CN112613453A - Method and system for checking violation of regulations on construction site of electric power infrastructure - Google Patents
Method and system for checking violation of regulations on construction site of electric power infrastructure Download PDFInfo
- Publication number
- CN112613453A CN112613453A CN202011601489.XA CN202011601489A CN112613453A CN 112613453 A CN112613453 A CN 112613453A CN 202011601489 A CN202011601489 A CN 202011601489A CN 112613453 A CN112613453 A CN 112613453A
- Authority
- CN
- China
- Prior art keywords
- sample
- violation
- training
- hidden danger
- image
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000010276 construction Methods 0.000 title claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 63
- 238000009440 infrastructure construction Methods 0.000 claims description 19
- 230000007547 defect Effects 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 238000013135 deep learning Methods 0.000 claims description 8
- 238000002372 labelling Methods 0.000 claims description 8
- 238000013024 troubleshooting Methods 0.000 claims description 8
- 238000011835 investigation Methods 0.000 claims description 2
- 238000007689 inspection Methods 0.000 description 9
- 238000013527 convolutional neural network Methods 0.000 description 8
- 238000007726 management method Methods 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000011176 pooling Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 230000002950 deficient Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000016507 interphase Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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
- G06F18/24—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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention relates to a method and a system for checking violation of regulations on a construction site of electric power infrastructure, which comprises the following steps: acquiring a historical sample image and establishing a sample library; carrying out data annotation on a sample image with hidden danger in a sample library; carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification; the collected field images are input into the training model to obtain the violation type and the violation position.
Description
Technical Field
The invention relates to the technical field of inspection of an electric power infrastructure construction site, in particular to a method and a system for checking violation of regulations in an electric power infrastructure construction site.
Background
The statements herein merely provide background information related to the present disclosure and may not necessarily constitute prior art.
In the operation site of the power equipment, besides the invasion of various weather conditions, the power equipment is also damaged by other external forces, such as the impact of engineering machinery on a tower, the suspension of kites and other objects on a lead to cause interphase short circuit and the like, and the safety operation of the power equipment can be threatened at any time. In order to ensure the safe operation of the power equipment, the patrol and inspection of violation of capital construction sites must be enhanced, people or objects which may threaten the power construction are foreseen, violation defects influencing the power construction and factors endangering the safe operation of the power equipment are discovered in time, and the violation operation can be stopped quickly or the remote violation early warning can be realized.
At present, a power infrastructure site mainly adopts various means such as site inspection, remote monitoring and the like to obtain a large amount of monitoring data, the related equipment is large in quantity and various in types, and each type of equipment has a lot of faults and defect types. The inventor finds that the identification and judgment of violation defects and hidden dangers mostly depend on the experience of inspection personnel, and in the process of inspecting image processing, the algorithms of intelligent image identification and automatic defect judgment are lacked, so that the automation degree of image data processing is low, the working efficiency of personnel is low, the labor intensity is high, and the requirements of modern power grid construction and development cannot be met.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides the method for checking the violation of regulations on the electric power infrastructure construction site, improves the intelligent identification efficiency and accuracy of the violation of regulations, and reduces the labor intensity of infrastructure safety management workers.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, the invention provides a violation checking method for a construction site of power infrastructure, which comprises the following steps:
acquiring a historical sample image and establishing a sample library;
carrying out data annotation on a sample image with hidden danger in a sample library;
carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification;
and inputting the acquired field image into a training model to obtain the violation type and the violation position.
Further, whether the collected sample image is a sample image with hidden danger or not is judged according to the pre-obtained typical characteristics of the construction hidden danger, and data annotation is carried out on the sample image with hidden danger.
Furthermore, the record format of the annotation file is required by adopting xml in the PASCAL VOC data set format as a standard, and all objects with defects and hidden dangers in the sample image are annotated with data.
Further, after data annotation is carried out on the sample images with the hidden danger in the sample library, data annotation is carried out on the typical characteristics of the sample images with the hidden danger in the sample library according to the image characteristics of the hidden danger.
Further, in the model training and tuning process and the model training and tuning process, the sample image data is identified based on the fast-RCNN network structure, training samples and test samples are randomly selected, and the training model is constructed and tuned by utilizing a deep learning algorithm.
Further, 90% of sample images of each type of hidden danger are randomly selected to construct training samples, and the rest 10% of sample images are used as test samples to construct training models.
Furthermore, after the collected field images are input into the training model to obtain the violation type and the violation position, the images are added into the sample library to update the sample library.
In a second aspect, the invention discloses a violation investigation system for a power infrastructure construction site, which comprises
A sample library establishing module: the system is used for acquiring historical sample images and establishing a sample library;
a data labeling module: the system is used for carrying out data annotation on the sample images with hidden danger in the sample library;
a training model establishing module: the training model is used for carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification;
an identification module: and the system is used for inputting the acquired field images into the training model to obtain the violation type and the violation position.
In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored on the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the steps of the method of the first aspect.
In a fourth aspect, the present invention provides a computer readable storage medium for storing computer instructions which, when executed by a processor, perform the steps of the method of the first aspect.
The invention has the beneficial effects that:
according to the method, the artificial intelligent deep learning algorithm is adopted to establish the training model, various violations of the electric power infrastructure construction site can be quickly identified by the training model, the intelligent identification efficiency and accuracy of violation operation can be improved, the construction team is guided to quickly carry out safety protection, the labor intensity of infrastructure safety management workers can be reduced, the violation identification management period is shortened, and the conversion of the electric power infrastructure safety management mode to the intelligent operation inspection mode is promoted.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the application and, together with the description, serve to explain the application and are not intended to limit the application.
FIG. 1 is a schematic flow chart of a method according to example 1 of the present invention;
FIG. 2 is a sample image without hidden trouble in embodiment 1 of the present invention;
FIG. 3 is a sample image with hidden danger in embodiment 1 of the present invention;
FIG. 4 is a diagram illustrating annotation of hidden danger sample image data according to embodiment 1 of the present invention;
FIG. 5 is a typical feature data annotation diagram of a hidden danger sample image according to embodiment 1 of the present invention;
FIG. 6 is a diagram illustrating a fast-RCNN network according to an embodiment 1 of the present invention;
FIG. 7 is a schematic diagram of a convolutional neural network according to embodiment 1 of the present invention;
FIG. 8 is a sample view of example 1 of the present invention;
FIG. 9 is a schematic view of the Loss curve in example 1 of the present invention;
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
As introduced by the background technology, most of the prior methods rely on the experience of inspection personnel in the aspects of identifying and judging violation defects and hidden dangers in the infrastructure construction site, the working efficiency of the personnel is low, the labor intensity is high, and the violation checking method for the electric power infrastructure construction site is provided by the application aiming at the problems.
In an exemplary embodiment of the present application, as shown in fig. 1, a method for troubleshooting violations at a construction site of power infrastructure includes the following steps:
step 1: and (3) analyzing hidden danger typical characteristics of the electric power infrastructure construction site in advance.
Taking construction machinery as an example, typical characteristics of construction hidden dangers are established, the covered construction machinery mainly comprises 12 marked target types such as a crane, a crane suspender, a pump truck arm, an excavator, a bulldozer, a tower crane, a pile driver rod, a rolling mill, a forklift, an excavator, a grader and the like, and the construction machinery hidden danger target has the following characteristics:
(1) the target size ratios differ: the crane boom, the pump truck arm, etc. are of the type that is elongated with an aspect ratio of less than 0.2 times when the target is placed vertically, with the aspect ratio of the crane, the pump truck, etc. vehicles being approximately 2 times closer, and with the aspect ratio of most targets being between 0.5 and 2 times.
(2) When the target is tilted, the target effective pixels in the marking box occupy 1/5 of the marking box, and most of the area is the background part.
(3) Rigidity target: all objects have sharp contours and edges and are essentially rigid objects.
Step 2: the method comprises the steps of obtaining historical sample images, establishing a sample library with high diversity, balance and quality, wherein image data is an important component for researching an artificial intelligent recognition algorithm, and in order to guarantee a good algorithm effect, the sample library with high diversity, balance and quality needs to be established. A complete and standard high-quality sample library is a premise of algorithm training, a large amount of standard and complete image data is a foundation of subsequent working data, and the following implementation steps are mainly adopted:
and screening sample data and screening out historical sample images.
Taking the transmission line as an example, the body image acquisition mainly depends on the helicopter and the unmanned aerial vehicle inspection task. The data mainly comprises picture files acquired by a visible light holder, and the resolution of the relatively visualized image is high, and is generally more than 2000 ten thousand pixels. The originally acquired picture contains a large number of non-defective images and partial defective images, and the defective images usually account for less than 1%.
The transmission channel image acquisition mainly depends on a visual monitoring shooting device deployed on a transmission line tower. The image monitoring shooting device is used as a field core device and is responsible for functions of image data acquisition, data transmission and the like, the images have the advantages of small flow occupation, high resolution and the like, the hidden danger typical characteristics of the electric power infrastructure construction field are analyzed in advance according to the step 1, the sample images are distinguished to be hidden danger sample images and hidden danger-free sample images, the hidden danger sample images are shown in a figure 2, and the hidden danger-free sample images are shown in a figure 3.
And step 3: standardizing a sample data annotation method, and carrying out data annotation on a sample image with hidden danger in a sample library;
currently, most of the algorithms based on deep learning object recognition are supervised learning, and the supervised learning is a machine learning task for deducing an object function from a labeled training data set. The larger the scale of the labeled training data set is, the stronger the diversity is, and the more accurate the labeling is, the stronger the generalization of the final training model is.
As shown in fig. 4, in order to meet the standardization of target identification, the xml in the PASCAL VOC data set format is adopted as the standard to request the recording format of the identification file, and all objects with defects and hidden dangers in the sample image are subjected to data annotation to be accurate to the target part.
In order to improve the generalization capability of the model and avoid overfitting, a large amount of training data needs to be labeled, and the diversity of sample images needs to be increased. The established sample library is to contain sample images of various scenes, weather, illumination and the like.
In the embodiment, a semi-automatic mode is adopted for data marking, namely, in the initial stage, data with set quantity of manual instruction marking are received, then the data are used for training a recognition model, and then the trained recognition model is used for recognizing the unmarked pictures to obtain marking information; and finally, receiving a manual instruction to assist in deleting a part of the error marked image. By the semi-automatic data labeling method, a large amount of manpower and material resources can be saved. And along with the gradual increase of the scale of the database, the identification precision of the trained model can be gradually improved, the identification result of the model can be directly used as the marking information at the later stage, the full-automatic data marking is realized, and the sample library of the power transmission hidden trouble sample image can be effectively expanded.
And 4, step 4: and carrying out data annotation on the image typical characteristics of the sample image with the hidden danger in the sample library.
Due to the similar appearance and shape of the objects, the objects are more, such as crane booms, pump truck arms and pile driver poles, bulldozers and excavators, and the like. When designing a network structure, the characteristics of different construction machines need to be fully considered, and data annotation of typical characteristics of the image is carried out. The typical features refer to unique features of the hidden danger target, and as shown in fig. 5, data annotation of typical features of an image is performed by taking an excavator as an example.
In the step 3 and the step 4, target labeling is carried out by using a uniform image labeling tool to form a standard xml labeling result. And the archiving storage is executed according to the specified naming and storage modes, so that the requirements of deep learning image analysis training and testing based on the convolutional neural network are met.
And 5: carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification; and constructing a training model by adopting image sample analysis based on deep learning.
The target identification method based on deep learning generally comprises two parts, wherein one part is a feature extraction network, which is generally also called a backbone network and is responsible for feature extraction of images, the network extraction is generally basic feature information, such as edge features, texture features and the like, the part is generally realized by a network structure in front of a classification layer of an image classification task, the image classification is also one of classic problems in the field of computer vision, and the most common image classification task data is an ImageNet data set and comprises 1000 images of different classes; the other part is a frame regression network, also called as an identification frame or a frame regression method, which is responsible for target positioning on the basis of image feature extraction, and the higher the identification result and the IOU of the labeled boundary frame are, the higher the positioning accuracy of the frame regression network is.
In this embodiment, the construction of the training model includes the following specific steps:
step 5.1 identifies the sample image data based on the fast-RCNN network structure.
The fast-RCNN is one of the most common models applied in the field of target recognition at present, has good effects on the aspects of target recognition accuracy, recognition speed and the like, and is an improved model from an RCNN framework step by step and belongs to the most representative model in an RCNN target recognition algorithm series. As a representative of the two-stage target detection model, the fast-RCNN achieves balance in target identification effect and calculation complexity relative to other models, and can be used as a backbone network for identifying violation and potential safety hazard in a power infrastructure construction monitoring image.
The Faster-RCNN further optimizes the network on the basis of the RCNN and the Fast-RCNN, so that the comprehensive performance is greatly improved, particularly the recognition speed is obvious, and the real-time requirement of the system can be basically met when a proper feature extraction network and parameter setting are selected. The fast-RCNN model can be largely generalized to 4 major components, including a Convolutional Neural Network (CNN) layer, a region suggestion network (RPN) layer, a region interest pooling (ropooling) layer, and a fully connected layer (Softmax).
As shown in fig. 6, the input image size of the fast-RCNN may be any size, the input image first passes through the convolutional neural network to extract a feature map of the image, and the extracted feature maps are respectively used for a subsequent RPN layer and a full link layer; the extracted feature map is used as input in an RPN layer, the RPN network is used for generating a candidate region (regionproposals) and a region score of an image, then the output of the proposals of the RPN layer and the feature map extracted by a convolutional neural network are input in Roi Pooling, regions with different sizes are normalized to a feature map with a fixed size, and finally the feature map is sent to a subsequent full-connection layer, the category of the proposals is calculated, and the final position information of an identification frame is obtained through frame regression.
The convolutional neural network structure is shown in fig. 7, and the convolutional neural network mainly comprises a convolutional layer, a pooling layer and a full-connection layer, wherein an original image enters the network from the leftmost input layer, then is continuously processed through the convolutional layer and the pooling layer, and finally is output through the full-connection layer to obtain the output of the network, and the output is generally called a feature map.
Step 5.2: and (4) carrying out model training and tuning by using a deep learning algorithm to obtain a training model for violation identification.
In this embodiment, the samples in the sample library are as shown in FIG. 8,
the method can be obtained by analyzing the characteristics of the defects and hidden danger targets such as color, shape, texture and the like in the sample library, and has the characteristics of various colors and irregular shapes, and the characteristics increase the identification difficulty. And finally completing the construction of a training and identification verification test set, wherein the sample scale is as follows: about 20 million images containing defect potential and 3 million images without defect potential.
And randomly selecting part of the images marked with defects and hidden dangers as a training set according to different scene requirements, and using the rest images as a test set. The training set and the test set, the scene without the defect hidden danger image and the scene with the defect hidden danger image are not included.
In this example, 90% of samples in each class were randomly selected to construct training samples, and the remaining 10 were used as test samples. And (3) using the classification performance of different backbone network test models, setting the initial learning rate to be 0.01, and reducing the learning rate by 0.1 every 30 epochs by adopting step-down variation. The network size is 224 × 224, the number of training iterations is 100epoch, the weights are initialized randomly, and the loss function is the classification loss and regression loss function in section 4.5.
After 100epoch training, the convergence of the model is shown in fig. 9.
As can be seen from the loss function loss variation curve in FIG. 9, the model has basically converged, the loss of the training sample and the testing sample is reduced to about 0.4, the precision is over 0.95, and a basic network structure is provided for the subsequent inspection image defect and hidden danger target identification algorithm.
In this embodiment, the method of classifying the size of the target sample image in the data set of the sample library is used to analyze the size of the target size: the small size targets are considered with the area of the labeled box smaller than 32 × 32, the large size targets are considered with the area of the labeled box larger than 96 × 96, and the medium size targets with the area between [32 × 32,96 × 96], as shown in the following table.
Target size analysis
The size and the form of each target type are different, the aspect ratio of each target type is approximately distributed in a certain range, and parameters of anchor points in network design are set according to the analyzed aspect ratio, so that a better training model is obtained.
Step 6: inputting the collected images of the construction site into a training model, obtaining the violation type and the violation position by using the training model, outputting the violation defect and hidden danger identification result, sending an alarm signal, adding the obtained images into a sample library, and updating the sample library.
The method of the embodiment can be used for rapidly identifying various violations on the electric power infrastructure construction site by utilizing the training model, can improve the intelligent identification efficiency and accuracy of violation operation, further guides a construction team to rapidly carry out safety protection, can reduce the labor intensity of infrastructure safety management workers, shortens the violation identification management period, and promotes the conversion of the electric power infrastructure safety management mode to the intelligent operation inspection mode.
Example 2:
the embodiment discloses a system for checking violations of regulations on construction site of electric power infrastructure, including:
a sample library establishing module: the system is used for acquiring historical sample images and establishing a sample library;
a data labeling module: the system is used for carrying out data annotation on the sample images with hidden danger in the sample library;
a training model establishing module: the training model is used for carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification;
an identification module: and the system is used for inputting the acquired field images into the training model to obtain the violation type and the violation position.
Example 3:
the embodiment provides electronic equipment, which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor executes the program to realize the violation checking method for the electric power infrastructure construction site, which is described in the embodiment 1.
Example 4:
the embodiment provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer-readable storage medium implements the violation checking method for the electric power infrastructure construction site, which is described in the first embodiment.
The steps involved in the second to fourth embodiments correspond to the first embodiment of the method, and the detailed description thereof can be found in the relevant description of the first embodiment. The term "computer-readable storage medium" should be taken to include a single medium or multiple media containing one or more sets of instructions; it should also be understood to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a processor and that cause the processor to perform any of the methods of the present invention.
Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using general purpose computer means, or alternatively, they can be implemented using program code that is executable by computing means, such that they are stored in memory means for execution by the computing means, or they are separately fabricated into individual integrated circuit modules, or multiple modules or steps of them are fabricated into a single integrated circuit module. The present invention is not limited to any specific combination of hardware and software.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (10)
1. The violation troubleshooting method for the electric power infrastructure construction site is characterized by comprising the following steps of:
acquiring a historical sample image and establishing a sample library;
carrying out data annotation on a sample image with hidden danger in a sample library;
carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification;
and inputting the acquired field image into a training model to obtain the violation type and the violation position.
2. The method for troubleshooting on the violating regulations of the power infrastructure construction site as claimed in claim 1, wherein whether the collected sample image is a sample image with hidden danger or not is judged according to the pre-obtained typical characteristics of the construction hidden danger, and the sample image with hidden danger is subjected to data annotation.
3. The method for troubleshooting on-site regulations for electric power capital construction according to claim 1, characterized in that an annotation file recording format is required by using xml in a PASCAL VOC data set format as a standard, and data annotation is performed on all objects with defects and hidden dangers in the sample image.
4. The method for troubleshooting regulations on the power infrastructure construction site as claimed in claim 1, wherein after the data annotation is performed on the sample image with the hidden danger in the sample library, the data annotation is performed on the typical characteristics of the sample image with the hidden danger in the sample library according to the image characteristics of the hidden danger.
5. The method for troubleshooting on-site regulations for electric power infrastructure construction as claimed in claim 1, wherein in the model training and tuning process, the sample image data is identified based on a fast-RCNN network structure, the training samples and the test samples are randomly selected, and the training model is constructed and tuned by utilizing a deep learning algorithm.
6. The method for troubleshooting on the violations of regulations in the power infrastructure construction site as claimed in claim 5, wherein 90% of sample images of each type of hidden danger are randomly selected to construct training samples, and the remaining 10% of the sample images are used as test samples to construct training models.
7. The method for troubleshooting violations at the power infrastructure construction site as recited in claim 1, wherein the collected site images are input into the training model to obtain the violation type and the violation position, and then the images are added into the sample library to update the sample library.
8. The utility model provides a be used for electric power capital construction job site investigation system violating regulations, its characterized in that includes:
a sample library establishing module: the system is used for acquiring historical sample images and establishing a sample library;
the first data labeling module: the system is used for carrying out data annotation on the sample images with hidden danger in the sample library;
the second data annotation module: the method comprises the steps of performing data annotation on typical characteristics of sample images with hidden dangers in a sample library;
a training model establishing module: the training model is used for carrying out model training and tuning on the sample image subjected to data annotation to obtain a training model for violation identification;
an identification module: and the system is used for inputting the acquired field images into the training model to obtain the violation type and the violation position.
9. An electronic device comprising a memory and a processor and computer instructions stored on the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the steps of any of the methods of claims 1-7.
10. A computer readable storage medium storing computer instructions which, when executed by a processor, perform the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011601489.XA CN112613453A (en) | 2020-12-29 | 2020-12-29 | Method and system for checking violation of regulations on construction site of electric power infrastructure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011601489.XA CN112613453A (en) | 2020-12-29 | 2020-12-29 | Method and system for checking violation of regulations on construction site of electric power infrastructure |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112613453A true CN112613453A (en) | 2021-04-06 |
Family
ID=75249078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011601489.XA Pending CN112613453A (en) | 2020-12-29 | 2020-12-29 | Method and system for checking violation of regulations on construction site of electric power infrastructure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112613453A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113824859A (en) * | 2021-08-17 | 2021-12-21 | 衢州光明电力投资集团有限公司赋腾科技分公司 | Construction hidden danger automatic identification and alarm device violating regulations |
| CN115454138A (en) * | 2022-10-11 | 2022-12-09 | 众芯汉创（北京）科技有限公司 | Construction violation determination method and system based on unmanned aerial vehicle image recognition technology |
| CN116108397A (en) * | 2022-12-22 | 2023-05-12 | 福建亿榕信息技术有限公司 | Electric power field operation violation identification method integrating multi-mode data analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107180504A (en) * | 2017-06-29 | 2017-09-19 | 广东电网有限责任公司佛山供电局 | A kind of the image recognition early warning system and method for the anti-big machinery destruction of transmission line of electricity |
| CN107679495A (en) * | 2017-10-09 | 2018-02-09 | 济南大学 | A kind of detection method of transmission line of electricity periphery activity engineering truck |
| CN110598757A (en) * | 2019-08-23 | 2019-12-20 | 国网山东省电力公司电力科学研究院 | Detection method for hidden danger of construction machinery of power transmission line |
| CN110705414A (en) * | 2019-09-24 | 2020-01-17 | 智洋创新科技股份有限公司 | Power transmission line construction machinery hidden danger detection method based on deep learning |
| CN110826514A (en) * | 2019-11-13 | 2020-02-21 | 国网青海省电力公司海东供电公司 | Construction site violation intelligent identification method based on deep learning |
| CN111695493A (en) * | 2020-06-10 | 2020-09-22 | 国网山东省电力公司电力科学研究院 | Method and system for detecting hidden danger of power transmission line |
-
2020
- 2020-12-29 CN CN202011601489.XA patent/CN112613453A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107180504A (en) * | 2017-06-29 | 2017-09-19 | 广东电网有限责任公司佛山供电局 | A kind of the image recognition early warning system and method for the anti-big machinery destruction of transmission line of electricity |
| CN107679495A (en) * | 2017-10-09 | 2018-02-09 | 济南大学 | A kind of detection method of transmission line of electricity periphery activity engineering truck |
| CN110598757A (en) * | 2019-08-23 | 2019-12-20 | 国网山东省电力公司电力科学研究院 | Detection method for hidden danger of construction machinery of power transmission line |
| CN110705414A (en) * | 2019-09-24 | 2020-01-17 | 智洋创新科技股份有限公司 | Power transmission line construction machinery hidden danger detection method based on deep learning |
| CN110826514A (en) * | 2019-11-13 | 2020-02-21 | 国网青海省电力公司海东供电公司 | Construction site violation intelligent identification method based on deep learning |
| CN111695493A (en) * | 2020-06-10 | 2020-09-22 | 国网山东省电力公司电力科学研究院 | Method and system for detecting hidden danger of power transmission line |
Non-Patent Citations (1)
| Title |
|---|
| 张骥 等: "基于深度学习的输电线路外破图像识别技术", 《计算机系统应用》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113824859A (en) * | 2021-08-17 | 2021-12-21 | 衢州光明电力投资集团有限公司赋腾科技分公司 | Construction hidden danger automatic identification and alarm device violating regulations |
| CN115454138A (en) * | 2022-10-11 | 2022-12-09 | 众芯汉创（北京）科技有限公司 | Construction violation determination method and system based on unmanned aerial vehicle image recognition technology |
| CN115454138B (en) * | 2022-10-11 | 2023-04-18 | 众芯汉创（北京）科技有限公司 | Construction violation determination method and system based on unmanned aerial vehicle image recognition technology |
| CN116108397A (en) * | 2022-12-22 | 2023-05-12 | 福建亿榕信息技术有限公司 | Electric power field operation violation identification method integrating multi-mode data analysis |
| CN116108397B (en) * | 2022-12-22 | 2024-01-09 | 福建亿榕信息技术有限公司 | Electric power field operation violation identification method integrating multi-mode data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112613453A (en) | Method and system for checking violation of regulations on construction site of electric power infrastructure | |
| CN110059694B (en) | Intelligent identification method for character data in complex scene of power industry | |
| CN111784685A (en) | Power transmission line defect image identification method based on cloud edge cooperative detection | |
| CN112115927B (en) | Intelligent machine room equipment identification method and system based on deep learning | |
| CN110070008A (en) | Bridge disease identification method adopting unmanned aerial vehicle image | |
| CN112613454A (en) | Electric power infrastructure construction site violation identification method and system | |
| CN112528979B (en) | Transformer substation inspection robot obstacle distinguishing method and system | |
| CN115272204A (en) | Bearing surface scratch detection method based on machine vision | |
| CN111899219A (en) | Image identification method and system for power transmission line machine patrol | |
| CN113723626A (en) | Subway line protection inspection method and device, computer equipment and storage medium | |
| CN115578326A (en) | Road disease identification method, system, equipment and storage medium | |
| CN115223043A (en) | Strawberry defect detection method and device, computer equipment and storage medium | |
| CN114863118A (en) | Self-learning identification system and method based on external hidden danger of power transmission line | |
| CN114724140A (en) | Strawberry maturity detection method and device based on YOLO V3 | |
| CN117114420B (en) | Image recognition-based industrial and trade safety accident risk management and control system and method | |
| CN110618129A (en) | Automatic power grid wire clamp detection and defect identification method and device | |
| CN111915565B (en) | Method for analyzing cracks of porcelain insulator of power transmission and transformation line in real time based on YOLACT algorithm | |
| CN113205511A (en) | Electronic component batch information detection method and system based on deep neural network | |
| CN112613560A (en) | Method for identifying front opening and closing damage fault of railway bullet train head cover based on Faster R-CNN | |
| CN111709991B (en) | Railway tool detection method, system, device and storage medium | |
| CN109325441B (en) | Method for identifying insulator object of power transmission line | |
| CN113554610A (en) | Photovoltaic module operation state detection method and application device thereof | |
| CN113361520A (en) | Transmission line equipment defect detection method based on sample offset network | |
| CN110781758A (en) | Dynamic video monitoring method and device for abnormal pantograph structure | |
| CN117274843B (en) | Unmanned aerial vehicle front end defect identification method and system based on lightweight edge calculation |
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: 20210406 |
|
| RJ01 | Rejection of invention patent application after publication |