CN110569717A - partial discharge detection method, device, system, equipment and readable storage medium - Google Patents
partial discharge detection method, device, system, equipment and readable storage medium Download PDFInfo
- Publication number
- CN110569717A CN110569717A CN201910681795.XA CN201910681795A CN110569717A CN 110569717 A CN110569717 A CN 110569717A CN 201910681795 A CN201910681795 A CN 201910681795A CN 110569717 A CN110569717 A CN 110569717A
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- region
- image
- network model
- detected
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 84
- 238000003860 storage Methods 0.000 title claims abstract description 15
- 230000001133 acceleration Effects 0.000 claims abstract description 71
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 69
- 230000002776 aggregation Effects 0.000 claims abstract description 54
- 238000004220 aggregation Methods 0.000 claims abstract description 54
- 238000000034 method Methods 0.000 claims description 58
- 238000012549 training Methods 0.000 claims description 33
- 238000010586 diagram Methods 0.000 claims description 21
- 238000007689 inspection Methods 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 19
- 230000002146 bilateral effect Effects 0.000 claims description 16
- 238000001914 filtration Methods 0.000 claims description 16
- 238000011176 pooling Methods 0.000 claims description 15
- 238000004891 communication Methods 0.000 claims description 9
- 238000003062 neural network model Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 description 19
- 230000008569 process Effects 0.000 description 15
- 238000013500 data storage Methods 0.000 description 14
- 230000009466 transformation Effects 0.000 description 10
- 238000009413 insulation Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 6
- 238000005457 optimization Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Entrepreneurship & Innovation (AREA)
- Biophysics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
the application provides a partial discharge detection method, a device, a system, equipment and a readable storage medium, wherein the partial discharge detection method comprises the following steps: acquiring an image of a region to be detected; denoising the image of the region to be detected through a context aggregation network model to obtain a denoised image; judging whether the to-be-detected region has partial discharge or not according to the de-noised image; and if the to-be-detected region has partial discharge, identifying the discharge region and the severity of the partial discharge of the to-be-detected region through an acceleration region convolutional neural network model according to the de-noised image. The partial discharge detection method can improve the efficiency and accuracy of partial discharge detection.
Description
Technical Field
The application relates to the technical field of transformer substation operation and maintenance, in particular to a partial discharge detection method, a device, a system, equipment and a readable storage medium.
Background
along with the development of economy and the improvement of life quality, the demands of residents and factories on power consumption are increased, and the normal and stable operation of power transformation equipment is particularly important, because the power transformation equipment can generate great influence on the power consumption of the residents and the factories.
the faults of the power transformation equipment are mainly classified into three types, namely mechanical faults, heating faults and insulation faults. Insulation faults are the most common type of faults, and partial discharge faults are one of the common faults in insulation faults. Although partial discharge does not immediately break down the insulation, once it occurs, it will continuously erode surrounding equipment, resulting in insulation cracking of the transformer and loss of electrical energy. Therefore, timely detection of partial discharge is very important for failure of the power transformation equipment.
in the conventional technology, the detection of partial discharge is mainly performed by manual periodic inspection. This method is inefficient.
Disclosure of Invention
in view of the foregoing, it is desirable to provide a partial discharge detection method, apparatus, system, device and readable storage medium.
in a first aspect, a partial discharge detection method, the method comprising:
Acquiring an image of a region to be detected;
denoising the image of the region to be detected through a context aggregation network model to obtain a denoised image;
Judging whether the to-be-detected region has partial discharge or not according to the de-noised image;
and if the to-be-detected region has partial discharge, identifying the discharge region and the severity of the partial discharge of the to-be-detected region through an acceleration region convolutional neural network model according to the de-noised image.
In one embodiment, the identifying the discharge region and the severity of the partial discharge of the region to be detected through the acceleration region convolutional neural network model according to the denoised image includes:
extracting a characteristic map of the de-noised image by using the convolution layer and the pooling layer;
generating a suggestion area through an area suggestion network according to the feature map;
combining the suggested region and the feature map through the pooling layer to extract a suggested feature map;
And evaluating the suggested characteristic diagram through the full connection layer to obtain the severity of the partial discharge of the region to be detected, and obtaining the discharge region of the partial discharge of the region to be detected by using bounding box regression.
in one embodiment, the method further comprises:
Constructing a convolutional neural network model of a preset acceleration region;
And training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
In one embodiment, the training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model includes:
acquiring partial discharge images of various known discharge areas and severity degrees, and marking to obtain a plurality of partial discharge image samples;
and inputting the plurality of partial discharge image samples into the preset acceleration region convolutional neural network model, and training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
In one embodiment, the context aggregation network model includes at least an input layer, a plurality of feature layers, and an output layer.
In one embodiment, the method further comprises:
Establishing a preset context aggregation network model;
And training the preset context aggregation network model to obtain the upper and lower aggregation network models.
In one embodiment, the training the preset context aggregation network model to obtain the upper and lower aggregation network models includes:
acquiring a sample image set and a bilateral filtering image set corresponding to the sample image set;
and inputting the sample image set and the bilateral filtering image set into the preset context aggregation network model, and learning the characteristic relation between the sample image set and the bilateral filtering image to obtain the upper and lower aggregation network models.
in one embodiment, the method further comprises:
And if the area to be detected has partial discharge, outputting alarm information.
in one embodiment, after identifying the discharge area and the severity of the partial discharge of the area to be detected through an accelerated area convolutional neural network model according to the denoised image, the method further includes:
And marking the de-noised image according to the discharge area and the severity of the partial discharge of the area to be detected to obtain a detection image sample.
in one embodiment, the method further comprises:
And inputting a detection image sample into the acceleration region convolution neural network model, and optimizing the acceleration region convolution neural network model.
in a second aspect, a partial discharge detection apparatus includes:
the image acquisition module is used for acquiring an image of a region to be detected;
The denoising module is used for denoising the image of the region to be detected through the context aggregation network model to obtain a denoised image;
The partial discharge judging module is used for judging whether partial discharge exists in the to-be-detected region or not according to the de-noised image;
And the result identification module is used for identifying the discharge area and the severity of the partial discharge of the area to be detected through an acceleration area convolutional neural network model according to the de-noised image if the area to be detected has the partial discharge.
In a third aspect, a partial discharge detection system includes:
The inspection robot is used for acquiring an image of a region to be detected;
the ground base station is in communication connection with the inspection robot and comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the steps of the method.
in a fourth aspect, a computer device comprises a memory storing a computer program and a processor implementing the steps of the method as described above when the processor executes the computer program.
In a fifth aspect, a computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method as described above.
according to the partial discharge detection method, the device, the system, the equipment and the readable storage medium, the image of the area to be detected is denoised through the context aggregation network model to obtain the denoised image, whether partial discharge exists in the area to be detected is judged according to the denoised image, and if partial discharge exists in the area to be detected, the discharge area and the severity of the partial discharge in the area to be detected are identified through the acceleration area convolutional neural network model according to the denoised image. The partial discharge detection method, the device, the system, the equipment and the readable storage medium provided by the embodiment can realize automatic identification and detection of partial discharge, manual inspection is not needed, and the identification efficiency is high. Meanwhile, the partial discharge detection method, the device, the system, the equipment and the readable storage medium provided by the embodiment denoise the image of the region to be detected through the context aggregation network model, so that the edge definition can be kept while noise is filtered, and the denoising effect is good. In addition, the partial discharge detection method, the device, the system, the equipment and the readable storage medium identify the discharge area and the severity of partial discharge of the area to be detected through the acceleration area convolutional neural network model according to the de-noised image, abandon the traditional sliding window and selective search method, and directly use the area suggestion network to generate the detection frame, so that the discharge area of the partial discharge is detected, the detection efficiency is improved, and the accuracy of the detection result is improved.
drawings
Fig. 1 is a schematic structural diagram of a partial discharge detection system in an application scenario according to an embodiment;
FIG. 2 is a flow diagram of a partial discharge detection method according to an embodiment;
FIG. 3 is a flowchart illustrating the operation of an accelerated regional convolutional neural network model according to an embodiment;
FIG. 4 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 5 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 6 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 7 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 8 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 9 is a flow diagram of a partial discharge detection method according to one embodiment;
FIG. 10 is a flow diagram of a method for partial discharge detection according to one embodiment;
fig. 11 is a schematic structural diagram of a partial discharge detection apparatus according to an embodiment;
Fig. 12 is a schematic structural diagram of a partial discharge detection apparatus according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The partial discharge detection method can be used for detecting partial discharge phenomena of equipment which is likely to generate partial discharge, such as high-voltage electrical equipment. For example, the partial discharge detection method provided by the application can be used for detecting the partial discharge phenomenon of the power transformation equipment of the substation. The embodiments provided in the present application are all described taking detection of partial discharge of a power transformation apparatus as an example. As shown in fig. 1, the partial discharge detection method provided in the embodiment of the present application may be specifically applied to the system using partial discharge detection as shown in fig. 1. The partial discharge detection system includes an inspection robot 100 and a ground base station 200. The inspection robot includes an inspection vehicle body and an image acquisition device 110. The image pickup device 110 is mounted on the inspection vehicle body. The image capturing device 110 is used to capture image information of a power transformation device, and the like, and the image capturing device may be, but is not limited to, an ultraviolet imager. The structure, the model and the like of the ultraviolet imager are not limited at all and can be selected according to actual requirements. The inspection vehicle body of the inspection robot 100 may include a data storage and processing module 120, a communication module 130, and a motion control module 140. The image capture device 110, the motion control module 140, and the communication module 130 are all connected to the data storage and processing module 120. The motion control module 140 is used to control the motion trajectory of the inspection robot 100. The communication module 130 is used for communication with other devices. In addition, the partial discharge detection system may further include a ground base station 200. The ground base station 200 communicates with the inspection robot 100 through the communication module 130. The data storage and processing module 120 includes a processor and a memory, the processor being capable of processing computer programs. The processor may be a Central Processing Unit (CPU) or a Micro Controller Unit (MCU). The ground base station may be a computer device that may be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The data storage and processing module 120 and the ground base station are each capable of executing computer programs to implement some or all of the steps of the partial discharge detection method provided herein.
those skilled in the art will appreciate that the configuration shown in fig. 1 is a block diagram of only a portion of the configuration associated with the present application and does not constitute a limitation on the devices to which the present application may be applied, and that a particular device may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
the following describes in detail the technical solutions of the present application and how the technical solutions of the present application solve the above technical problems by embodiments and with reference to the drawings. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 2 is a flowchart of a partial discharge detection method according to an embodiment. The execution subject of this embodiment may be the data storage and processing module 120 shown in fig. 1, may also be the ground base station 200, and may also be executed jointly by the data storage and processing module 120 and the ground base station 200, that is, the execution subject is a partial discharge detection system. The following embodiments all take the implementation subject as an example of a partial discharge detection system, and describe the partial discharge detection method. The present embodiment relates to a specific process of identifying whether a partial discharge exists and the location and severity of the partial discharge by a partial discharge detection system. As shown in fig. 2, the method includes:
And S10, acquiring the image of the area to be detected.
The image of the area to be detected refers to an image of a part or all of the area of the power transformation equipment to be detected. The motion control module of the partial discharge detection system controls the inspection robot to inspect the vehicle body, the inspection robot carries the ultraviolet imager to walk along the planned path, and the ultraviolet imager images the substation equipment to be detected and transmits the substation equipment to the data storage and processing module.
S20, denoising the image of the region to be detected through a Context Aggregation Network (CAN) model to obtain a denoised image.
The image obtained by the ultraviolet imager is influenced by factors such as distance, gain and visual angle in the imaging process, so that noise exists. The preprocessing of the image of the area to be detected can eliminate redundant and interference information, obtain useful information and improve the efficiency and the precision of image recognition, segmentation and extraction. The context aggregation network model may be a pre-trained model for denoising the image. The data storage and processing module carries out denoising processing on the to-be-detected region image collected by the ultraviolet imager based on the context aggregation network model, removes noise points to obtain a denoised image, can keep edge definition while filtering noise, and has good denoising effect.
and S30, judging whether the local discharge exists in the region to be detected according to the denoising image.
And judging whether the partial discharge exists in the region to be detected or not by a data storage and processing module of the partial discharge detection system according to the de-noised image obtained in the S20. There are various methods for determining and identifying partial discharge by the partial discharge detection system, such as comparing historical images, finding image bright spots, and the like. Taking the searching of the image bright points as an example, the data storage and processing module searches in the de-noised image according to a preset algorithm to determine whether bright point information exists in the image, and if the bright point information exists, it is indicated that partial discharge exists in the area to be detected and insulation fault exists in the power transformation equipment; if the denoised image does not have the bright spot information, the fact that the local discharge does not exist in the to-be-detected area is indicated, and the insulation fault does not exist in the power transformation equipment.
And S40, if the partial discharge exists in the region to be detected, identifying the discharge region and the severity of the partial discharge in the region to be detected through an acceleration region convolutional neural network model (fast-RCNN) according to the de-noised image.
And if the to-be-detected region has partial discharge, the data storage and processing module further judges and identifies the de-noised image, and identifies the partial discharge region and the partial discharge severity. Of course, if the partial discharge exists in the area to be detected, the data storage and processing module can also transmit the denoising image to the ground base station through the communication module, and the ground base station inputs the denoising image into a pre-trained acceleration area convolutional neural network model for processing so as to identify the specific discharge position and the discharge severity of the partial discharge in the area to be detected. The discharge area and the severity of partial discharge of the area to be detected are identified through the acceleration area Convolutional neural network model, the whole process is completed through a Convolutional Neural Network (CNN), an area suggestion network (RPN) is directly used for generating a detection frame, a target area is detected, and the discharge area of the partial discharge of the area to be detected is obtained.
the partial discharge detection method provided in this embodiment denoises an image of a to-be-detected region through a context aggregation network model to obtain a denoised image, determines whether a partial discharge exists in the to-be-detected region according to the denoised image, and identifies a discharge region and a severity of the partial discharge in the to-be-detected region through an acceleration region convolutional neural network model according to the denoised image if the partial discharge exists in the to-be-detected region. The partial discharge detection method provided by the embodiment can realize automatic identification and detection of partial discharge, does not need manual inspection, and is high in identification efficiency. Meanwhile, the method provided by the embodiment denoises the image of the region to be detected through the context aggregation network model, so that the edge definition can be kept while the noise is filtered, and the denoising effect is good. In addition, the method identifies the discharge area and the severity of partial discharge of the area to be detected through an acceleration area convolutional neural network model according to the denoised image, abandons the traditional sliding window and selective search (SELECTIVESEARCH, SS) method, and directly uses RPN to generate a detection frame, so that the discharge area of the partial discharge is detected, the detection efficiency is improved, and the accuracy of the detection result is improved.
referring to fig. 3 and fig. 4, in an embodiment, fig. 3 is a flowchart of a possible implementation manner of S40 in fig. 2, and the embodiment relates to a possible implementation manner of identifying a discharging area and a severity of partial discharge in an area to be detected through an acceleration area convolutional neural network model according to a denoised image if the area to be detected has partial discharge. Specifically, the acceleration region convolutional neural network model at least includes a convolutional layer, a pooling layer, a risk priority network, and a full link layer, and S40 includes:
and S410, extracting a feature map of the de-noised image by using the convolution layer and the pooling layer.
and the fast-RCNN utilizes the convolution layer and the draft layer to extract the characteristic diagram of the input de-noised image for the subsequent RPN network and the full connection layer. In one embodiment, padding is added before convolution to fill the border, so that the size of the convolved picture is ensured to be unchanged.
And S420, generating a suggestion area through an area suggestion network according to the feature diagram.
and generating a region suggestion by the RPN, scoring the anchors, evaluating whether the anchors belong to the foreground or the background, and then performing more accurate correction on the positions and the types of the anchors by using bounding box regression to generate a suggested region.
In a specific embodiment, S420 is implemented by the following process:
1) obtaining a characteristic diagram through convolution and pooling of S410, generating anchors on the characteristic diagram, wherein each point is provided with 9 anchors;
2) Further extracting features from the convolution layer with kernel size 3, stride 1 and padding 1, and outputting a feature map with a constant size, which is called an RPN feature map;
3) on the RPN feature map, prediction output is performed using two convolution layers whose kernel size is 1 × 1, and the number of channels represents the predicted class score and offset. And one of the two output channels obtains classification scores of the foreground and the background through sofxmax, wherein the scores represent the probability of the foreground and the probability of the background. And the other path predicts the other four numerical values of the anchor frame through frame regression, namely the coordinate value of the upper left corner and the frame length and width, so as to obtain a more accurate suggested region proposl and generate the suggested region.
And S430, combining the suggested region and the feature map through the pooling layer, and extracting the suggested feature map.
and the ROI pooling layer combines the suggested region obtained in the step S420 and the feature map obtained in the step S410 to extract a suggested feature map.
and S440, evaluating the suggested characteristic diagram through the full connection layer to obtain the severity of the partial discharge of the region to be detected, and obtaining the discharge region of the partial discharge of the region to be detected by using the regression of the bounding box.
And the full connection layer evaluates the recommended categories by using the recommended feature map obtained in the step S430 to obtain the severity of partial discharge of the region to be detected. And meanwhile, the bounding box regression is used again to obtain a more accurate position, namely a discharge region of the partial discharge of the region to be detected.
in this embodiment, the feature map of the denoised image is extracted by using the convolutional layer and the pooling layer, the proposed region is generated through the region suggestion network according to the feature map, the proposed region and the feature map are combined through the pooling layer, the proposed feature map is extracted, the proposed feature map is evaluated through the full connection layer, the severity of the partial discharge of the region to be detected is obtained, the boundary frame regression is used, the discharge region of the partial discharge of the region to be detected is obtained, the recognition efficiency is high, and the accuracy is high. According to the method provided by the embodiment, the recommended region is generated through the region recommendation network according to the characteristic diagram, the traditional sliding window and selective search method are abandoned, and the RPN is directly used for generating the detection frame, so that the discharge region of the partial discharge is detected, the detection efficiency is improved, and the accuracy of the detection result is improved.
Referring to fig. 5, the embodiment relates to a specific process for constructing and training an acceleration region convolutional neural network model. Specifically, the method further comprises:
and S510, constructing a convolutional neural network model of a preset acceleration region.
The preset acceleration region convolutional neural network model is a preliminarily established convolutional neural network model which is not trained.
s520, training the convolutional neural network model of the preset acceleration region to obtain the convolutional neural network model of the acceleration region.
And training the convolutional neural network model in the acceleration region, wherein in the training process, each part of the model not only learns how to complete the task of the model, but also independently learns how to cooperate with each other.
Referring to fig. 6, in an embodiment, the training process of the acceleration region convolutional neural network model includes the following processes, i.e., S520 includes:
and S521, acquiring partial discharge images of various known discharge areas and severity degrees, and marking to obtain a plurality of partial discharge image samples.
Partial discharge images of known discharge areas and severity can be acquired by a patrol robot or manually. Collected images are labeled and classified, and the content of the label may include, but is not limited to, acquisition time, acquisition location, discharge area, and the like. And the severity of the discharges of the images are classified and labeled. Therefore, a plurality of partial discharge image samples are obtained and stored in a data storage and processing module of the inspection robot or a memory of the ground base station.
s522, inputting the partial discharge image samples into a preset acceleration region convolutional neural network model, and training the preset acceleration region convolutional neural network model to obtain an acceleration region convolutional neural network model.
And inputting the partial discharge image sample into a preset acceleration region convolutional neural network model, comparing the result output by the model with a known mark, continuously adjusting the model parameters, training and improving the model structure, and finally obtaining the acceleration region convolutional neural network model. It should be noted that the greater the number of partial discharge image samples, the more training the model, and the more accurate the obtained acceleration region convolution neural network model.
In the embodiment, the acceleration region convolutional neural network model is obtained by training the preset acceleration region convolutional neural network model, so that the accuracy of the model is improved, and the accuracy of the local discharge recognition is improved.
Referring to fig. 7, in an embodiment, the method further includes:
S610, marking the de-noised image according to the discharge area and the severity of the partial discharge of the area to be detected to obtain a detection image sample.
And marking the denoised image according to the result identified in the step S50, wherein the marking content comprises but is not limited to image acquisition time, acquisition position, discharge area and severity, obtaining a detection image sample, and adding the detection image sample into the original partial discharge image database as a new partial discharge image sample.
And S620, inputting the detected image sample into the acceleration region convolution neural network model, and optimizing the acceleration region convolution neural network model.
and inputting the detection image sample into the acceleration region convolutional neural network model for further training and optimization, wherein the training and optimization process is similar to that of S522 and is not described herein again.
in the embodiment, the de-noising image is marked according to the discharge area and the severity of the partial discharge of the area to be detected to obtain the detection image sample, the detection image sample is input into the acceleration area convolutional neural network model, and the acceleration area convolutional neural network model is optimized, so that the accuracy of the acceleration area convolutional neural network model is further improved, and the accuracy of the subsequent partial discharge detection result is improved.
The context aggregation network model and the training process thereof are further described below with reference to the embodiments. In one embodiment, the context aggregation network model includes at least an input layer, a plurality of feature layers, and an output layer.
Referring to fig. 8, in an embodiment, the process of constructing and training the context aggregation network model is as follows, that is, the method further includes:
s710, establishing a preset context aggregation network model.
The preset context aggregation network model refers to a preliminarily established and untrained network model.
S720, training the preset context aggregation network model to obtain an upper aggregation network model and a lower aggregation network model.
referring to fig. 9, the training process for the preset context aggregation network model may be as follows, that is, S720 includes:
s721, acquiring a sample image set and a bilateral filtering image set corresponding to the sample image set;
and S722, inputting the sample image set and the bilateral filtering image set into a preset context aggregation network model, and learning the characteristic relation between the sample image set and the bilateral filtering image to obtain an upper aggregation network model and a lower aggregation network model.
In this embodiment, the upper aggregation network model and the lower aggregation network model are obtained by obtaining a sample image set and a bilateral filtering image set corresponding to the sample image set, inputting the sample image set and the bilateral filtering image set to the preset context aggregation network model, and learning a characteristic relationship between the sample image set and the bilateral filtering image. The trained model does not need to run an operator of bilateral filtering, so that the processing time of the image is reduced, and a better operation result is achieved. In addition, the context aggregation network model is obtained by training the preset context aggregation network model, and the accuracy of the model is improved, so that the denoising accuracy is improved, and the accuracy of partial discharge detection is improved.
referring to fig. 10, in an embodiment, the method further includes:
and S80, if the partial discharge exists in the area to be detected, outputting alarm information.
If the judgment result of S30 shows that partial discharge exists in the area to be detected and the equipment to be detected has an insulation fault, the data storage and processing module of the inspection robot outputs alarm information to remind the user and improve the user experience.
It should be understood that, although the steps in the flowcharts of fig. 2-3 and 5-10 are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps of fig. 2-3, 5-10 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential.
referring to fig. 11, in one embodiment, a partial discharge detection apparatus 30 is provided, which includes: an image acquisition module 310, a denoising module 320, a partial discharge judgment module 330, and a result identification module 340, wherein:
The image acquisition module 310 is configured to acquire an image of a region to be detected;
The denoising module 320 is configured to denoise the to-be-detected region image through the context aggregation network model to obtain a denoised image;
The partial discharge judgment module 330 is configured to judge whether a to-be-detected region has partial discharge according to the denoised image;
And the result identification module 340 is configured to identify a discharge region and a severity of the partial discharge in the region to be detected through an acceleration region convolutional neural network model according to the denoised image if the region to be detected has the partial discharge.
In one embodiment, the acceleration region convolutional neural network model at least includes a convolutional layer, a pooling layer, a risk priority network, and a full link layer, and the result identification module 340 is further specifically configured to extract a feature map of the denoised image by using the convolutional layer and the pooling layer; generating a suggestion area through an area suggestion network according to the feature map; combining the suggested region and the feature map through the pooling layer to extract a suggested feature map; and evaluating the suggested characteristic diagram through the full connection layer to obtain the severity of the partial discharge of the region to be detected, and obtaining the discharge region of the partial discharge of the region to be detected by using bounding box regression.
Referring to fig. 12, in an embodiment, the partial discharge detection apparatus 30 further includes a first model building training module 350, configured to build a convolutional neural network model in a preset acceleration region; and training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
in one embodiment, the first model building training module 350 is specifically configured to obtain partial discharge images of various known discharge regions and severity degrees, and mark the partial discharge images to obtain a plurality of partial discharge image samples; and inputting the plurality of partial discharge image samples into the preset acceleration region convolutional neural network model, and training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
in one embodiment, the context aggregation network model includes at least an input layer, a plurality of feature layers, and an output layer.
In one embodiment, the partial discharge detection apparatus 30 further includes a second model building training module 360, configured to build a preset context aggregation network model; and training the preset context aggregation network model to obtain the upper and lower aggregation network models.
in an embodiment, the second model construction training module 360 is further specifically configured to obtain a sample image set and a bilateral filtered image set corresponding to the sample image set; and inputting the sample image set and the bilateral filtering image set into the preset context aggregation network model, and learning the characteristic relation between the sample image set and the bilateral filtering image to obtain the upper and lower aggregation network models.
In one embodiment, the partial discharge detection apparatus 30 further includes an alarm output module 370, configured to output an alarm message if the area to be detected has partial discharge.
in an embodiment, the partial discharge detection apparatus 30 further includes a model optimization module 380, configured to mark the denoised image according to the discharge area and the severity of the partial discharge in the area to be detected, so as to obtain a detection image sample.
In one embodiment, the model optimization module 380 is further configured to input the detection image sample into the acceleration region convolutional neural network model, and optimize the acceleration region convolutional neural network model.
for specific limitations of the partial discharge detection device 30, reference may be made to the above limitations of a method for identifying an optical cable intrusion construction event, and details thereof are not repeated herein. The modules in the partial discharge detection apparatus 30 may be implemented in whole or in part by software, hardware, or a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
An embodiment of the present application provides a partial discharge detection system, including:
The inspection robot is used for acquiring an image of a region to be detected;
The ground base station is in communication connection with the inspection robot and comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the steps of the method.
the structure and advantageous effects of the partial discharge detection system are as described above, and are not described herein again.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
Acquiring an image of a region to be detected;
denoising the image of the region to be detected through a context aggregation network model to obtain a denoised image;
Judging whether the to-be-detected region has partial discharge or not according to the de-noised image;
And if the to-be-detected region has partial discharge, identifying the discharge region and the severity of the partial discharge of the to-be-detected region through an acceleration region convolutional neural network model according to the de-noised image.
The implementation principle and technical effect of the computer device provided by the above embodiment are similar to those of the above method embodiment, and are not described herein again.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, the computer program, when executed by a processor, further implementing the steps of:
acquiring an image of a region to be detected;
Denoising the image of the region to be detected through a context aggregation network model to obtain a denoised image;
Judging whether the to-be-detected region has partial discharge or not according to the de-noised image;
and if the to-be-detected region has partial discharge, identifying the discharge region and the severity of the partial discharge of the to-be-detected region through an acceleration region convolutional neural network model according to the de-noised image.
The implementation principle and technical effect of the computer-readable storage medium provided by the above embodiments are similar to those of the above method embodiments, and are not described herein again.
it will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link (SLDRAM), Rambus (Rambus) direct RAM (RDRAM), direct bused dynamic RAM (DRDRAM), and bused dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
the above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (14)
1. A partial discharge detection method, the method comprising:
Acquiring an image of a region to be detected;
denoising the image of the region to be detected through a context aggregation network model to obtain a denoised image;
Judging whether the to-be-detected region has partial discharge or not according to the de-noised image;
And if the to-be-detected region has partial discharge, identifying the discharge region and the severity of the partial discharge of the to-be-detected region through an acceleration region convolutional neural network model according to the de-noised image.
2. the method according to claim 1, wherein the acceleration region convolutional neural network model at least comprises a convolutional layer, a pooling layer, a risk priority network and a full link layer, and the identifying the discharge region and the severity of the partial discharge of the region to be detected through the acceleration region convolutional neural network model according to the denoised image comprises:
extracting a characteristic map of the de-noised image by using the convolution layer and the pooling layer;
Generating a suggestion area through an area suggestion network according to the feature map;
Combining the suggested region and the feature map through the pooling layer to extract a suggested feature map;
and evaluating the suggested characteristic diagram through the full connection layer to obtain the severity of the partial discharge of the region to be detected, and obtaining the discharge region of the partial discharge of the region to be detected by using bounding box regression.
3. The method of claim 1, further comprising:
constructing a convolutional neural network model of a preset acceleration region;
And training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
4. The method of claim 3, wherein the training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model comprises:
acquiring partial discharge images of various known discharge areas and severity degrees, and marking to obtain a plurality of partial discharge image samples;
And inputting the plurality of partial discharge image samples into the preset acceleration region convolutional neural network model, and training the preset acceleration region convolutional neural network model to obtain the acceleration region convolutional neural network model.
5. The method of claim 1, wherein the context aggregation network model comprises at least an input layer, a plurality of feature layers, and an output layer.
6. the method of claim 1, further comprising:
establishing a preset context aggregation network model;
and training the preset context aggregation network model to obtain the upper and lower aggregation network models.
7. the method according to claim 6, wherein the training the preset context aggregation network model to obtain the upper and lower aggregation network models comprises:
Acquiring a sample image set and a bilateral filtering image set corresponding to the sample image set;
And inputting the sample image set and the bilateral filtering image set into the preset context aggregation network model, and learning the characteristic relation between the sample image set and the bilateral filtering image to obtain the upper and lower aggregation network models.
8. the method of claim 1, further comprising:
And if the area to be detected has partial discharge, outputting alarm information.
9. the method according to claim 1, wherein after identifying the discharge area and the severity of the partial discharge of the area to be detected by the acceleration area convolutional neural network model according to the de-noised image, the method further comprises:
And marking the de-noised image according to the discharge area and the severity of the partial discharge of the area to be detected to obtain a detection image sample.
10. The method of claim 9, further comprising:
and inputting a detection image sample into the acceleration region convolution neural network model, and optimizing the acceleration region convolution neural network model.
11. A partial discharge detection apparatus, comprising:
The image acquisition module is used for acquiring an image of a region to be detected;
The denoising module is used for denoising the image of the region to be detected through the context aggregation network model to obtain a denoised image;
the partial discharge judging module is used for judging whether partial discharge exists in the to-be-detected region or not according to the de-noised image;
and the result identification module is used for identifying the discharge area and the severity of the partial discharge of the area to be detected through an acceleration area convolutional neural network model according to the de-noised image if the area to be detected has the partial discharge.
12. A partial discharge detection system, comprising:
the inspection robot is used for acquiring an image of a region to be detected;
A ground base station in communication with the inspection robot, the ground base station including a memory and a processor, the memory storing a computer program, the processor implementing the steps of the method of any one of claims 1 to 10 when executing the computer program.
13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor realizes the steps of the method of any one of claims 1 to 10 when executing the computer program.
14. a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910681795.XA CN110569717A (en) | 2019-07-26 | 2019-07-26 | partial discharge detection method, device, system, equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910681795.XA CN110569717A (en) | 2019-07-26 | 2019-07-26 | partial discharge detection method, device, system, equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110569717A true CN110569717A (en) | 2019-12-13 |
Family
ID=68773339
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910681795.XA Pending CN110569717A (en) | 2019-07-26 | 2019-07-26 | partial discharge detection method, device, system, equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110569717A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191313A (en) * | 2021-05-20 | 2021-07-30 | 国能大渡河沙坪发电有限公司 | Video stream discharge identification method and device based on hydraulic power plant and computer equipment |
| CN115184744A (en) * | 2022-06-27 | 2022-10-14 | 上海格鲁布科技有限公司 | GIS ultrahigh frequency discharge signal detection device and method based on fast-RCNN |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106291278A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of partial discharge of switchgear automatic testing method based on many visual systemes |
| CN106556781A (en) * | 2016-11-10 | 2017-04-05 | 华乘电气科技（上海）股份有限公司 | Shelf depreciation defect image diagnostic method and system based on deep learning |
| US20170206431A1 (en) * | 2016-01-20 | 2017-07-20 | Microsoft Technology Licensing, Llc | Object detection and classification in images |
| CN108921839A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN109596955A (en) * | 2018-12-30 | 2019-04-09 | 国网北京市电力公司 | Shelf depreciation state determines method and device |
| CN208780775U (en) * | 2018-08-21 | 2019-04-23 | Abb瑞士股份有限公司 | Electric discharge for switchgear monitors system and switchgear |
| CN109740586A (en) * | 2018-12-19 | 2019-05-10 | 南京华科和鼎信息科技有限公司 | A kind of anti-dazzle certificate automatic reading system and method |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
| CN109871746A (en) * | 2018-12-30 | 2019-06-11 | 国网辽宁省电力有限公司电力科学研究院 | A kind of mode identification method based on the diagnosis of GIS insulation defect |
-
2019
- 2019-07-26 CN CN201910681795.XA patent/CN110569717A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170206431A1 (en) * | 2016-01-20 | 2017-07-20 | Microsoft Technology Licensing, Llc | Object detection and classification in images |
| CN106291278A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of partial discharge of switchgear automatic testing method based on many visual systemes |
| CN106556781A (en) * | 2016-11-10 | 2017-04-05 | 华乘电气科技（上海）股份有限公司 | Shelf depreciation defect image diagnostic method and system based on deep learning |
| CN108921839A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN208780775U (en) * | 2018-08-21 | 2019-04-23 | Abb瑞士股份有限公司 | Electric discharge for switchgear monitors system and switchgear |
| CN109740586A (en) * | 2018-12-19 | 2019-05-10 | 南京华科和鼎信息科技有限公司 | A kind of anti-dazzle certificate automatic reading system and method |
| CN109596955A (en) * | 2018-12-30 | 2019-04-09 | 国网北京市电力公司 | Shelf depreciation state determines method and device |
| CN109871746A (en) * | 2018-12-30 | 2019-06-11 | 国网辽宁省电力有限公司电力科学研究院 | A kind of mode identification method based on the diagnosis of GIS insulation defect |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
Non-Patent Citations (2)
| Title |
|---|
| SONG, HUI ET AL: "GIS Partial Discharge Pattern Recognition via Deep Convolutional Neural Network under Complex Data Sources", 《IEEE TRANSACTIONS ON DIELECTRICS AND ELECTRICAL INSULATION》 * |
| 吕启深: "GA-BP神经网络在GIS典型缺陷局部放电识别中的应用", 《电瓷避雷器》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191313A (en) * | 2021-05-20 | 2021-07-30 | 国能大渡河沙坪发电有限公司 | Video stream discharge identification method and device based on hydraulic power plant and computer equipment |
| CN115184744A (en) * | 2022-06-27 | 2022-10-14 | 上海格鲁布科技有限公司 | GIS ultrahigh frequency discharge signal detection device and method based on fast-RCNN |
| CN115184744B (en) * | 2022-06-27 | 2023-09-05 | 上海格鲁布科技有限公司 | GIS ultrahigh frequency discharge signal detection device and method based on fast-RCNN |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106803071B (en) | Method and device for detecting object in image | |
| CN109086785B (en) | Training method and device for image calibration model | |
| CN110175982B (en) | Defect detection method based on target detection | |
| CN110390666B (en) | Road damage detection method, device, computer equipment and storage medium | |
| CN110619330A (en) | Recognition model training method and device, computer equipment and recognition method | |
| CN111797890A (en) | Method and system for detecting defects of power transmission line equipment | |
| CN112560722B (en) | Airplane target identification method and device, computer equipment and storage medium | |
| CN110826476A (en) | Image detection method and device for identifying target object, electronic equipment and storage medium | |
| CN112861917B (en) | Weak supervision target detection method based on image attribute learning | |
| Zhou et al. | Exploring faster RCNN for fabric defect detection | |
| CN111598889B (en) | Identification method and device for inclination fault of equalizing ring and computer equipment | |
| CN106778650A (en) | Scene adaptive pedestrian detection method and system based on polymorphic type information fusion | |
| CN111368636A (en) | Object classification method and device, computer equipment and storage medium | |
| CN110598698A (en) | Natural scene text detection method and system based on adaptive regional suggestion network | |
| CN110334775B (en) | Unmanned aerial vehicle line fault identification method and device based on width learning | |
| CN114399484A (en) | Photovoltaic module defect detection method, device, equipment and medium | |
| CN113065379B (en) | Image detection method and device integrating image quality and electronic equipment | |
| CN111832561A (en) | Character sequence recognition method, device, equipment and medium based on computer vision | |
| CN111382638B (en) | Image detection method, device, equipment and storage medium | |
| CN115170882A (en) | Optimization method of rail wagon part detection network and guardrail breaking fault identification method | |
| CN110781887A (en) | License plate screw detection method and device and computer equipment | |
| CN114842243A (en) | Fault detection method, system and equipment based on large-span inspection robot | |
| CN110569717A (en) | partial discharge detection method, device, system, equipment and readable storage medium | |
| CN108985216B (en) | Pedestrian head detection method based on multivariate logistic regression feature fusion | |
| CN116229236A (en) | Bacillus tuberculosis detection method based on improved YOLO v5 model |
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: 20191213 |
|
| RJ01 | Rejection of invention patent application after publication |