keeping the parameters unchanged, using the ResNet18 as a universal image feature extractor to perform feature extraction on a multi-view image, outputting N +1 feature graphs with the size of 512 × 7, and expanding the feature graphs into one-dimensional feature vectors;
s422, respectively encoding the one-dimensional feature vectors corresponding to the view images by using a multi-head attention mechanism model to obtain N +1 one-dimensional high-grade feature vectors with the length of 1024;
s423, respectively inputting the N +1 one-dimensional high-level feature vectors obtained after the coding in the S422 into a multi-layer perceptron classifier, and outputting the predicted fault type probability distribution corresponding to the N +1 view images;
s424, introducing a confidence gating mechanism to the N +1 probability distributions in S423, calculating a maximum probability category corresponding to the probability distributions by using an argmax function, knowing that the probability value of each category is between 0 and 1, and if the probability value corresponding to the maximum probability category of the distribution is smaller than a gating threshold β, discarding a probability distribution result predicted by the view image, where the specific calculation result is shown in formula (2):
where p represents the probability distribution of a certain view prediction, x is the class corresponding to the maximum probability in the probability distribution,
a probability value for the category; for example, if the gating threshold β is set to 0.8, and the probability value of the xth class in the probability distribution of a certain view prediction is 0.75 at most, the requirement of the confidence gating mechanism is not satisfied, and the probability distribution result of the view image prediction is discarded.
S425, calculating information divergence between every two of the probability distributions subjected to the confidence gating mechanism in the step S424, and constraining distribution consistency of the probability distributions so as to enable the probability distributions of image prediction of all the view angles to be similar as much as possible, and determining the prediction fault category of the model by the image information of the view angles;
s426, voting fault categories obtained by predicting a plurality of view images by using an integrated learning idea, comprehensively studying and judging to obtain a final result of prediction of a fault classification model, wherein the fault categories belong to known fault types defined in the S41, the multi-head attention mechanism model receives image characteristics extracted by a pre-training model ResNet18 from different views, including N unmanned aerial vehicle inspection view angles and view angles of a fixed camera, and carries out multi-time coding, and finally, a multi-layer sensing machine is used for mapping the fused characteristics to a lower dimensionality and outputting fault category probability distribution corresponding to a picture;
s43: the training process of the fault classification model is constrained using a loss function, which comprises two parts,
the first part is that probability distribution obtained by carrying out pre-training model feature extraction, multi-head attention mechanism model coding and multi-layer perceptron model mapping on N unmanned aerial vehicle inspection visual angle images and fixed visual angle images and screening through a confidence gating mechanism is subjected to information divergence-based constraint, so that the N +1 probability distributions of different visual angles are kept consistent as much as possible, compatibility and complementation between image information of different visual angles are realized, and the loss function of the part is defined as
The formula is shown as (3),
wherein N represents the number of the unmanned aerial vehicle inspection visual angles, x and y represent two different visual angles respectively,
and
respectively representing the probability distribution of the corresponding visual angles, wherein the value of the loss function is equal to the sum of information divergence between every two visual angles of the N +1 visual angles;
the second part is that the probability distribution of the N +1 visual angles is voted by adopting the idea of integrated learning, the result of each visual angle prediction has the same weight, the prediction category with the highest vote number is taken as the integral fault classification result and is subjected to loss calculation with the result of artificial comprehensive research and judgment marking, and the loss function of the part is defined as
The formula is shown as (4),
wherein
Representing the probability distribution corresponding to the fault manual labeling category,
and representing the class probability distribution of the fault obtained through prediction of a classification model, and calculating the geometric distance between the two as a value of a loss function. And (5) integrating the loss functions of the two parts as the training loss function of the fault classification model, wherein the total loss function formula is shown as (5),
wherein
For balancing the ratio of the two partial loss functions,
the value range of (1) is between 0 and 1, the reasonable value can prevent the model from excessively fitting the optimization of a single loss function, and the training of the model should take into account the optimization of the two loss functions.
As shown in fig. 2, a system for the fine routing inspection method of the power transmission line with dynamic and static coordination comprises a fixed acquisition device, a dynamic detection module, a processing module and a judgment module,
the system comprises a fixed acquisition device, a control center and a control center, wherein the fixed acquisition device comprises a temperature sensor, a humidity sensor, a smoke concentration sensor, an environmental gas sensor and a fixed camera, the sensors and the fixed camera are arranged on corresponding monitoring nodes according to the monitoring requirements of a power transmission line, and the fixed acquisition device transmits acquired routing inspection information to the control center through a communication module;
the dynamic detection module comprises an inspection unmanned aerial vehicle, and the inspection unmanned aerial vehicle receives a control center signal through an unmanned aerial vehicle control module;
the processing module comprises a control center of the machine nest and a communication module connected with the control center, the communication module is connected with the fixed acquisition equipment and the dynamic detection module, the static judgment module is a software module used for judging faults and calling the unmanned aerial vehicle and subsequent data processing, and the software module is loaded on the control center of the machine nest.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described above, or equivalents may be substituted for elements thereof. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. A fine routing inspection method for a power transmission line with dynamic and static coordination is characterized by comprising the following steps:
s1: the fixed acquisition equipment acquires the returned static information: receiving static information acquired by fixed acquisition equipment of a power transmission line of a target inspection section of a power system, and uploading the static information so as to perform preliminary state judgment in the next step;
s2: the multilayer perceptron integrates multi-class static information to evaluate the fault degree: according to static information acquired by fixed acquisition equipment, the current equipment state is evaluated by fusing multi-class data through a multilayer perceptron, and whether the routing inspection unmanned aerial vehicle is called for fine routing inspection is determined according to an evaluation result;
s3: patrol and examine unmanned aerial vehicle and carry out the meticulous patrol and examine and upload multi-angle image information to transmission line: if the non-emergency fault occurs, calling an inspection unmanned aerial vehicle to perform fine inspection, acquiring inspection information of the inspection unmanned aerial vehicle, and transmitting the inspection information to a target data control center through a communication module;
s4, fusing fault classification models of multi-view and fixed-view image information of the unmanned aerial vehicle: the static image information of a fixed camera and the multi-angle image information of the fine routing inspection of the unmanned aerial vehicle are fused, a multi-head attention mechanism model based on image characteristics is used for carrying out image fault classification, and a confidence gating mechanism and information divergence constraint are introduced to realize multi-view learning.
2. The dynamic and static cooperative power transmission line refinement routing inspection method according to claim 1, wherein the process of the step S1 further comprises:
s11: the method comprises the steps that static information under a power transmission and transformation scene is collected by fixed collection equipment located on a power transmission line; the control center of the machine nest is communicated with the fixed acquisition equipment through a communication module, and receives and stores the static information acquired by the fixed acquisition equipment;
s12: the control center of the cell is used as the core part of the system to control the transmission of information and the interaction among all modules; the collector communication module provides an information channel for data transmission of the fixed collection equipment and the control center of the machine nest, and ensures that the control center of the machine nest can receive various types of static information collected by the fixed collection equipment;
s13: the method comprises the steps of collecting on-site multi-type static information through fixed collection equipment of the power transmission line, and uploading the multi-type static information to a control center of a machine nest through a communication module to preliminarily judge the running state of the power transmission line.
3. The fine routing inspection method for the dynamic and static cooperative power transmission line according to claim 1, wherein the process of the step S2 further comprises:
s21: the software part included in the step comprises a nest control center and a static judgment module, wherein the nest static judgment module based on static information is carried on the control center of the nest, receives the static information acquired by fixed acquisition equipment, preliminarily judges the state of target routing inspection area equipment through a multilayer sensor model according to the static information, determines whether to call a routing inspection unmanned aerial vehicle for fine routing inspection according to an evaluation result, and generates trigger information if a non-emergency fault occurs; the airframe control center receives the trigger information generated by the static judgment module, and further issues a target instruction to the airframe control module to schedule the inspection unmanned aerial vehicle for fine inspection;
s22: the machine nest static state judgment module adopts a multilayer perceptron model to synthesize multi-class static information to judge the current equipment state, and comprises the following processes:
the method comprises the steps of constructing a multi-class information equipment state judgment data set through static information acquired by fixed acquisition equipment and fault class historical records corresponding to the static information, constructing data set data through the static information acquired by the fixed acquisition equipment and integrating the corresponding equipment state information, marking corresponding fault classes by adopting three types of non-fault, emergent fault and non-emergent fault, namely equipment state labels corresponding to the multi-class static information at the same moment, training a multi-layer perceptron model through the data set, integrating four types of static information including temperature, humidity, smoke concentration and environmental gas concentration acquired by the fixed acquisition equipment to serve as input of the multi-layer perceptron model, judging faults through the multi-layer perceptron, outputting three types of non-fault, emergent fault and non-emergent fault classes, and defining loss function
The training process is constrained by the constraints of the training process,
the formula is defined as follows:
wherein, the first and the second end of the pipe are connected with each other,
a label that indicates the level of the fault,
a failure level indicating the predicted output is indicated,
represents the total number of samples;
training the multilayer sensor model based on the training step, deploying the trained multilayer sensor model in a nest static judgment module, and performing primary judgment on the equipment state;
s23: and respectively taking measures to deal with the fault levels output by the multilayer perceptron model deployed in the nest static judgment module: aiming at the emergency fault, as the emergency fault needs to be processed and maintained in time, related workers need to be sent to the site for further investigation and processing; aiming at non-emergency faults, the inspection unmanned aerial vehicle is adopted for further fine inspection, and trigger information is generated by the nest static judgment module and is sent to the nest control center, so that the inspection unmanned aerial vehicle is further dispatched for fine inspection;
s24: judging whether a fault occurs according to a nest static judging module under a nest control center, and if the fault occurs, dispatching related workers to the site in time for further troubleshooting; if non-emergency faults occur, the trigger information is generated by the machine nest static judging module and is transmitted back to the machine nest control center so as to call the inspection unmanned aerial vehicle for fine inspection.
4. The dynamic and static cooperative power transmission line refinement routing inspection method according to claim 1, wherein the process of the step S3 further comprises:
s31: the hardware part included in the step comprises an inspection unmanned aerial vehicle, and multi-angle and high-precision image acquisition of all equipment and parts around the power transmission line and the tower is realized;
s32: the software part included in the step comprises a nest control module, a nest control center, an unmanned aerial vehicle communication module and a data storage center, wherein the nest control module receives trigger information issued by the nest control center and schedules the inspection unmanned aerial vehicle to perform fine inspection; the unmanned aerial vehicle communication module is used for receiving the instruction from the nest control center and returning the routing inspection information to the nest control center; the data storage center is built based on a database technology, receives all fine routing inspection information of the unmanned aerial vehicle from the nest control center, and stores the generated data file and the generated flight log into a database.
5. The fine routing inspection method for the dynamic and static cooperative power transmission line according to claim 1, wherein the process of the step S4 further comprises:
s41: the method comprises the following steps of constructing a multi-view image fault classification data set based on a database obtained by a fine inspection result of an unmanned aerial vehicle, carrying out multi-angle image acquisition on each part around a power transmission line and a tower in a close range through the inspection unmanned aerial vehicle, and marking the fault type of each part by adopting a manual comprehensive judgment method based on large-scale static image information returned by a high-definition camera carried by the inspection unmanned aerial vehicle, wherein the defined known fault types are as follows: the method comprises the following steps that insulator faults, grading ring faults, vibration damper faults, pole tower structure faults, transmission line damage, the fact that a transmission line is close to a barrier, birds nest, sample pairs of a data set comprise multi-view images and fault type marking results, specifically, the multi-view images comprise N unmanned aerial vehicle inspection view images, and the multi-view images are obtained by carrying out multi-angle image acquisition on the transmission line or the pole tower and other components; the multi-view image also comprises large-scale static image information collected by a camera fixed on the tower;
s42: image fault classification is carried out by using a multi-head attention mechanism model based on image characteristics, and a confidence gating mechanism and information divergence constraint are introduced for improvement, so that multi-view learning is realized, and the fault classification accuracy is improved;
s43: the method comprises the steps of using a loss function to constrain the training process of a fault classification model, wherein the loss function comprises two parts, the first part is to carry out information divergence-based constraint on probability distribution obtained by carrying out pre-training model feature extraction, multi-head attention mechanism model coding and multi-layer perceptron model mapping on N unmanned aerial vehicle inspection visual angle images and fixed visual angle images and screening through a confidence gating mechanism, so that N +1 probability distributions of different visual angles are kept consistent as much as possible, compatible complementation between image information of different visual angles is realized, and the loss function of the part is defined as
The formula is shown as (3),
wherein N represents the number of the unmanned aerial vehicle inspection visual angles, x and y represent two different visual angles respectively,
and with
Respectively representing the probability distribution of the corresponding visual angles, wherein the value of the loss function is equal to the sum of information divergence between every two visual angles of the N +1 visual angles; the second part is to adopt the idea of ensemble learning to vote on the probability distribution of the N +1 visual angles, and the result of prediction of each visual angle has the same valueThe prediction category with the highest number of votes is used as the whole fault classification result, loss calculation is carried out on the fault classification result and the result of manual comprehensive research and judgment marking, and the loss function of the part is defined as
The formula is shown as (4),
wherein
Representing the probability distribution corresponding to the fault manual labeling category,
representing the class probability distribution of the fault obtained by prediction of a classification model, calculating the geometric distance between the two as the value of a loss function,
and (5) integrating the loss functions of the two parts as the training loss function of the fault classification model, wherein the total loss function formula is shown as (5),
wherein
For balancing the ratio of the two partial loss functions,
is in the range of 0 to 1.
6. The dynamic and static cooperative power transmission line refinement routing inspection method according to claim 5, wherein the step S42 specifically comprises the following processes:
s421, calling a ResNet18 residual network pre-training model pre-trained on a large-scale public data set, fixing parameters of ResNet18 during training, keeping the parameters unchanged, using the ResNet18 as a universal image feature extractor to perform feature extraction on a multi-view image, outputting N +1 feature graphs with the size of 512 × 7, and expanding the feature graphs into one-dimensional feature vectors;
s422, respectively encoding the one-dimensional feature vectors corresponding to the view images by using a multi-head attention mechanism model to obtain N +1 one-dimensional high-grade feature vectors with the length of 1024;
s423, respectively inputting the N +1 one-dimensional high-level feature vectors obtained after the coding in the S422 into a multi-layer perceptron classifier, and outputting the predicted fault type probability distribution corresponding to the N +1 view images;
s424, a confidence gating mechanism is introduced to the N +1 probability distributions in S423, the argmax function calculates a maximum probability category corresponding to the probability distributions, and if the probability value of each category is known to be between 0 and 1, and if the probability value corresponding to the maximum probability category corresponding to the probability distributions calculated by the argmax function is smaller than the gating threshold β, the maximum probability distribution result of the view image is discarded, and the specific calculation result is shown in formula (2):
where p represents the probability distribution of a view prediction, x is the class corresponding to the maximum probability in the probability distribution,
a probability value for the category;
s425, calculating information divergence between every two of the plurality of probability distributions subjected to the confidence gating mechanism in the step S424, and constraining distribution consistency of the probability distributions so as to enable the prediction probability distributions of the images of all the view angles to be similar as much as possible, wherein the prediction fault types of the model are determined by the image information of the plurality of view angles together;
s426, voting is carried out on the fault categories obtained by predicting the images of the plurality of visual angles by using the idea of ensemble learning, the final result of the prediction of the fault classification model is obtained by comprehensive study and judgment, the fault categories belong to the known fault types defined in the S41, the multi-head attention mechanism model receives image characteristics extracted by a pre-training model ResNet18 from different visual angles, the image characteristics comprise N unmanned aerial vehicle inspection visual angles and visual angles of fixed cameras, multiple coding is carried out, finally, the fused characteristics are mapped to a lower dimensionality by using a multilayer sensing machine, and the probability distribution of the fault categories corresponding to the images is output.
7. A system for realizing the fine routing inspection method of the power transmission line with dynamic and static coordination of any one of claims 1 to 6 is characterized in that: comprises a fixed acquisition device, a dynamic detection module, a processing module and a judgment module,
the system comprises a fixed acquisition device, a control center and a control center, wherein the fixed acquisition device comprises a temperature sensor, a humidity sensor, a smoke concentration sensor, an environmental gas sensor and a fixed camera, the sensors and the fixed camera are arranged on corresponding monitoring nodes according to the monitoring requirements of a power transmission line, and the fixed acquisition device transmits acquired routing inspection information to the control center through a communication module;
the dynamic detection module comprises an inspection unmanned aerial vehicle, and the inspection unmanned aerial vehicle receives a control center signal through an unmanned aerial vehicle control module;
the processing module comprises a control center of the machine nest, a static judging module and a communication module, the communication module is connected with the control center of the machine nest, the static judging module is communicated with the control center, the communication module is connected with the fixed acquisition equipment and the dynamic detection module, the static judging module is a software module used for judging faults and calling the unmanned aerial vehicle and subsequent data processing, and the software module is carried on the control center of the machine nest.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211140267.1A CN115220479B (en) | 2022-09-20 | 2022-09-20 | Dynamic and static cooperative power transmission line refined inspection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211140267.1A CN115220479B (en) | 2022-09-20 | 2022-09-20 | Dynamic and static cooperative power transmission line refined inspection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115220479A true CN115220479A (en) | 2022-10-21 |
| CN115220479B CN115220479B (en) | 2022-12-13 |
Family
ID=83617636
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211140267.1A Active CN115220479B (en) | 2022-09-20 | 2022-09-20 | Dynamic and static cooperative power transmission line refined inspection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115220479B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660596A (en) * | 2022-11-03 | 2023-01-31 | 创启科技(广州)有限公司 | Data interaction method of mobile terminal and mobile terminal |
| CN116363537A (en) * | 2023-05-31 | 2023-06-30 | 广东电网有限责任公司佛山供电局 | Method and system for identifying hidden danger of hanging objects outside transformer substation |
| US11836968B1 (en) * | 2022-12-08 | 2023-12-05 | Sas Institute, Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
| US12002256B1 (en) | 2022-12-08 | 2024-06-04 | Sas Institute Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102211670A (en) * | 2011-05-17 | 2011-10-12 | 云南电网公司普洱供电局 | Fixed-wing unmanned plane fixed-point shooting system and method for troubleshooting transmission lines thereof |
| KR20190044203A (en) * | 2017-10-20 | 2019-04-30 | 주식회사 삼진엘앤디 | Uav patrol system and patrol method to maintain safety in the designated district |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
| CN111275216A (en) * | 2020-03-11 | 2020-06-12 | 山东科技大学 | Layered collaborative optimization routing inspection method for high-voltage transmission whole line |
| CN111784685A (en) * | 2020-07-17 | 2020-10-16 | 国网湖南省电力有限公司 | Power transmission line defect image identification method based on cloud edge cooperative detection |
| CN112491982A (en) * | 2020-11-13 | 2021-03-12 | 国网天津市电力公司 | Refined sensing method based on cloud edge cooperative power transmission line |
| CN112668696A (en) * | 2020-12-25 | 2021-04-16 | 杭州中科先进技术研究院有限公司 | Unmanned aerial vehicle power grid inspection method and system based on embedded deep learning |
| CN112731960A (en) * | 2020-12-02 | 2021-04-30 | 国网辽宁省电力有限公司阜新供电公司 | Unmanned aerial vehicle remote power transmission line intelligent inspection system and method |
| CN114610050A (en) * | 2022-03-14 | 2022-06-10 | 广东电网有限责任公司 | Inspection method and device for power system, electronic equipment and storage medium |
| CN114970605A (en) * | 2022-05-06 | 2022-08-30 | 大连理工大学 | Multi-mode feature fusion neural network refrigeration equipment fault diagnosis method |
-
2022
- 2022-09-20 CN CN202211140267.1A patent/CN115220479B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102211670A (en) * | 2011-05-17 | 2011-10-12 | 云南电网公司普洱供电局 | Fixed-wing unmanned plane fixed-point shooting system and method for troubleshooting transmission lines thereof |
| KR20190044203A (en) * | 2017-10-20 | 2019-04-30 | 주식회사 삼진엘앤디 | Uav patrol system and patrol method to maintain safety in the designated district |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
| CN111275216A (en) * | 2020-03-11 | 2020-06-12 | 山东科技大学 | Layered collaborative optimization routing inspection method for high-voltage transmission whole line |
| CN111784685A (en) * | 2020-07-17 | 2020-10-16 | 国网湖南省电力有限公司 | Power transmission line defect image identification method based on cloud edge cooperative detection |
| CN112491982A (en) * | 2020-11-13 | 2021-03-12 | 国网天津市电力公司 | Refined sensing method based on cloud edge cooperative power transmission line |
| CN112731960A (en) * | 2020-12-02 | 2021-04-30 | 国网辽宁省电力有限公司阜新供电公司 | Unmanned aerial vehicle remote power transmission line intelligent inspection system and method |
| CN112668696A (en) * | 2020-12-25 | 2021-04-16 | 杭州中科先进技术研究院有限公司 | Unmanned aerial vehicle power grid inspection method and system based on embedded deep learning |
| CN114610050A (en) * | 2022-03-14 | 2022-06-10 | 广东电网有限责任公司 | Inspection method and device for power system, electronic equipment and storage medium |
| CN114970605A (en) * | 2022-05-06 | 2022-08-30 | 大连理工大学 | Multi-mode feature fusion neural network refrigeration equipment fault diagnosis method |
Non-Patent Citations (2)
| Title |
|---|
| ET AL.: "Multi-fitting Detection on Transmission Line based on Cascade Reasoning Graph Network", 《IEEE》 * |
| 姜萍等: "基于改进InceptionV3网络的光伏组件航拍红外图像故障分类方法", 《激光杂志》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660596A (en) * | 2022-11-03 | 2023-01-31 | 创启科技(广州)有限公司 | Data interaction method of mobile terminal and mobile terminal |
| US11836968B1 (en) * | 2022-12-08 | 2023-12-05 | Sas Institute, Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
| US12002256B1 (en) | 2022-12-08 | 2024-06-04 | Sas Institute Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
| CN116363537A (en) * | 2023-05-31 | 2023-06-30 | 广东电网有限责任公司佛山供电局 | Method and system for identifying hidden danger of hanging objects outside transformer substation |
| CN116363537B (en) * | 2023-05-31 | 2023-10-24 | 广东电网有限责任公司佛山供电局 | Method and system for identifying hidden danger of hanging objects outside transformer substation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115220479B (en) | 2022-12-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115220479B (en) | Dynamic and static cooperative power transmission line refined inspection method and system | |
| CN108037770B (en) | Unmanned aerial vehicle power transmission line inspection system and method based on artificial intelligence | |
| CN106026393B (en) | A kind of power network line Intelligent line patrolling and operational safety Risk Forecast System and method | |
| CN110492607A (en) | A kind of intelligent substation condition monitoring system based on ubiquitous electric power Internet of Things | |
| CN112101088B (en) | Unmanned aerial vehicle electric power automatic inspection method, device and system | |
| CN105913604A (en) | Fire occurrence determining method and device based on unmanned aerial vehicle | |
| CN106600887A (en) | Video monitoring linkage system based on substation patrol robot and method thereof | |
| CN109376660B (en) | Target monitoring method, device and system | |
| CN110807460B (en) | Transformer substation intelligent patrol system based on image recognition and application method thereof | |
| CN112087528B (en) | Intelligent water environment monitoring system and method based on deep learning | |
| CN113589837A (en) | Electric power real-time inspection method based on edge cloud | |
| CN112671104A (en) | Transformer substation multidimensional scene control platform facing complex scene | |
| CN113850562A (en) | Intelligent side station supervision method and system | |
| CN115201218A (en) | Vehicle-mounted pavement disease intelligent detection method and system | |
| CN112233270A (en) | Unmanned aerial vehicle is intelligence around tower system of patrolling and examining independently | |
| CN112614130A (en) | Unmanned aerial vehicle power transmission line insulator fault detection method based on 5G transmission and YOLOv3 | |
| CN112613438A (en) | Portable online citrus yield measuring instrument | |
| CN116846059A (en) | Edge detection system for power grid inspection and monitoring | |
| CN114445782A (en) | Power transmission line image acquisition system based on edge AI and Beidou short messages | |
| CN112260402B (en) | Monitoring method for state of intelligent substation inspection robot based on video monitoring | |
| CN113569956A (en) | Mountain fire disaster investigation and identification method based on AI algorithm | |
| CN112819988A (en) | Unmanned aerial vehicle power station intelligent inspection method and system based on 5G and network side server | |
| CN115664006B (en) | Intelligent management and control integrated platform for incremental power distribution network | |
| CN216647401U (en) | Safety helmet recognition device | |
| CN115933750A (en) | Data processing-based power inspection method and power inspection system |
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 |