CN110765858A - Non-invasive fault arc monitoring method based on convolutional neural network - Google Patents
Non-invasive fault arc monitoring method based on convolutional neural network Download PDFInfo
- Publication number
- CN110765858A CN110765858A CN201910867785.5A CN201910867785A CN110765858A CN 110765858 A CN110765858 A CN 110765858A CN 201910867785 A CN201910867785 A CN 201910867785A CN 110765858 A CN110765858 A CN 110765858A
- Authority
- CN
- China
- Prior art keywords
- neural network
- monitoring
- convolutional neural
- image
- abnormal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1218—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing using optical methods; using charged particle, e.g. electron, beams or X-rays
-
- 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]
Abstract
The invention relates to a non-invasive fault arc monitoring method based on a convolutional neural network, which comprises the steps of dividing a monitoring area, starting to collect line videos in the monitoring area after initialization, synchronizing a video stream by a control end when any one frame of video image is abnormal, confirming an abnormal area, inputting all processed video images into the convolutional neural network after confirming the abnormal area of N frames of video images, extracting and classifying the characteristics of the abnormal area, alarming when the characteristic of the fault arc is confirmed to be met based on a classification result, and simultaneously confirming the position where a power supply corresponding to the current fault arc can be cut off to cut off the current. The invention has low cost, only needs to install a conventional sampling camera, has simple installation and convenient debugging, does not need to utilize manual selection characteristics, adopts computer automatic processing, and has high processing speed and high monitoring efficiency.
Description
Technical Field
The invention belongs to data identification; a data representation; a record carrier; the technical field of processing of record carriers, in particular to a non-invasive fault arc monitoring method based on a convolutional neural network for reading or recognizing printed or written characters or for recognizing patterns.
Background
The electric fire monitoring system comprises electric fire monitoring equipment, a residual current type electric fire monitoring detector and a temperature measurement type electric fire monitoring detector, can monitor the current, residual current and temperature in a protected circuit, finds electric fire hidden dangers in time and prevents electric fire from happening.
In fact, however, many serious fire accidents are caused only by fault arcs in the line below the rated current or expected short-circuit current, which may occur in the case of improperly designed or aged power supply lines, appliance plugs and power lines of domestic appliances, internal wiring harnesses or component insulation; when a fault electric arc occurs, the protection devices on the circuit such as electric leakage, overcurrent and short circuit can not detect or can not act rapidly to cut off a power supply, the main hazard of the fault electric arc is fire, when the fault electric arc occurs, the central temperature of the fault electric arc reaches 3000-4000 ℃, metal melt splashes out, the high temperature and high heat generated by the fault electric arc easily ignites a circuit insulating layer to cause the circuit to fire, and if combustible substances exist near a fault point, the combustible substances are also easily ignited to cause the fire.
Therefore, from the perspective of fire monitoring, it is very important to monitor the fault arc, and it is also important to monitor the fire.
In the prior art, a mainstream fault arc judgment method mainly adopts a mode of installing fault arc monitoring equipment on a line, and is invasive detection, and although the monitoring effect is good, the problems of high cost and complex installation exist generally.
Disclosure of Invention
The invention solves the problems of high installation cost and complex installation of the fault arc monitoring device mainly adopted in the fault arc judgment method in the prior art, and provides an optimized non-invasive fault arc monitoring method based on a convolutional neural network.
The technical scheme adopted by the invention is that a convolution neural network-based non-invasive fault arc monitoring method comprises the following steps:
step 1: dividing a monitoring area and initializing; starting to collect line videos in a monitoring range;
step 2: when any frame of video image is abnormal, synchronizing video stream from the current frame in real time;
and step 3: confirming an abnormal area;
and 4, step 4: if the current frame is the Nth frame, the next step is carried out, otherwise, the next frame image is obtained, and the step 3 is carried out; n is more than 1;
and 5: inputting all images of the confirmed abnormal areas into a convolutional neural network, extracting characteristics of the abnormal areas and classifying the abnormal areas;
step 6: based on the classification result, if the fault arc characteristics are confirmed to be met, alarming is carried out, the next step is carried out, otherwise, the to-be-confirmed state is prompted, and the step 2 is returned;
and 7: and confirming the position where the power supply can be cut off corresponding to the current fault arc, and cutting off the current.
Preferably, in step 1, dividing the monitoring area includes the following steps:
step 1.1: confirming that any monitoring picture acquisition equipment monitors all lines in the space; if the monitoring blind area exists, increasing monitoring picture acquisition equipment or subdividing a monitoring area corresponding to any monitoring picture acquisition equipment in the same space;
step 1.2: carrying out initial image acquisition and background reservation on a monitoring area corresponding to any monitoring picture acquisition equipment to enable the whole monitoring area to be used as a background image;
step 1.3: and confirming the input end and the input end of each wire in the background image, and marking the position where the power supply can be cut off.
Preferably, the step 2 comprises the steps of:
step 2.1: taking any frame of video image as a current frame;
step 2.2: taking the difference between the video image of the previous frame of the current frame and the video image of the current frame to obtain a difference image C;
step 2.3: if C is not equal to 0, judging that the current frame is abnormal, and performing the next step, otherwise, taking the next frame of the current frame as the current frame, and returning to the step 2.2;
step 2.4: and (3) taking the current frame video image for binarization processing, if three channel values of RGB are all 255 areas on the original line, considering that a fault arc exists, synchronizing the video stream from the current frame in real time, otherwise, prompting to-be-confirmed, and returning to the step 2.
Preferably, the step 3 comprises the steps of:
step 3.1: processing the synchronous video stream to obtain a foreground binary image;
step 3.2: carrying out connected domain detection on the foreground binary image to obtain a minimum external rectangle;
step 3.3: and intercepting an image area corresponding to the minimum circumscribed rectangle in the video image as an abnormal area.
Preferably, said step 3.2 comprises the steps of:
step 3.2.1: carrying out size planning on the obtained foreground binary image;
step 3.2.2: detecting connected domains on the foreground binary image, and filtering the connected domains with the filtering areas smaller than a threshold value S;
step 3.2.3: obtaining a minimum circumscribed rectangle of the connected domain;
step 3.2.4: and carrying out size calibration on the minimum circumscribed rectangle.
Preferably, said step 3.3 comprises the steps of:
step 3.3.1: intercepting an image area corresponding to a minimum external rectangle in a video image;
step 3.3.2: adjusting the image area to the corresponding size of the feature extraction block;
step 3.3.3: and taking the image area after the size adjustment as an abnormal area of the feature to be extracted.
Preferably, the step 5 comprises the steps of:
step 5.1: extracting features of abnormal regions of all video images by using the trained convolutional neural network;
step 5.2: enhancing the extracted features;
step 5.3: and outputting a classification result by the trained convolutional neural network.
Preferably, in the step 5.1, the feature includes radian and/or inflection point features of a region in which the RGB channel values are 255 in the abnormal region.
Preferably, said step 5.2 comprises the steps of:
step 5.2.1: partitioning the region corresponding to the extracted features into blocks;
step 5.2.2: calculating the mean value of the gradient values of each block;
step 5.2.3: differentiating each pixel point in each block, increasing the gradient value of the pixel point with the gradient value above a threshold value, and reducing the gradient value of the pixel point with the gradient value below the threshold value;
step 5.2.4: connecting the adjusted pixel points in each block in series to obtain an adjusted block;
step 5.2.5: and connecting all the adjusted blocks in series to obtain the enhanced overall characteristics of the abnormal area.
Preferably, the video collected by the monitoring picture collecting device takes 7 days as a coverage period.
The invention provides an optimized non-invasive fault arc monitoring method based on a convolutional neural network, which comprises the steps of dividing a monitoring area, starting to collect line videos in the monitoring area after initialization, synchronizing a video stream by a control terminal when any one frame of video image is abnormal, confirming an abnormal area, inputting all processed video images into the convolutional neural network after the abnormal area of N frames of video images is confirmed, extracting and classifying the characteristics of the abnormal area, alarming when the characteristic of the fault arc is confirmed to be met based on a classification result, confirming the position where a power supply corresponding to the current fault arc can be cut off, and cutting off the current.
The invention has low cost, only needs to install a conventional sampling camera, has simple installation and convenient debugging, does not need to utilize manual selection characteristics, adopts computer automatic processing, and has high processing speed and high monitoring efficiency.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The present invention is described in further detail with reference to the following examples, but the scope of the present invention is not limited thereto.
The invention relates to a convolutional neural network-based non-invasive fault arc monitoring method, which comprises the following steps.
Step 1: dividing a monitoring area and initializing; and starting to collect the line video in the monitoring range.
In the step 1, dividing the monitoring area includes the following steps:
step 1.1: confirming that any monitoring picture acquisition equipment monitors all lines in the space; if the monitoring blind area exists, increasing monitoring picture acquisition equipment or subdividing a monitoring area corresponding to any monitoring picture acquisition equipment in the same space;
step 1.2: carrying out initial image acquisition and background reservation on a monitoring area corresponding to any monitoring picture acquisition equipment to enable the whole monitoring area to be used as a background image;
step 1.3: and confirming the input end and the input end of each wire in the background image, and marking the position where the power supply can be cut off.
The video collected by the monitoring picture collecting device takes 7 days as a coverage period.
In the invention, all lines in the monitoring range need to be ensured to be in the monitoring range, so that any opportunity for troubleshooting the fault arc is not missed; if the electric wire at any position is not in the monitoring range, the space to be monitored needs to be redistributed.
In the invention, the background diagram is confirmed in step 1, and the input end and the output end of each line of electric wire are confirmed and marked simultaneously, so that the lines between adjacent monitoring areas are matched, the input electrodes are sequentially guided, each line of electric wire corresponds to the input electrodes, the input electrodes are marked as the positions where the power supply can be cut off, the specific circuit can be traced back to the occurrence of the abnormity, such as the electric arc, and the further discharge of the electric arc position is placed after reasonable cutting off while the alarm is ensured.
In the invention, in order to ensure a sufficient monitoring space, a mode of covering an old monitoring video with a new monitoring video can be adopted, and certainly, in order to ensure more accurate tracing of fault arcs and the like, a certain time is required to be reserved for a covering time period, generally, 7 days are taken as a covering period, and certainly, a person skilled in the art can set the covering period according to the requirement.
Step 2: and when any frame of video image is abnormal, synchronizing the video stream from the current frame in real time.
The step 2 comprises the following steps:
step 2.1: taking any frame of video image as a current frame;
step 2.2: taking the difference between the video image of the previous frame of the current frame and the video image of the current frame to obtain a difference image C;
step 2.3: if C is not equal to 0, judging that the current frame is abnormal, and performing the next step, otherwise, taking the next frame of the current frame as the current frame, and returning to the step 2.2;
step 2.4: and (3) taking the current frame video image for binarization processing, if three channel values of RGB are all 255 areas on the original line, considering that a fault arc exists, synchronizing the video stream from the current frame in real time, otherwise, prompting to-be-confirmed, and returning to the step 2.
In the invention, generally, monitoring areas are all dark occasions, and the occurrence of electric arcs is obvious in video frames due to the characteristics of electric discharge, heat generation and the like, and can be easily found by adopting video images to make difference.
In the invention, because the arc is generally in a luminous form due to the discharge characteristic of the arc, after the video image of the current frame is subjected to binarization processing, the position where the arc appears is necessarily in white, that is, the three channel values of RGB are all 255.
In the invention, an exceptional situation exists, namely, other dynamic states of non-electric arcs appear in a monitoring image, or even electric wires move, in this situation, the result of binarization generally does not appear as a white area when electric arcs occur, so that the result is taken as a situation to be confirmed, and an operator can specifically check the situation when the operator thinks that the situation is needed.
And step 3: and confirming the abnormal area.
The step 3 comprises the following steps:
step 3.1: processing the synchronous video stream to obtain a foreground binary image;
step 3.2: carrying out connected domain detection on the foreground binary image to obtain a minimum external rectangle;
said step 3.2 comprises the steps of:
step 3.2.1: carrying out size planning on the obtained foreground binary image;
step 3.2.2: detecting connected domains on the foreground binary image, and filtering the connected domains with the filtering areas smaller than a threshold value S;
step 3.2.3: obtaining a minimum circumscribed rectangle of the connected domain;
step 3.2.4: and carrying out size calibration on the minimum circumscribed rectangle.
Step 3.3: and intercepting an image area corresponding to the minimum circumscribed rectangle in the video image as an abnormal area.
Said step 3.3 comprises the steps of:
step 3.3.1: intercepting an image area corresponding to a minimum external rectangle in a video image;
step 3.3.2: adjusting the image area to the corresponding size of the feature extraction block;
step 3.3.3: and taking the image area after the size adjustment as an abnormal area of the feature to be extracted.
In the invention, the step 3 is mainly used for confirming the abnormal area, namely, the image for the convolutional neural network processing is reduced, so that the identification time is greatly shortened.
In the invention, because the initial image acquisition is carried out on the monitoring area corresponding to the monitoring picture acquisition equipment and is reserved as the background image, a foreground binary image can be obtained, and the area to be detected is defined in a luminous area.
In the invention, the obtained foreground binary image is subjected to size division to obtain the image size of the foreground binary image, the position and the length and the width of a light-emitting point of the foreground binary image can be obtained based on debugging in an initialization process, and based on the detection, connected domain detection is carried out on the foreground binary image to obtain the minimum external rectangle of a main light-emitting region.
In the invention, the minimum external rectangle of the connected domain is subjected to size calibration, so that the approximate height and width of the electric arc can be obtained, and further the specification of the electric arc can be obtained.
In the invention, an image area corresponding to the minimum circumscribed rectangle in the video image is intercepted, and the image area is adjusted to the specified size of the feature extraction block, so that the subsequent processing of the convolutional neural network is facilitated, and the processed image is taken as an abnormal area of the feature to be extracted.
And 4, step 4: if the current frame is the Nth frame, the next step is carried out, otherwise, the next frame image is obtained, and the step 3 is carried out; n is more than 1.
In the invention, in order to ensure the continuity of arc detection, abnormal regions of a plurality of frame images need to be extracted and are integrally input into a convolutional neural network, so that the accuracy of a detection result is ensured.
In the present invention, N is generally 60 to 120, i.e., at least 5 to 10 seconds.
And 5: and inputting the images of all confirmed abnormal areas into a convolutional neural network, and extracting and classifying the characteristics of the abnormal areas.
The step 5 comprises the following steps:
step 5.1: extracting features of abnormal regions of all video images by using the trained convolutional neural network;
in the step 5.1, the features include radian and/or inflection point features of a region in which the RGB channel values are all 255 in the abnormal region.
Step 5.2: enhancing the extracted features;
the step 5.2 comprises the following steps:
step 5.2.1: partitioning the region corresponding to the extracted features into blocks;
step 5.2.2: calculating the mean value of the gradient values of each block;
step 5.2.3: differentiating each pixel point in each block, increasing the gradient value of the pixel point with the gradient value above a threshold value, and reducing the gradient value of the pixel point with the gradient value below the threshold value;
step 5.2.4: connecting the adjusted pixel points in each block in series to obtain an adjusted block;
step 5.2.5: and connecting all the adjusted blocks in series to obtain the enhanced overall characteristics of the abnormal area.
Step 5.3: and outputting a classification result by the trained convolutional neural network.
In the invention, firstly, the trained convolutional neural network is used for extracting the characteristics of the abnormal regions of all video images, wherein the convolutional neural network is a conventional technology in the field of neural networks, and the training can be completed by a person skilled in the art.
In the invention, the characteristics are mainly concentrated on the radian of the electric arc, the discharge capacity of partial electric arc is large, and the characteristics such as inflection points and the like can be possibly generated; the features may also include other kinds, such as radioactive features, waveform features, etc., which are easily understood by those skilled in the art, and should be processed as training samples during the training process of the convolutional neural network, so that the trained convolutional neural network can identify diversified arc features.
In the invention, because the arc has an obvious luminous state in the discharging process, the extracted features need to be enhanced, an enhancement mode of the directional gradient histogram features is introduced, the regions corresponding to the extracted features are subjected to partition differentiation processing, so that the difference of pixel points in each block can be rapidly increased, screening is carried out in a threshold value mode, the parts which are bright but not the brightest are weakened, the obvious discharging regions which exceed the threshold value are gained, the obvious discharging regions are brighter, the adjusted images of different blocks can be finally obtained, and then splicing between the blocks is carried out, so that the features are enhanced on one hand, and the processing speed is accelerated on the other hand.
In the invention, the classification result is finally output by the trained convolutional neural network.
Step 6: and (3) based on the classification result, if the fault arc characteristics are confirmed to be met, giving an alarm, carrying out the next step, otherwise, prompting to-be-confirmed, and returning to the step 2.
And 7: and confirming the position where the power supply can be cut off corresponding to the current fault arc, and cutting off the current.
In the invention, if the light-emitting phenomenon exists but the characteristics of the electric arc are not met, if open fire occurs, the state is a state to be confirmed, otherwise, direct alarm is required to cut off the line, and unnecessary loss is prevented to the maximum extent.
The method comprises the steps of dividing a monitoring area, initializing, starting to collect line videos in the monitoring area, synchronizing a video stream by a control end when any one frame of video image is abnormal, confirming an abnormal area, inputting all processed video images into a convolutional neural network after the abnormal area of N frames of video images is confirmed, extracting and classifying the characteristics of the abnormal area, alarming when the characteristic of a fault arc is confirmed to be met based on a classification result, confirming the position of a power supply corresponding to the current fault arc to be cut off at the same time, and cutting off the current.
The invention has low cost, only needs to install a conventional sampling camera, has simple installation and convenient debugging, does not need to utilize manual selection characteristics, adopts computer automatic processing, and has high processing speed and high monitoring efficiency.
Claims (10)
1. A non-invasive fault arc monitoring method based on a convolutional neural network is characterized in that: the method comprises the following steps:
step 1: dividing a monitoring area and initializing; starting to collect line videos in a monitoring range;
step 2: when any frame of video image is abnormal, synchronizing video stream from the current frame in real time;
and step 3: confirming an abnormal area;
and 4, step 4: if the current frame is the Nth frame, the next step is carried out, otherwise, the next frame image is obtained, and the step 3 is carried out; n is more than 1;
and 5: inputting all images of the confirmed abnormal areas into a convolutional neural network, extracting characteristics of the abnormal areas and classifying the abnormal areas;
step 6: based on the classification result, if the fault arc characteristics are confirmed to be met, alarming is carried out, the next step is carried out, otherwise, the to-be-confirmed state is prompted, and the step 2 is returned;
and 7: and confirming the position where the power supply can be cut off corresponding to the current fault arc, and cutting off the current.
2. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 1, wherein: in the step 1, dividing the monitoring area includes the following steps:
step 1.1: confirming that any monitoring picture acquisition equipment monitors all lines in the space; if the monitoring blind area exists, increasing monitoring picture acquisition equipment or subdividing a monitoring area corresponding to any monitoring picture acquisition equipment in the same space;
step 1.2: carrying out initial image acquisition and background reservation on a monitoring area corresponding to any monitoring picture acquisition equipment to enable the whole monitoring area to be used as a background image;
step 1.3: and confirming the input end and the input end of each wire in the background image, and marking the position where the power supply can be cut off.
3. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 1, wherein: the step 2 comprises the following steps:
step 2.1: taking any frame of video image as a current frame;
step 2.2: taking the difference between the video image of the previous frame of the current frame and the video image of the current frame to obtain a difference image C;
step 2.3: if C is not equal to 0, judging that the current frame is abnormal, and performing the next step, otherwise, taking the next frame of the current frame as the current frame, and returning to the step 2.2;
step 2.4: and (3) taking the current frame video image for binarization processing, if three channel values of RGB are all 255 areas on the original line, considering that a fault arc exists, synchronizing the video stream from the current frame in real time, otherwise, prompting to-be-confirmed, and returning to the step 2.
4. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 1, wherein: the step 3 comprises the following steps:
step 3.1: processing the synchronous video stream to obtain a foreground binary image;
step 3.2: carrying out connected domain detection on the foreground binary image to obtain a minimum external rectangle;
step 3.3: and intercepting an image area corresponding to the minimum circumscribed rectangle in the video image as an abnormal area.
5. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 4, wherein: said step 3.2 comprises the steps of:
step 3.2.1: carrying out size planning on the obtained foreground binary image;
step 3.2.2: detecting connected domains on the foreground binary image, and filtering the connected domains with the filtering areas smaller than a threshold value S;
step 3.2.3: obtaining a minimum circumscribed rectangle of the connected domain;
step 3.2.4: and carrying out size calibration on the minimum circumscribed rectangle.
6. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 4, wherein: said step 3.3 comprises the steps of:
step 3.3.1: intercepting an image area corresponding to a minimum external rectangle in a video image;
step 3.3.2: adjusting the image area to the corresponding size of the feature extraction block;
step 3.3.3: and taking the image area after the size adjustment as an abnormal area of the feature to be extracted.
7. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 1, wherein: the step 5 comprises the following steps:
step 5.1: extracting features of abnormal regions of all video images by using the trained convolutional neural network;
step 5.2: enhancing the extracted features;
step 5.3: and outputting a classification result by the trained convolutional neural network.
8. The convolutional neural network-based non-invasive arc fault monitoring method according to claim 7, wherein: in the step 5.1, the features include radian and/or inflection point features of a region in which the RGB channel values are all 255 in the abnormal region.
9. The convolutional neural network-based non-invasive arc fault monitoring method according to claim 7, wherein: the step 5.2 comprises the following steps:
step 5.2.1: partitioning the region corresponding to the extracted features into blocks;
step 5.2.2: calculating the mean value of the gradient values of each block;
step 5.2.3: differentiating each pixel point in each block, increasing the gradient value of the pixel point with the gradient value above a threshold value, and reducing the gradient value of the pixel point with the gradient value below the threshold value;
step 5.2.4: connecting the adjusted pixel points in each block in series to obtain an adjusted block;
step 5.2.5: and connecting all the adjusted blocks in series to obtain the enhanced overall characteristics of the abnormal area.
10. The convolutional neural network-based non-invasive fault arc monitoring method according to claim 1, wherein: the video collected by the monitoring picture collecting device takes 7 days as a coverage period.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910867785.5A CN110765858B (en) | 2019-09-14 | 2019-09-14 | Non-invasive fault arc monitoring method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910867785.5A CN110765858B (en) | 2019-09-14 | 2019-09-14 | Non-invasive fault arc monitoring method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110765858A true CN110765858A (en) | 2020-02-07 |
| CN110765858B CN110765858B (en) | 2022-07-05 |
Family
ID=69329636
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910867785.5A Active CN110765858B (en) | 2019-09-14 | 2019-09-14 | Non-invasive fault arc monitoring method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110765858B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111458599A (en) * | 2020-04-16 | 2020-07-28 | 福州大学 | Series arc fault detection method based on one-dimensional convolutional neural network |
| US20220215643A1 (en) * | 2021-01-07 | 2022-07-07 | Canon Kabushiki Kaisha | Image processing device, imaging device, image processing method, and recording medium |
| CN115085160A (en) * | 2022-08-22 | 2022-09-20 | 南京弘特浩电气有限公司 | Arc light protection system |
| CN116503858A (en) * | 2023-05-06 | 2023-07-28 | 华东师范大学 | Immunofluorescence image classification method and system based on generation model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102208018A (en) * | 2011-06-01 | 2011-10-05 | 西安工程大学 | Method for recognizing fire disaster of power transmission line based on video variance analysis |
| KR20130017611A (en) * | 2011-08-11 | 2013-02-20 | (주) 예스텔레콤 | Apparatus for detecting arc fault using arc pulse timing |
| CN105337180A (en) * | 2015-10-16 | 2016-02-17 | 常熟开关制造有限公司(原常熟开关厂） | Power distribution cabinet and power distribution system |
| CN106934418A (en) * | 2017-03-09 | 2017-07-07 | 国家电网公司 | A kind of insulator infrared diagnostics method based on convolution Recursive Networks |
| CN109613401A (en) * | 2019-01-31 | 2019-04-12 | 福州大学 | A kind of cable arc fault early warning system based on the infrared thermovision of intelligence low cost |
| CN110135269A (en) * | 2019-04-18 | 2019-08-16 | 杭州电子科技大学 | A kind of fire image detection method based on blend color model and neural network |
-
2019
- 2019-09-14 CN CN201910867785.5A patent/CN110765858B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102208018A (en) * | 2011-06-01 | 2011-10-05 | 西安工程大学 | Method for recognizing fire disaster of power transmission line based on video variance analysis |
| KR20130017611A (en) * | 2011-08-11 | 2013-02-20 | (주) 예스텔레콤 | Apparatus for detecting arc fault using arc pulse timing |
| CN105337180A (en) * | 2015-10-16 | 2016-02-17 | 常熟开关制造有限公司(原常熟开关厂） | Power distribution cabinet and power distribution system |
| CN106934418A (en) * | 2017-03-09 | 2017-07-07 | 国家电网公司 | A kind of insulator infrared diagnostics method based on convolution Recursive Networks |
| CN109613401A (en) * | 2019-01-31 | 2019-04-12 | 福州大学 | A kind of cable arc fault early warning system based on the infrared thermovision of intelligence low cost |
| CN110135269A (en) * | 2019-04-18 | 2019-08-16 | 杭州电子科技大学 | A kind of fire image detection method based on blend color model and neural network |
Non-Patent Citations (2)
| Title |
|---|
| SHIZE HUANG ET.AL: "Arc detection and recognition in pantograph–catenary system based on convolutional neural network", 《INFORMATION SCIENCES》 * |
| 余琼芳 等: "基于AlexNet深度学习网络的串联故障电弧检测方法", 《电子测量与仪器学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111458599A (en) * | 2020-04-16 | 2020-07-28 | 福州大学 | Series arc fault detection method based on one-dimensional convolutional neural network |
| US20220215643A1 (en) * | 2021-01-07 | 2022-07-07 | Canon Kabushiki Kaisha | Image processing device, imaging device, image processing method, and recording medium |
| US11854239B2 (en) * | 2021-01-07 | 2023-12-26 | Canon Kabushiki Kaisha | Image processing device, imaging device, image processing method, and recording medium |
| CN115085160A (en) * | 2022-08-22 | 2022-09-20 | 南京弘特浩电气有限公司 | Arc light protection system |
| CN115085160B (en) * | 2022-08-22 | 2022-11-01 | 南京弘特浩电气有限公司 | Arc light protection system |
| CN116503858A (en) * | 2023-05-06 | 2023-07-28 | 华东师范大学 | Immunofluorescence image classification method and system based on generation model |
| CN116503858B (en) * | 2023-05-06 | 2024-01-23 | 华东师范大学 | Immunofluorescence image classification method and system based on generation model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110765858B (en) | 2022-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110765858B (en) | Non-invasive fault arc monitoring method based on convolutional neural network | |
| CN108307146B (en) | System and method for detecting potential safety hazard of high-voltage transmission line | |
| US11754607B2 (en) | Detection of electric discharges that precede fires in electrical wiring | |
| JP4729610B2 (en) | Smoke detector | |
| CN107272637A (en) | A kind of video monitoring system fault self-checking self- recoverage control system and method | |
| JP4653207B2 (en) | Smoke detector | |
| CN104093017B (en) | A kind of method being realized three road signal monitorings by contrast | |
| CN106597222B (en) | A kind of transmission line of electricity overvoltage fault type recognition method based on ground line current | |
| CN107103330A (en) | A kind of LED status recognition methods and device | |
| CN104469305A (en) | Fault detecting method and device of power grid video monitoring device | |
| CN107169966B (en) | Power transmission line forest fire identification method based on temperature distribution | |
| CN112001327B (en) | Valve hall equipment fault identification method and system | |
| CN115017444A (en) | Positioning and identifying method for early faults of tree line | |
| CN111738142B (en) | Method and system for judging air switch state | |
| JP2018151376A (en) | Evaluation of phase-resolved partial discharge | |
| CN111899210A (en) | Arc detection method based on image recognition | |
| US20220057271A1 (en) | Apparatus for monitoring a switchgear | |
| CN107577997A (en) | The discrimination method that mountain fire is invaded in a kind of electric transmission line channel | |
| CN104021576A (en) | Method and system for tracking moving objects in scene | |
| CN113920057A (en) | Method, device and system for identifying color of product indicator light | |
| CN112067961A (en) | Arc fault detection method, system and storage medium | |
| CN104427210A (en) | A method and device for detecting a random abnormal blinking dot | |
| CN112634590B (en) | Method and device for detecting fault of platform area based on visual recognition technology | |
| CN114550046A (en) | Fire hazard identification method, device, processor, storage medium and electronic equipment | |
| CN105911433B (en) | The recognition methods of transmission line lightning stroke type |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |