CN112884018A - Power grid line fault recognition model training method and power grid line inspection method - Google Patents
Power grid line fault recognition model training method and power grid line inspection method Download PDFInfo
- Publication number
- CN112884018A CN112884018A CN202110121070.2A CN202110121070A CN112884018A CN 112884018 A CN112884018 A CN 112884018A CN 202110121070 A CN202110121070 A CN 202110121070A CN 112884018 A CN112884018 A CN 112884018A
- Authority
- CN
- China
- Prior art keywords
- fault
- model
- power grid
- grid line
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000007689 inspection Methods 0.000 title claims abstract description 31
- 238000012360 testing method Methods 0.000 claims abstract description 87
- 230000006870 function Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000012423 maintenance Methods 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013102 re-test Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C1/00—Registering, indicating or recording the time of events or elapsed time, e.g. time-recorders for work people
- G07C1/20—Checking timed patrols, e.g. of watchman
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The application provides a power grid line fault recognition model training method and a power grid line inspection method, wherein the training method comprises the following steps: constructing a training sample set and a testing sample set; training a fault recognition model by adopting sample pictures in a training sample set and matched feature labels until a loss function is minimum; inputting sample pictures in a test sample set into the trained fault recognition model to obtain a test classification result; calculating the accuracy of the model by using the test classification result and the feature labels of the sample pictures in the test sample set; and under the condition that the accuracy of the model does not meet the set requirement, repeatedly executing the training and testing steps until the model meeting the set requirement is obtained. The fault recognition model has the capability of recognizing the line fault of the power grid by training a pre-established initialization fault recognition model. The fault identification model is deployed in practical application, and can replace experienced professionals to identify the shot power grid line characteristic pictures.
Description
Technical Field
The application relates to the technical field of power grid inspection, in particular to a power grid line fault model training method and a power grid line inspection method.
Background
In the power industry, in order to avoid a large-area power failure fault caused by faults of a main power grid line and a tower, the faults of the main power grid and the tower need to be inspected, and the faults are timely found and eliminated.
The adoption of unmanned aerial vehicles for fault routing inspection of power grid lines is common in the power industry; however, at present, after the image data obtained by the routing inspection shooting of the unmanned aerial vehicle is adopted, a professional is still required to identify faults based on the image data; the accuracy of fault identification is directly related to the energy state of professionals after the professionals experience, because the reliability of fault identification by looking up image data of the professionals is directly related to the personal state and the identifiability of the image data; it is also difficult for the relevant professional to concentrate on the identification of faults for a long time. How to reduce the dependence on the experience of professionals in actual operation is the core for realizing the rapid elimination of the grid line fault.
Disclosure of Invention
In order to solve the above technical problem or at least partially solve the above technical problem, the present application provides.
In one aspect, the present application provides a power grid line fault recognition model training method, including:
s101: constructing a training sample set and a testing sample set; the training sample set and the testing sample set both comprise normal sample pictures of components in a power grid line and fault sample pictures of the components when various faults occur; the normal sample picture and the fault sample picture both have matched feature labels;
s102: training a fault recognition model by adopting sample pictures in a training sample set and matched feature labels until a loss function is minimum;
s103: inputting sample pictures in a test sample set into the trained fault recognition model to obtain a test classification result;
s104: calculating model accuracy by using the test classification result and the feature labels of the sample pictures in the test sample set;
s105: in the case where the model accuracy does not meet the set requirement, the steps S102-S104 are repeatedly executed until a fault identification model meeting the set requirement is obtained.
Optionally, step S104 calculates model accuracy using the test classification result and the feature labels of the sample pictures in the test sample set, including:
counting the number of correct fault identifications and the number of all fault identifications in the test classification result;
calculating the accuracy according to the correct fault identification number and the number of samples with the characteristic labels in the test sample set as fault labels;
calculating a recall rate according to the correct fault identification number and the total fault identification number;
and calculating the model accuracy according to the precision and the recall rate.
Optionally, at least two types of fault identification models are included;
the step S105 further includes, under the condition that the model accuracy of at least two fault recognition models meets the set requirement, selecting the fault recognition model with the highest model accuracy as the actual application model.
Optionally, two types of the fault identification models are included, and the two types of the fault identification models are a convolutional neural network model and a multilayer perceptron model respectively.
Optionally, counting the number of times of performing the steps S102-S104;
and when the repetition times exceed the preset times, modifying the framework of the fault identification model and then repeatedly executing the steps S102-S104.
On the other hand, the application provides a power grid line inspection method, which comprises the following steps:
acquiring a power grid line characteristic picture to be identified by adopting a manual inspection and/or unmanned aerial vehicle inspection method;
processing the power grid line characteristic picture to be identified by adopting the fault identification model obtained by the power grid line fault identification model training method to obtain a fault judgment result;
and determining the part needing to be overhauled according to the fault judgment result.
Optionally, the method further comprises: after the part needing to be overhauled is overhauled, acquiring the picture of the part needing to be overhauled again;
processing the picture of the part to be overhauled by adopting the fault identification model to obtain an overhauled fault judgment result;
and determining the maintenance quality according to the fault judgment result after maintenance.
In another aspect, the present application provides a power grid line fault recognition model training device, including:
the sample set constructing unit is used for constructing a training sample set and a testing sample set; the training sample set and the testing sample set both comprise normal sample pictures of components in a power grid line and fault sample pictures of the components when various faults occur; the normal sample picture and the fault sample picture both have matched feature labels;
the model training unit is used for training the fault recognition model by adopting the sample pictures in the training sample set and the matched feature labels until the loss function is minimum;
the model testing unit is used for inputting sample pictures in a test sample set into the trained fault recognition model to obtain a test classification result;
the model accuracy evaluation unit is used for calculating the model accuracy by adopting the test classification result and the feature labels of the sample pictures in the test sample set;
and the training end judging unit is used for triggering the model training unit, the model testing unit and the model accuracy evaluating unit to repeatedly execute operation under the condition that the model accuracy does not meet the set requirement until a fault recognition model meeting the set requirement is obtained.
In another aspect, the present application provides a power grid line inspection device, including:
the image acquisition unit is used for acquiring a power grid line characteristic picture to be identified;
the fault judgment unit is used for processing the power grid line characteristic picture to be recognized by adopting the fault recognition model obtained by the power grid line fault recognition model training method to obtain a fault judgment result;
and the overhauling position determining unit is used for determining the position to be overhauled according to the fault judgment result.
According to the training method of the power grid line fault recognition model, the fault recognition model has the capability of recognizing the power grid line fault by training the pre-established initial fault recognition model. The fault recognition model is deployed in practical application, and can replace experienced professionals to recognize the shot power grid line characteristic picture and determine the power grid line fault problem represented by the characteristic picture; by adopting the model, the working intensity of professionals can be reduced.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and together with the description, serve to explain the principles of the application.
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a flowchart of a power grid line fault recognition model training method provided in an embodiment of the present application;
fig. 2 is a flowchart of a power grid line inspection method provided in the embodiment of the present application;
FIG. 3 is a schematic structural diagram of a power grid line fault recognition model training device provided in an embodiment of the present application;
fig. 4 is a schematic structural diagram of a power grid line inspection device provided in the embodiment of the present application;
fig. 5 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;
wherein: 11-a sample set construction unit, 12-a model training unit, 13-a model testing unit, 14-a model accuracy evaluation unit, 15-a training end judgment unit, 21-an image acquisition unit, 22-a fault judgment unit, 23-an overhaul position determination unit, 31-a processor, 32-a memory, 33-a communication interface and 34-a bus system.
Detailed Description
In order that the above-mentioned objects, features and advantages of the present application may be more clearly understood, the solution of the present application will be further described below. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, but the present application may be practiced in other ways than those described herein; it is to be understood that the embodiments described in this specification are only some embodiments of the present application and not all embodiments.
Fig. 1 is a flowchart of a power grid line fault recognition model training method provided in an embodiment of the present application. As shown in fig. 1, the power grid line fault recognition model training method provided by the embodiment of the present application includes steps S101 to S106.
S101: and constructing a training sample set and a testing sample set.
In the embodiment of the application, the training sample set and the testing sample set are sets including power grid line pictures obtained in a large number of power grid line patrols and examines. In specific application, a power grid line picture can be obtained from historical routing inspection data, and a training sample, namely a test sample set, is constructed.
It should be noted that, in order to ensure that the subsequent training model recognizes various possible faults in the power grid line as much as possible and does not recognize pictures characterizing normal line features as fault pictures as much as possible, both the training sample set and the test sample set have normal sample pictures and fault sample pictures when various components have faults. Wherein the normal sample pictures and the fault sample pictures include pictures of various component parts such as tower poles, wires, bottom wires, drainage wires, and the like, which constitute the power grid line.
The normal sample pictures and the fault sample pictures in the training sample set and the testing sample set respectively comprise corresponding feature labels, and the feature labels are used for representing the labels of fault types. For a normal sample picture, the feature label can only indicate the component part contained in the normal sample picture, but does not set the identification characters such as the character normal of the fault identification field, and the corresponding picture is determined to be a normal sample picture directly according to the condition that the characters of the component part and the fault identification field are null during model training at the later stage. For the fault sample picture, the feature label comprises identifiers used for distinguishing fault conditions, such as component parts, defect types and defect grades and the like contained in the fault sample picture.
It should be noted that the number of various identifiers for distinguishing fault types from feature labels determines the number of nodes of the output layer of the subsequent fault identification model.
In the embodiment of the application, in order to adapt to the input layer setting of the fault identification model, the normal sample picture and the fault sample picture are required to be subjected to standardization processing, and the standardization processing comprises operations of cutting, scaling and the like of the pictures, so that the pictures can include sufficient information of the parts to be identified, and the sizes of the pictures are set to be the same dimension and the channel number.
It should also be noted that, in the embodiment of the present application, the number of sample pictures in the training sample set should be greater than the number of sample pictures in the test sample set; in practical application, the sample ratio of the two samples can be set to be about 4:1, so that both a sufficient amount of sample pictures can be used for training the fault recognition model and a sufficient amount of sample pictures can be used for testing the model. At this time, the number of samples, whether training sample set or testing sample set, should be sufficient to improve the generalization capability of the model.
S102: and training the fault recognition model by adopting sample pictures in the training sample set and the matched feature labels until the loss function is minimum.
Step S102 is a training step of the fault recognition model. In the step, sample pictures in the training sample set are used as input and input into the fault recognition model, so that the probability that the pictures are recognized as various feature labels is obtained, and an output result vector is constructed by utilizing the probability recognized as various feature labels. In addition, a discrimination vector needs to be constructed according to the feature tag of a certain picture. And then, calculating a loss function by using the output result vector and the discrimination vector, and correcting the network parameters of the fault identification model by adopting an error back propagation algorithm until the loss function is minimum.
And after the calculation loss function is minimum, determining that the fault recognition model is trained completely. In the embodiment of the present application, the loss function is minimum, which means that the loss function meets the set order requirement, or the error back propagation algorithm is used for the set number of times.
S103: and inputting sample pictures in the test sample set into the trained fault recognition model to obtain a test classification result.
After the trained fault recognition model is obtained, the sample pictures in the test sample data set can be used as input, and the fault recognition model is used for obtaining a test classification result. The test classification result actually takes the classification label corresponding to the maximum probability in the output result vector as the test classification result.
It should be noted that, in step S103, all sample pictures in the test sample set are input to the fault identification model, and a corresponding test classification result is obtained.
S104: and calculating the accuracy of the model by using the test classification result and the feature labels of the sample pictures in the test sample set.
In the embodiment of the application, the calculation of the model accuracy is a process of comparing and counting data by using the test classification result and the feature labels of the sample images in the sample image set, and taking the statistical result as the accuracy of the representation model.
In an application of the embodiment of the present application, the step S104 may be detailed as steps S1041-S1044.
S1041: and counting the number of correct fault identifications and the number of all fault identifications in the test classification result.
The correct fault identification number is the number of test classification results satisfying the following conditions in all test classification results: (1) testing the classification result as a fault result; (2) and the fault type in the test classification result is the same as the feature label of the corresponding sample picture.
The total fault identification number is the number of test classification results of which the test results are fault results.
S1042: and calculating the accuracy according to the correct fault identification number and the sample number of the test sample set characteristic labels as fault labels.
In the embodiment of the application, the precision represents the ratio of the identified fault samples in all the fault samples in the test samples, and the ratio is obtained by adopting the ratio of the number of correct fault samples to the number of samples in which the test samples, namely the medium feature labels, are fault labels.
If the number of correct failure identifications is a, and the number of samples with feature labels as failure labels is b, the precision is precision a/b.
S1043: and calculating the recall rate according to the correct fault identification number and the total fault identification number.
The recall table characterizes the ratio of the number of identified faults to the number of total faults identified in the test sample. If the total fault identification number is c, the recall rate recall is a/c.
S1044: calculating model accuracy from accuracy and recall
In the embodiment of the present application, the accuracy of the model is calculated according to the accuracy and the recall ratio, according to a formula F1 ═ precision × recall)/(precision + recall, where F1 is the accuracy of the model.
S105: judging whether the accuracy of the model meets the set requirement; if yes, go to step S106, otherwise, go back to step S102.
S106: and finishing the model training.
In the embodiment of the application, if the accuracy of the model meets the set requirement, the model does not need to be retrained; if the model accuracy does not meet the requirements, the steps of steps S102-S104 need to be re-executed.
In the embodiment of the present application, a precision threshold may be preset in step S105, and only when the model precision exceeds the precision threshold, the model is determined to have a better generalization capability, which can be used for actual operation.
The embodiment of the application provides a training method of a power grid line fault recognition model, which enables the fault recognition model to have the capability of recognizing the power grid line fault by training a pre-established initialization fault recognition model. The fault recognition model is deployed in practical application, and can replace experienced professionals to recognize the shot power grid line characteristic picture and determine the power grid line fault problem represented by the characteristic picture; by adopting the model, the working intensity of professionals can be reduced.
In the embodiment of the application, there may be a plurality of fault recognition models that are actually constructed, and in the process of executing step S103, each fault recognition model is trained, and the accuracy of each fault recognition model is tested through step S104. And in step S105, the following steps are performed: under the condition that the fault recognition models meet the accuracy requirement, the fault recognition model with the highest model accuracy is selected as the practical application model, and at the moment, even if the accuracy of some models does not meet the requirement, the models are not trained.
In a specific application, the at least two types of fault identification models can be a convolutional neural network model and a multilayer perceptron model; of course, in other applications, other models used in the field of deep learning may also be employed.
In a specific application of the embodiment of the present application, in addition to the foregoing steps S101 to S106, the method may further include the following steps: counting the repetition times of the steps S102-S104, and judging whether the repetition times exceed the preset times; if yes, the steps S102-S104 are repeatedly executed after the framework of the fault identification model is modified. In specific application, modifying the architecture of the fault identification model comprises changing the number of layers of the model, changing the number of nodes of a certain hidden layer in the model, changing the type of an activation function of each node and the like.
Besides providing the aforementioned power grid line fault recognition model training method, the embodiment of the present application further provides a power grid line polling method. Fig. 2 is a flowchart of a power grid line inspection method provided in the embodiment of the present application. As shown in fig. 2, the power grid line inspection method includes steps S201 to S203.
S201: and obtaining a power grid line characteristic picture to be identified.
The power grid line characteristic picture to be identified is obtained by adopting a manual inspection method or an unmanned aerial vehicle inspection method and comprises the picture of the power grid line characteristic. And after the picture is obtained, processing the picture (including operations such as cutting, zooming and the like) according to the steps in the model training method to obtain the processed characteristic picture to be recognized.
S202: and inputting the power grid line characteristic picture to be identified into a fault identification model to obtain a fault judgment result.
S203: and determining the part to be overhauled according to the fault judgment result.
In step S202, the result of the fault determination is to determine whether the corresponding picture includes fault information, and if the corresponding picture includes the fault information, step S203 determines a power grid line portion corresponding to the picture according to the picture tracing source including the fault information, and determines the portion as an inspection part.
In practical application, the parts to be overhauled can be determined by adopting the methods of the steps S201 to S203, and then whether the corresponding grid line characteristic picture is a fault part picture is determined by adopting a manual recheck method, and then whether the determined overhauled parts are overhauled is determined.
In practical application, after the pictures of the manual retest are confirmed and the feature labels are standardized, the pictures are added into a database, and then the pictures of the database are continuously utilized to retrain the fault recognition model.
In the embodiment of the application, the feature label of the fault identification model includes type information such as a defect part, a defect grade and the like, and the actually obtained fault determination result may also include such information. Therefore, the parts which are preferentially overhauled can be determined according to the defect parts and the defect grades, and the defect grading early warning is realized.
In the specific application of the embodiment of the present application, in addition to the foregoing steps S201 to S203, the method for inspecting a power grid line may further include steps S204 to S206.
S204: and after the part needing to be overhauled is overhauled, acquiring the picture of the part needing to be overhauled again.
S205: processing a picture of a part to be overhauled by adopting a fault identification model to obtain an overhauled fault judgment result;
s206: and determining the maintenance quality according to the fault judgment result after maintenance.
In addition to providing the circuit line fault recognition model training method, the embodiment of the application also provides a power grid line fault recognition model training device. Fig. 3 is a schematic structural diagram of a power grid line fault recognition model training device provided in an embodiment of the present application, and as shown in fig. 3, the training device includes a sample set construction unit 11, a model training unit 12, a model testing unit 13, a model accuracy evaluation unit 14, and a training end determination unit 15.
The sample set constructing unit 11 is configured to construct a training sample set and a testing sample set; the training sample set and the testing sample set both comprise normal sample pictures of components in the power grid line and fault sample pictures of the components when various faults occur; the normal sample picture and the fault sample picture both have matched feature labels;
the model training unit 12 is configured to train the fault recognition model by using the sample pictures in the training sample set and the matched feature labels until the loss function is minimum;
the model testing unit 13 is configured to input a sample picture in a test sample set to the trained fault identification model to obtain a test classification result;
the model accuracy evaluation unit 14 calculates the model accuracy by using the test classification result and the feature labels of the sample pictures in the test sample set;
the training end judging unit 15 is used for triggering the model training unit 12, the model testing unit 13 and the model accuracy evaluating unit 14 to repeatedly execute operations under the condition that the model accuracy does not meet the set requirement until a fault recognition model meeting the set requirement is obtained.
In one specific application, the model accuracy evaluation unit 14 determines the accuracy of the fault identification model using the following method: counting the number of correct fault identifications and the number of all fault identifications in the test classification result; calculating the accuracy according to the correct fault identification number and the sample number of the test sample set characteristic labels as fault labels; calculating the recall rate according to the correct fault identification number and the total fault identification number; and calculating the accuracy of the model according to the accuracy and the recall rate.
In a specific application of the embodiment of the application, the model training unit 12 may train at least medium-type fault recognition models, the corresponding model testing unit 13 tests each fault recognition model to obtain a test classification result, the model accuracy evaluation unit 14 evaluates the model accuracy of each fault recognition model, and the training end determination unit 15 is further configured to select a fault recognition model that has the highest accuracy and meets the setting requirement as the final application model.
Based on the invention concept of the power grid line inspection method provided in the foregoing, the embodiment of the present application further provides a power grid line inspection device. Fig. 4 is a schematic structural diagram of a power grid line inspection device provided in the embodiment of the present application; as shown in fig. 4, the power grid line inspection device includes an image acquisition unit 21, a fault determination unit 22, and a maintenance location determination unit 23.
The image obtaining unit 21 is configured to obtain a to-be-identified power grid line feature picture.
The fault judgment unit 22 is configured to process the to-be-recognized power grid line characteristic picture by using a fault recognition model obtained by a power grid line fault recognition model training method to obtain a fault judgment result;
the inspection portion determining unit 23 is configured to determine a portion to be inspected according to the failure determination result.
Based on the inventive concept, the application also provides an electronic device. Fig. 5 is a schematic structural diagram of an electronic device provided in an embodiment of the present application. As shown in fig. 5, the first server comprises at least one processor 31, at least one memory 32 and at least one communication interface 33. And a communication interface 33 for information transmission with an external device.
The various components in the first server are coupled together by a bus system 34. Understandably, the bus system 34 is used to enable connective communication between these components. The bus system 34 includes a power bus, a control bus, and a status signal bus in addition to the data bus. For clarity of illustration, the various buses are labeled as bus system 34 in fig. 5.
It will be appreciated that the memory 32 in this embodiment can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. In some embodiments, memory 32 stores elements, executable units or data structures, or a subset thereof, or an expanded set thereof: an operating system and an application program.
The operating system includes various system programs, such as a framework layer, a core library layer, a driver layer, and the like, and is used for implementing various basic tasks and processing hardware-based tasks. The application programs include various application programs such as a media player (MediaPlayer), a Browser (Browser), etc. for implementing various application tasks. The program for implementing the power grid line fault recognition model training method and/or the power grid line inspection method provided by the embodiment of the application can be included in the application program.
In the embodiment of the present application, the processor 31 is configured to execute the steps of the grid line fault recognition model training method and/or the grid line polling method provided in the embodiment of the present application by calling a program or an instruction stored in the memory 32, which may be specifically a program or an instruction stored in an application program.
The power grid line fault recognition model training method and/or the power grid line inspection method provided by the embodiment of the application can be applied to the processor 31, or can be realized by the processor 31. The processor 31 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 31. The Processor 31 may be a general-purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The steps of the power grid line fault recognition model training method and/or the power grid line inspection method provided by the embodiment of the application can be directly implemented by a hardware decoding processor, or implemented by combining hardware and software units in the decoding processor. The software elements may be located in ram, flash, rom, prom, or eprom, registers, among other storage media that are well known in the art. The storage medium is located in the memory 32, and the processor 31 reads the information in the memory 32 and performs the steps of the method in combination with the hardware thereof.
The embodiments of the present application further provide a non-transitory computer-readable storage medium, where the non-transitory computer-readable storage medium stores a program or an instruction, where the program or the instruction causes a computer to execute the steps of each embodiment of the power grid line fault identification model training method and/or the power grid line polling method, and in order to avoid repeated descriptions, the steps are not described herein again.
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The above description is merely exemplary of the present application and is presented to enable those skilled in the art to understand and practice 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 (10)
1. A power grid line fault recognition model training method is characterized by comprising the following steps:
s101: constructing a training sample set and a testing sample set; the training sample set and the testing sample set both comprise normal sample pictures of components in a power grid line and fault sample pictures of the components when various faults occur; the normal sample picture and the fault sample picture both have matched feature labels;
s102: training a fault recognition model by adopting sample pictures in a training sample set and matched feature labels until a loss function is minimum;
s103: inputting sample pictures in a test sample set into the trained fault recognition model to obtain a test classification result;
s104: calculating model accuracy by using the test classification result and the feature labels of the sample pictures in the test sample set;
s105: in the case where the model accuracy does not meet the set requirement, the steps S102-S104 are repeatedly executed until a fault identification model meeting the set requirement is obtained.
2. The power grid line fault recognition model training method of claim 1, wherein step S104 is implemented to calculate model accuracy using the test classification result and the feature labels of the sample pictures in the test sample set, and comprises:
counting the number of correct fault identifications and the number of all fault identifications in the test classification result;
calculating the accuracy according to the correct fault identification number and the number of samples with the characteristic labels in the test sample set as fault labels;
calculating a recall rate according to the correct fault identification number and the total fault identification number;
and calculating the model accuracy according to the precision and the recall rate.
3. The power grid line fault recognition model training method of claim 1, wherein: at least two types of fault recognition models are included;
the step S105 further includes, under the condition that the model accuracy of at least two fault recognition models meets the set requirement, selecting the fault recognition model with the highest model accuracy as the actual application model.
4. The grid line fault recognition model training method according to claim 3,
the fault recognition method comprises two types of fault recognition models, wherein the two types of fault recognition models are a convolutional neural network model and a multilayer perceptron model respectively.
5. The power grid line fault recognition model training method of claim 1, further comprising:
counting the number of times of executing the steps S102-S104;
and when the repetition times exceed the preset times, modifying the framework of the fault identification model and then repeatedly executing the steps S102-S104.
6. A power grid line inspection method is characterized by comprising the following steps:
acquiring a power grid line characteristic picture to be identified by adopting a manual inspection and/or unmanned aerial vehicle inspection method;
processing the power grid line characteristic picture to be identified by adopting a fault identification model obtained by the power grid line fault identification model training method according to any one of claims 1 to 5 to obtain a fault judgment result;
and determining the part needing to be overhauled according to the fault judgment result.
7. The power grid line patrol method according to claim 6, further comprising:
after the part needing to be overhauled is overhauled, acquiring the picture of the part needing to be overhauled again;
processing the picture of the part to be overhauled by adopting the fault identification model to obtain an overhauled fault judgment result;
and determining the maintenance quality according to the fault judgment result after maintenance.
8. A power grid line fault recognition model training device is characterized by comprising:
the sample set constructing unit is used for constructing a training sample set and a testing sample set; the training sample set and the testing sample set both comprise normal sample pictures of components in a power grid line and fault sample pictures of the components when various faults occur; the normal sample picture and the fault sample picture both have matched feature labels;
the model training unit is used for training the fault recognition model by adopting the sample pictures in the training sample set and the matched feature labels until the loss function is minimum;
the model testing unit is used for inputting sample pictures in a test sample set into the trained fault recognition model to obtain a test classification result;
the model accuracy evaluation unit is used for calculating the model accuracy by adopting the test classification result and the feature labels of the sample pictures in the test sample set;
and the training end judging unit is used for triggering the model training unit, the model testing unit and the model accuracy evaluating unit to repeatedly execute operation under the condition that the model accuracy does not meet the set requirement until a fault recognition model meeting the set requirement is obtained.
9. The utility model provides a power grid line inspection device which characterized in that includes:
the image acquisition unit is used for acquiring a power grid line characteristic picture to be identified;
the fault judgment unit is used for processing the power grid line characteristic picture to be identified by adopting the fault identification model obtained by the power grid line fault identification model training method according to any one of claims 1 to 5 to obtain a fault judgment result;
and the overhauling position determining unit is used for determining the position to be overhauled according to the fault judgment result.
10. An electronic device comprising a processor and a memory;
the processor is adapted to perform the steps of the method of any one of claims 1 to 7 by calling a program or instructions stored in the memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121070.2A CN112884018A (en) | 2021-01-28 | 2021-01-28 | Power grid line fault recognition model training method and power grid line inspection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121070.2A CN112884018A (en) | 2021-01-28 | 2021-01-28 | Power grid line fault recognition model training method and power grid line inspection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112884018A true CN112884018A (en) | 2021-06-01 |
Family
ID=76053268
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110121070.2A Pending CN112884018A (en) | 2021-01-28 | 2021-01-28 | Power grid line fault recognition model training method and power grid line inspection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112884018A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113689111A (en) * | 2021-08-20 | 2021-11-23 | 北京百度网讯科技有限公司 | Fault recognition model training method, fault recognition device and electronic equipment |
| CN113759968A (en) * | 2021-09-01 | 2021-12-07 | 中国南方电网有限责任公司超高压输电公司贵阳局 | Unmanned aerial vehicle-based power grid line patrol planning method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107067026A (en) * | 2017-02-16 | 2017-08-18 | 国网江苏省电力公司电力科学研究院 | Electrical equipment fault detection method based on deep neural network |
| CN110580475A (en) * | 2019-11-12 | 2019-12-17 | 江西博微新技术有限公司 | line diagnosis method based on unmanned aerial vehicle inspection, electronic device and storage medium |
| US20200403406A1 (en) * | 2019-06-19 | 2020-12-24 | King Fahd University Of Petroleum And Minerals | Distribution grid fault analysis under load and renewable energy uncertainties |
| CN112232133A (en) * | 2020-09-18 | 2021-01-15 | 许继集团有限公司 | Power transmission line image identification method and device based on deep convolutional neural network |
-
2021
- 2021-01-28 CN CN202110121070.2A patent/CN112884018A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107067026A (en) * | 2017-02-16 | 2017-08-18 | 国网江苏省电力公司电力科学研究院 | Electrical equipment fault detection method based on deep neural network |
| US20200403406A1 (en) * | 2019-06-19 | 2020-12-24 | King Fahd University Of Petroleum And Minerals | Distribution grid fault analysis under load and renewable energy uncertainties |
| CN110580475A (en) * | 2019-11-12 | 2019-12-17 | 江西博微新技术有限公司 | line diagnosis method based on unmanned aerial vehicle inspection, electronic device and storage medium |
| CN112232133A (en) * | 2020-09-18 | 2021-01-15 | 许继集团有限公司 | Power transmission line image identification method and device based on deep convolutional neural network |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113689111A (en) * | 2021-08-20 | 2021-11-23 | 北京百度网讯科技有限公司 | Fault recognition model training method, fault recognition device and electronic equipment |
| CN113759968A (en) * | 2021-09-01 | 2021-12-07 | 中国南方电网有限责任公司超高压输电公司贵阳局 | Unmanned aerial vehicle-based power grid line patrol planning method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10885618B2 (en) | Inspection apparatus, data generation apparatus, data generation method, and data generation program | |
| CN110021005B (en) | Method and device for screening defects of circuit board and computer readable recording medium | |
| EP3502966A1 (en) | Data generation apparatus, data generation method, and data generation program | |
| US11029260B2 (en) | Solder paste printing quality inspection system and method | |
| WO2019051941A1 (en) | Method, apparatus and device for identifying vehicle type, and computer-readable storage medium | |
| CN109060817B (en) | Artificial intelligence reinspection system and method thereof | |
| US11727553B2 (en) | Vision analysis and validation system for improved inspection in robotic assembly | |
| CN112884018A (en) | Power grid line fault recognition model training method and power grid line inspection method | |
| CN112528975A (en) | Industrial quality inspection method, device and computer readable storage medium | |
| CN114445746A (en) | Model training method, railway contact net abnormity detection method and related device | |
| US20220343642A1 (en) | High-precision identification method and system for substations | |
| CN112529109A (en) | Unsupervised multi-model-based anomaly detection method and system | |
| CN112419268A (en) | Method, device, equipment and medium for detecting image defects of power transmission line | |
| US11120308B2 (en) | Vehicle damage detection method based on image analysis, electronic device and storage medium | |
| CN111507483A (en) | Rework board detection apparatus, method, and computer-readable storage medium | |
| CN117372424B (en) | Defect detection method, device, equipment and storage medium | |
| CN111415326A (en) | Method and system for detecting abnormal state of railway contact net bolt | |
| CN113592864B (en) | Transformer monitoring method, device, system and medium based on convolutional neural network | |
| CN113447572B (en) | Steel rail flaw detection method, electronic device, steel rail flaw detection vehicle and readable storage medium | |
| CN116383710A (en) | Label determining method, device, electronic equipment and storage medium | |
| KR20210037199A (en) | Apparatus for dividing, tagging an image and for detecting defect of facilities using the same | |
| CN115294505A (en) | Risk object detection and model training method and device and electronic equipment | |
| CN114897863A (en) | Defect detection method, device and equipment | |
| CN114155412A (en) | Deep learning model iteration method, device, equipment and storage medium | |
| CN113284141A (en) | Model determination method, device and equipment for defect detection |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210601 |
|
| RJ01 | Rejection of invention patent application after publication |