CN114627388B - Foreign matter detection equipment and foreign matter detection method for power transmission line - Google Patents
Foreign matter detection equipment and foreign matter detection method for power transmission line Download PDFInfo
- Publication number
- CN114627388B CN114627388B CN202210288689.7A CN202210288689A CN114627388B CN 114627388 B CN114627388 B CN 114627388B CN 202210288689 A CN202210288689 A CN 202210288689A CN 114627388 B CN114627388 B CN 114627388B
- Authority
- CN
- China
- Prior art keywords
- image
- processor
- cloud server
- transmission line
- state image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 162
- 230000005540 biological transmission Effects 0.000 title claims abstract description 139
- 238000012545 processing Methods 0.000 claims abstract description 114
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000004891 communication Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 54
- 230000009545 invasion Effects 0.000 claims description 20
- 238000003709 image segmentation Methods 0.000 claims description 7
- 238000005265 energy consumption Methods 0.000 abstract description 8
- 238000012552 review Methods 0.000 description 22
- 238000012549 training Methods 0.000 description 20
- 230000008569 process Effects 0.000 description 17
- 238000012544 monitoring process Methods 0.000 description 8
- 238000007781 pre-processing Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000012423 maintenance Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000010191 image analysis Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000002372 labelling Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000000428 dust Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001627 detrimental effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012285 ultrasound imaging Methods 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The utility model relates to a transmission line foreign matter check out test set and foreign matter check out test set method thereof, transmission line foreign matter check out test set includes image acquisition device, treater and high in the clouds server, the treater is connected to image acquisition device, treater and high in the clouds server communication, image acquisition device and treater all set up in transmission line, image acquisition device is used for gathering transmission line's state image and sends to the treater, the treater is used for carrying out target detection to state image, after detecting the state image that has the foreign matter, send processing signal to high in the clouds server, high in the clouds server is used for judging whether release the warning according to processing signal. The image acquisition device and the processor are used for identifying whether foreign matters exist on the primary power transmission line or not, so that the number of images sent to the cloud server can be reduced, the dependence on a high-reliability network is reduced, the network cost is saved, the energy consumption is reduced, and the use is reliable.
Description
Technical Field
The application relates to the technical field of power grid maintenance, in particular to a power transmission line foreign matter detection device and a foreign matter detection method thereof.
Background
With the development of power industry in China and the implementation of strategies such as east and west electric power transmission, the safe operation of a power transmission line is one of important links for guaranteeing the stability of a power grid. However, the invasion of foreign matters on the wires is a phenomenon commonly existing in the power transmission line, and the easily floatable objects such as kites, reflective films, dust screens and the like become a great hidden trouble threatening the stable operation of the high-voltage power transmission line. Therefore, the invasion of foreign matters of the transmission line is required to be monitored, effective early warning is realized when the wire is hung with the foreign matters, and measures are taken to remove the wire hanging foreign matters.
The traditional transmission line foreign matter intrusion monitoring method is based on camera monitoring, and the state of the transmission line is shot through a camera installed on a transmission tower or other positions, and the picture is transmitted to a server through a mobile network to carry out foreign matter detection. However, the method needs to continuously upload the pictures to the cloud server, has high requirement on network state, is easy to generate unstable transmission, has high network communication cost and high use cost, is not suitable for long-term use, and is unreliable in work.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a power transmission line foreign matter detection apparatus and a foreign matter detection method thereof that are low in power consumption and reliable in use.
In a first aspect, the application provides a foreign matter detection device for a power transmission line, including an image acquisition device, a processor and a cloud server, wherein the image acquisition device is connected with the processor, the processor communicates with the cloud server, and the image acquisition device and the processor are both arranged on the power transmission line;
the image acquisition device is used for acquiring a state image of the power transmission line and sending the state image to the processor, the processor is used for carrying out target detection on the state image, after detecting the state image with foreign matters, the processor sends a processing signal to the cloud server, and the cloud server is used for judging whether to issue an alarm according to the processing signal.
In one embodiment, the processor includes a microcontroller unit and an embedded neural network processing unit, the image acquisition device and the embedded neural network processing unit are both connected to the microcontroller unit, and the microcontroller unit communicates with the cloud server;
the microcontroller unit is used for receiving the state image, transmitting the state image to the embedded neural network processing unit, performing target detection on the state image by the embedded neural network processing unit, and transmitting a processing signal to the cloud server through the microcontroller unit after detecting the state image with foreign matters.
In one embodiment, the image acquisition device comprises an RGB camera and an infrared camera, and the RGB camera and the infrared camera are both connected with the processor.
In one embodiment, the processor is configured to divide the status image from the RGB camera, and then perform object detection on the divided sub-images respectively.
In one embodiment, the processor is configured to divide the status image from the RGB camera using a halving or octaving method.
In one embodiment, the processor is configured to perform target detection on the status image based on a first weight, and the cloud server is configured to perform target review on the received status image with the foreign object based on a second weight, where a parameter of the first weight is less than a parameter of the second weight.
In a second aspect, the application further provides a foreign matter detection method of the foreign matter detection device of the power transmission line, the foreign matter detection device of the power transmission line comprises an image acquisition device, a processor and a cloud server, the image acquisition device is connected with the processor, the processor is communicated with the cloud server, and the image acquisition device and the processor are both arranged on the power transmission line;
The foreign matter detection method of the transmission line foreign matter detection device is applied to the processor, and comprises the following steps:
acquiring a state image of the power transmission line;
performing target detection on the state image;
after detecting a state image with foreign matters, sending a processing signal to the cloud server, and judging whether to issue an alarm or not by the cloud server according to the processing signal.
In one embodiment, the performing object detection on the status image includes:
and after the state image is segmented, respectively carrying out target detection on the segmented sub-images.
In a third aspect, the present application further provides a foreign object detection method of a foreign object detection device of a power transmission line, where the foreign object detection device of the power transmission line includes an image acquisition device, a processor and a cloud server, the image acquisition device is connected with the processor, the processor communicates with the cloud server, and the image acquisition device and the processor are both disposed on the power transmission line;
the foreign matter detection method of the transmission line foreign matter detection equipment is applied to the cloud server, and the method comprises the following steps:
receiving a processing signal; the processing signal is subjected to target detection on a state image by the processor, the state image is sent after the state image with foreign matters is detected, and the state image is acquired by the image acquisition device through the power transmission line;
And judging whether to issue an alarm according to the processing signal.
In one embodiment, the image acquisition device includes an RGB camera and an infrared camera, and the determining whether to issue an alarm according to the processing signal includes:
and when the processing signal comprises a first detection result obtained after the processor performs target detection on the state image from the RGB camera and a second detection result obtained after the processor performs target detection on the state image from the infrared camera, an alarm is issued.
The power transmission line foreign matter detection device comprises an image acquisition device, a processor and a cloud server, wherein the image acquisition device is connected with the processor, the processor is communicated with the cloud server, the image acquisition device and the processor are both arranged on the power transmission line, the image acquisition device is used for acquiring a state image of the power transmission line and sending the state image to the processor, the processor is used for carrying out target detection on the state image, after the state image with foreign matters is detected, a processing signal is sent to the cloud server, and the cloud server is used for judging whether to issue an alarm according to the processing signal. The image acquisition device and the processor are used for identifying whether foreign matters exist on the primary power transmission line or not, if so, the processing signals are sent to the cloud server, and the cloud server judges whether to issue an alarm or not according to the processing signals, so that the number of images sent to the cloud server can be reduced, the dependence on a high-reliability network is reduced, the network cost is saved, the energy consumption is reduced, and the use is reliable.
Drawings
FIG. 1 is a block diagram of a transmission line foreign matter detection apparatus in one embodiment;
FIG. 2 is a schematic diagram of halving an image in one embodiment;
FIG. 3 is a schematic diagram of an octant of an image in one embodiment;
FIG. 4 is a flowchart of the operation of the transmission line foreign matter detection apparatus in one embodiment;
FIG. 5 is a flow diagram of weight training in one embodiment;
FIG. 6 is a flow chart of a foreign object detection method of a transmission line foreign object detection apparatus in one embodiment;
fig. 7 is a flowchart of a foreign matter detection method of a transmission line foreign matter detection apparatus in another embodiment;
fig. 8 is a flowchart of a foreign matter detection method of the transmission line foreign matter detection apparatus in still another embodiment;
fig. 9 is a flowchart of a foreign matter detection method of a transmission line foreign matter detection apparatus in still another embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
In one embodiment, a transmission line foreign matter detection device is provided for detecting a transmission line and determining whether a foreign matter invades the transmission line. The foreign matter is different from the power transmission line, and may affect the normal operation of the power transmission line. Such as kites, kongming lights, reflective films, dust screens, and the like. Referring to fig. 1, the power transmission line foreign matter detection device includes an image acquisition device 300, a processor 100 and a cloud server 200, the image acquisition device 300 is connected with the processor 100, the processor 100 communicates with the cloud server 200, both the image acquisition device 300 and the processor 100 are disposed on a power transmission line, the image acquisition device 300 is used for acquiring a status image of the power transmission line and transmitting the status image to the processor 100, the processor 100 is used for performing target detection on the status image, after detecting that the status image has foreign matters, a processing signal is transmitted to the cloud server 200, and the cloud server 200 is used for judging whether to issue an alarm according to the processing signal. The image acquisition device 300 and the processor 100 are used for identifying whether foreign matters exist on the primary power transmission line or not, if so, a processing signal is sent to the cloud server 200, and the cloud server 200 judges whether to issue an alarm or not according to the processing signal, so that the number of images sent to the cloud server 200 can be reduced, the dependence on a high-reliability network is reduced, the network cost is saved, the energy consumption is reduced, and the use is reliable.
Specifically, the image capturing device 300 is disposed on or near a power transmission line, for example, a power transmission tower. The image acquisition device 300 is used for acquiring a state image of the power transmission line, and can shoot the power transmission line in real time or according to a preset time interval to obtain the state image of the power transmission line, the real-time acquisition of the state image can effectively avoid the occurrence of the condition of missing detection, and the acquisition of the state image according to the preset time interval can reduce the workload of the image acquisition device 300. The type of the image capturing device 300 is not unique, for example, the image capturing device may be an optical camera or an infrared camera 320, and the optical camera may capture a visible light image of the power transmission line as a status image, which is simple to implement and visual to image. The infrared camera 320 can collect an infrared image of the power transmission line as a status image, and because the temperature of the power transmission line is generally inconsistent with the temperature of the foreign object, the infrared image of the power transmission line can also be used as a basis for detecting whether the foreign object exists later. It will be appreciated that in other embodiments, the image acquisition device 300 may also be of other types, such as an ultrasound imaging device, etc., as long as those skilled in the art recognize that it can be implemented.
The processor 100 is also disposed on or near the power transmission line, for example, on a power transmission tower, and is configured to perform target detection on the status image, and send the detected status image with the foreign object to the cloud server 200. After the processor 100 acquires the status image, it determines whether or not foreign objects exist on the power transmission line by performing image analysis on the status image. The image analysis process is not unique, and may include gray scale processing and contour analysis, and by analyzing the gray scale and the object contour of each target object of the state image, it is determined whether or not foreign objects exist on the power transmission line. When the processor 100 analyzes that the foreign matter exists in the state image, it is considered that the foreign matter invasion of the transmission line is detected for the first time, and a processing signal is sent to the cloud server 200 for subsequent processing. The type of the processor 100 is not unique, and for example, the processor may be a micro control unit, and the micro control unit may implement data processing, and is small in size and high in integration degree. It is understood that in other embodiments, the micro-control unit may be of other configurations or types as long as one skilled in the art deems it to be practical.
The cloud server 200 is also called a cloud server, and a more advanced algorithm can be stored in the cloud server, so that the accuracy of analysis and detection results can be improved when the images are analyzed. In this embodiment, after receiving the processing signal, the cloud server 200 determines whether to issue an alarm according to the processing signal. The process of alarm judgment by the cloud server 200 is different according to the type of the processing signal. For example, when the image capturing apparatus 300 includes an RGB camera, the processor 100 analyzes a status image from the RGB camera, and after detecting that a status image of a foreign object exists, sends the status image of the foreign object to the cloud server 200, and the processing signal at this time is the status image of the foreign object. The cloud server 200 performs target review on the state image with the foreign object to determine whether the foreign object is actually present in the state image with the foreign object. If foreign matter exists, an alarm is issued. Specifically, the mode of performing the target review is not unique, for example, the target review may be performed by using a common image processing process, including image gray level processing, contour extraction, analysis, and the like, or may be implemented by using an artificial neural network, for example, a convolutional neural network, which may be specifically selected according to actual needs, so long as those skilled in the art consider that the target review may be implemented. Generally, an algorithm used by the cloud server 200, such as an artificial neural network, is complex, the target review precision is high, the types of objects which can be identified are more, and the accuracy of the foreign matter detection result can be further ensured.
Or, when the image capturing device 300 includes the RGB camera 310 and the infrared camera 320, after the processor 100 analyzes the status image from the RGB camera and analyzes the status image from the infrared camera 320, and detects that the status image has foreign objects, the processor 100 sends an alarm signal to the cloud server 200, the processing signal at this time is the alarm signal, and the cloud server may not analyze the image after receiving the alarm signal, and directly issue an alarm, so as to save a control flow. It can be understood that when the image capturing device 300 includes an RGB camera and an infrared camera 320, when the processor 100 analyzes a status image from the RGB camera and a status image from the infrared camera 320, and detects that only a status image from one camera has a foreign object, the status image with the foreign object is sent to the cloud server 200 for target review, and at this time, the processing signal is the status image with the foreign object, the cloud server 200 further determines whether the transmission line does have a foreign object invasion condition, and if it does, issues an alarm. Thus, the influence of the working error of the image acquisition device 300 on the detection result can be reduced, so that the working reliability of the foreign matter detection equipment of the power transmission line can be improved.
Further, the way in which the cloud server 200 issues the alarm is not unique, for example, to send a short message or make a call to an operation and maintenance person, and inform the operation and maintenance person that the transmission line has foreign matter invasion, and needs to be processed in time. Alternatively, the cloud server 200 may also send an alarm signal to a connected alarm device, which may be an audible and visual alarm or a display screen, etc., to inform the foreign matter invasion in time.
In one embodiment, referring to fig. 1, the processor 100 includes a micro-controller unit 110 and an embedded neural network processing unit 120, the image acquisition device 300 and the embedded neural network processing unit 120 are both connected to the micro-controller unit 110, the micro-controller unit 110 is in communication with the cloud server 200, the micro-controller unit 110 is configured to receive a status image and transmit the status image to the embedded neural network processing unit 120, the embedded neural network processing unit 120 performs object detection on the status image, and after detecting that a status image with a foreign object exists, a processing signal is sent to the cloud server 200 through the micro-controller unit 110.
Specifically, in this embodiment, the micro controller unit 110 may act as a repeater, and is responsible for transmitting the received status image to the embedded neural network processing unit 120, and transmitting the image processed by the embedded neural network processing unit 120 to the cloud server 200. The embedded neural network processing unit 120 is mainly used as the image processor 100, and different devices can realize different works, so that the overlarge workload of a single device can be avoided, and the service life of the device is influenced. The embedded neural network processing unit 120 adopts a decision mode similar to the human brain, has a deep learning function, and has small computing capacity and small volume. When the embedded neural network processing unit 120 performs object detection on the state image, a convolutional neural network algorithm may be used, and an appropriate object detection algorithm needs to be selected according to the specificity of the hardware of the embedded neural network processing unit 120. Further, the embedded neural network processing unit 120 may be embedded in the micro control unit, so as to improve the integration degree of the processor 100 and reduce the volume of the processor 100.
In one embodiment, the image capture device 300 includes an RGB camera 310 and an infrared camera 320, each of the RGB camera 310 and the infrared camera 320 coupled to the processor 100.
Specifically, the image capturing device 300 may include an RGB camera 310 and an infrared camera 320, where the RGB camera 310 captures three basic color components of red, green and blue through three different cables, and typically three separate CCD sensors are used to capture three color signals, which may be used to capture an accurate color image. The infrared camera 320 can collect an infrared image of the power transmission line as a status image, and because the temperature of the power transmission line is generally inconsistent with the temperature of the foreign object, the infrared image of the power transmission line can also be used as a basis for detecting whether the foreign object exists later. Further, the number of the RGB cameras 310 may be more than two, and the number of the infrared cameras 320 may be more than two, and each RGB camera 310 and each infrared camera 320 may be disposed at different positions, so as to realize multi-position and multi-angle status image acquisition of the power transmission line, thereby being beneficial to improving accuracy of foreign object detection results. In addition, the RGB camera 310 and the infrared camera 320 have the same shooting angle, the RGB camera 310, the infrared camera 320 and the detection result can be mutually verified, the detection error is reduced, and the working performance of the transmission line foreign matter detection device is improved.
In one embodiment, the processor 100 is configured to segment the status image from the RGB camera 310, and then perform object detection on the segmented sub-images. The processor 100 may further send the detected state image in which the sub-image with the foreign object exists to the cloud server 200, where the cloud server 200 segments the received state image with the foreign object, and then performs target review on the segmented sub-image, and if the foreign object is identified, issues an alarm.
Specifically, in the on-line monitoring scenario of the power transmission line, the scenario is often open, and the object to be monitored occupies a relatively small area in the picture, so when the processor 100 performs target detection on the state image from the RGB camera 310, the state image is first divided to obtain more than two sub-images, and then the divided sub-images are detected, so that the accuracy of the detection result can be improved. If the foreign matter exists in the sub-image, the foreign matter is considered to be detected in the state image where the sub-image is located, and the state image where the sub-image is located is used as a processing signal and sent to the cloud server 200 for further processing. It will be appreciated that when the processor 100 performs object detection on the status image from the infrared camera 320, the status image from the infrared camera 320 is not usually high in resolution, so that the object detection can be directly performed without performing an image segmentation step. Similarly, after receiving the processing signal, the cloud server 200 performs target review on the state image with the foreign object, divides the state image with the foreign object to obtain more than two sub-images, then detects the divided sub-images, and if the foreign object is detected in the sub-images, considers that the foreign object is actually present in the state image with the foreign object in which the sub-image is located, and issues an alarm. By adopting the mode of firstly dividing the image and then analyzing and processing the sub-images, the small target in the image can be detected more easily, which is beneficial to reducing the omission ratio.
In one embodiment, the processor 100 is configured to divide the status image from the RGB camera 310 using a halving or octaving method.
Specifically, when the processor 100 segments the status image from the RGB camera 310, the status image is generally segmented into more than two sub-images with equal areas, so that the same or similar analysis method can be used for each sub-image, which is beneficial to reducing the image processing error. Further, in this embodiment, the processor 100 may divide the state image by using a bisection or octaion method, and the sub-image after cutting is close to 1:1, the proportion of the objects in the state image can be effectively prevented from being changed, and the object detection is facilitated. The cloud server 200 may divide the status image with the foreign object by using a halving or octaving method, and the sub-image after the cutting is close to 1:1, can effectively prevent the proportion of the object in the state image with foreign matters from changing, and is more beneficial to target detection. Taking the image capturing apparatus 300 as the RGB camera 310 as an example, the resolution of the image that can be captured by the RGB camera 310 is 2048×1080, the image can be divided into 1024×1080 halves (see fig. 2) or 512×540 eight halves (see fig. 3), and the target detection is performed on the divided images, so as to record the target detection result.
In one embodiment, the processor 100 is configured to perform target detection on the status image based on a first weight, and the cloud server 200 is configured to perform target review on the received status image with foreign objects based on a second weight, where the parameter of the first weight is less than the parameter of the second weight.
Specifically, before the processor 100 performs the target detection on the state image, the cloud server 200 may perform the pre-training of the artificial neural network model weight before performing the target review on the received state image with the foreign object, please refer to fig. 5, and the training process includes: firstly, constructing an artificial neural network structure and initializing weights, then, performing forward training, calculating a loss function, taking the weights at the moment as the weights when the precision reaches the expectation, performing back propagation when the precision does not reach the expectation, updating the weights, and then returning to the forward training step. After the image capturing device 300 obtains the status image, the object appearing in the image may be labeled with an invader, for example, to identify a forest fire or a kite, i label the forest fire or the kite appearing in the image, and if no related object appears in the image, the labeling process is generally artificial. After the invader labeling, image preprocessing is carried out, wherein the image preprocessing is an important step in the artificial neural network training process, the image is subjected to operations such as zooming and combining, and the image preprocessing method can influence the training precision and is an important part of model training. After image preprocessing, the process returns to the forward training step as described above.
The process of training weights by the artificial neural network model used by the processor 100 and the cloud server 200 is substantially the same, and the difference is that the artificial neural network used by the cloud server 200 is generally complex, the target detection accuracy is high, and the types of objects that can be identified are more. The artificial neural network used in the processor 100 has a simpler network structure, and after the weights are obtained through training, quantization and format conversion of the weights are needed to realize low-energy-consumption real-time monitoring. Weight training is generally performed on a computer with better performance, and weights obtained by training are released to the cloud server 200 and the processor 100 for use.
In this embodiment, the processor 100 is configured to perform target detection on the status image based on a first weight, and the cloud server 200 is configured to perform target review on the received status image with foreign objects based on a second weight, where the parameters of the first weight are less than those of the second weight. The first weight and the second weight are different because the network depth used by the processor 100 is shallower than that used on the cloud server 200, the parameters of the first weight are smaller than those of the second weight, less memory is required by the processor 100 for performing object detection on the state image based on the first weight, and the energy consumption required for one time of object detection is lower.
When the image capturing apparatus 300 includes the RGB camera 310 and the infrared camera 320, the processor 100 may use different weights for the RGB image and the infrared image, for example, the weight used for the RGB image is a weight a, and the weight used for the infrared image is a weight a. Correspondingly, the weight used by the cloud server for the RGB image is weight B, and the weight used for the infrared image is weight B.
Above-mentioned transmission line foreign matter detection equipment, transmission line foreign matter detection equipment includes image acquisition device 300, processor 100 and high in the clouds server 200, image acquisition device 300 connects processor 100, processor 100 communicates with high in the clouds server 200, image acquisition device 300 and processor 100 all set up in transmission line, image acquisition device 300 is used for gathering transmission line's state image and sends to processor 100, processor 100 is used for carrying out target detection to state image, after detecting the state image that has the foreign matter, send processing signal to high in the clouds server 200, high in the clouds server 200 is used for judging whether issuing the warning according to processing signal. The image acquisition device 300 and the processor 100 are used for identifying whether foreign matters exist on the primary power transmission line or not, if so, a processing signal is sent to the cloud server 200, and the cloud server 200 judges whether to issue an alarm or not according to the processing signal, so that the number of images sent to the cloud server 200 can be reduced, the dependence on a high-reliability network is reduced, the network cost is saved, the energy consumption is reduced, and the use is reliable.
For a better understanding of the above embodiments, a detailed explanation is provided below in connection with a specific embodiment. In one embodiment, the transmission line foreign matter detection apparatus includes an image acquisition device 300, a processor 100 and a cloud server 200, the processor 100 includes a microcontroller unit 110 and an embedded neural network processing unit 120, and the image acquisition device 300 includes an RGB camera 310 and an infrared camera 320. Referring to fig. 4, the RGB camera 310 and the infrared camera 320 are both connected to the micro-controller unit 110 to realize multi-angle shooting of the current status of the transmission line, wherein the RGB camera 310 and the infrared camera 320 have the same shooting angle. The embedded neural network processing unit 120 processes the photographs to obtain preliminary prediction results, and solves the problem of small target detection by means of image segmentation. If early warning objects appear in the predicted results of the pictures shot by the RGB camera 310 and the pictures shot by the infrared camera 320, the early warning objects directly issue warning information to the cloud server 200; if an early warning object appears in the prediction results of one of the two, the images acquired by the two are uploaded to the cloud server 200, and finally whether the transmission line is invaded by foreign matters or not is judged and an alarm is issued. The neural network of the cloud server 200 itself is more advanced. The reason for the primary identification at the side equipment is to reduce the dependence on a high-reliability network and save network charge.
The pre-training of the artificial neural network model weights is required before the embedded neural network processing unit 120 and the cloud server 200 perform the target detection task, and the overall flow is shown in fig. 5. The image annotation refers to the annotation of objects appearing in an image, such as a mountain fire and a kite to be identified, the mountain fire and the kite appearing in the image are annotated, and if related objects do not appear in the image, the objects are not annotated. All labeling processes can be human. Image preprocessing is a step necessary for the training process of the artificial neural network, and the operations of scaling, combining and the like are performed on the images. The preprocessing method of the image can influence the training precision and is an important part of model training. The loss function and accuracy are expected to have no fixed value depending on the actual situation. Weight training is performed on a computer with better performance, and weights obtained by training are issued to the embedded processing unit and the cloud server 200. The output result is the coordinates and the type of the foreign matter identified in the image, and if the foreign matter is not in the image, no output is made.
The process of embedding the artificial neural network model weights used by the neural network processing unit 120 and the cloud server 200 is substantially the same, and the difference is that the artificial neural network used by the cloud server 200 is complex, the target detection accuracy is high, and the types of objects that can be identified are more. The artificial neural network used in the embedded neural network processing unit 120 has a simpler network structure, and after the weights are obtained by training, quantization and format conversion of the weights are needed to realize real-time monitoring with low energy consumption. The weight used by the embedded neural network processing unit 120 is a weight a, i.e., a first weight, and the weight used by the cloud server 200 is a weight B, i.e., a second weight. The artificial neural network used in the embedded neural network processing unit 120 and the artificial neural network used in the cloud server 200 may be convolutional neural networks or others. Further, when the image capturing device 300 includes the RGB camera 310 and the infrared camera 320, the weight used by the embedded neural network processing unit 120 for the RGB image is a weight a, and the weight used for the infrared image is a weight a; the cloud server 200 uses a weight B for the RGB image, and a weight B for the infrared image.
Under the on-line monitoring scene of the power transmission line, the scene is usually quite clear, the object to be monitored occupies smaller space in the picture, and the mode of dividing the image and then detecting is adopted for small target detection under the scene of the power transmission line. The resolution of an image which can be acquired by an RGB camera used for the transmission line integrated sensor is 2048 x 1080, the image is divided into 512 x 540 eight equal parts or 1024 x 1080 two equal parts, target detection is respectively carried out on the divided image, a target detection result is recorded, and an image division schematic diagram is shown. The proportion of pictures input into the model is defined, typically 1:1, a part of the above-mentioned image preprocessing is to scale the non-scaled image. Although the scaled image may also be input into the model, the scale of the object has changed, which is detrimental to object recognition. E.g. 2048 x 1080 pictures, scaled to 1:1, 1080 x 1080, the whole picture is narrowed, and the objects in the picture are also narrowed, so that the recognition effect is affected. A division of 2, 8 equal divisions is more suitable, and the cut small figures are close to 1: 1.
Target detection is carried out on the sub-images after the segmentation, the resolution of the sub-images after the segmentation is reduced, and the proportion of the images is close to 1:1, the effect of scaling the picture on the target is small during preprocessing, and the small target in the image is easy to detect. The eight-equal-part and two-equal-part segmentation modes are adopted, so that on one hand, the influence of image segmentation on the detection of a large target object can be avoided, and on the other hand, the small target object can be prevented from being segmented just in the middle of a segmentation line. Since the resolution of the picture taken by the infrared camera 320 is generally low, the above-described process may not be performed.
When the side performs foreign matter intrusion monitoring, if the picture shot by the RGB camera 310 and the picture shot by the infrared camera 320 both detect the early warning object, early warning information is directly issued to the cloud server 200, if an early warning object appears in the prediction result of one of the images, the power transmission line state picture used for the detection is uploaded to the cloud server 200 for further detection, whether the power transmission line is invaded by the foreign matter or not is finally judged, and an alarm is issued, if the cloud server 200 also judges that the early warning object exists in the picture, the alarm is issued, and the specific flow is shown in fig. 4. The edge is on the edge side, i.e. the transmission line side, and the opposite is on the cloud side, i.e. the cloud server 200 side.
The transmission line foreign matter detection device realizes transmission line foreign matter intrusion monitoring through the micro control unit, the embedded neural network processing unit 120, the plurality of RGB cameras 310, the infrared camera 320 and the cloud server 200, wherein the RGB cameras 310 and the infrared camera 320 have the same shooting angle. The plurality of RGB cameras 310 are connected with the micro-controller unit 110 to realize multi-angle shooting of the current situation of the transmission line, and the embedded neural network processing unit 120 processes the photos to obtain preliminary prediction results and solves the problem of small target detection in an image segmentation mode. If early warning objects appear in the predicted results of the pictures shot by the RGB camera 310 and the pictures shot by the infrared camera 320, the early warning objects directly issue warning information to the cloud server; if an early warning object appears in the prediction results of one of the two, uploading the images acquired by the two to a cloud server, and finally judging whether the transmission line is invaded by foreign matters or not and issuing an alarm.
By means of embedding the neural network processing unit 120, the plurality of RGB cameras 310 and the infrared camera 320, real-time and low-energy-consumption edge target detection is achieved, the problem of small target detection under a large scene is solved by dividing a high-resolution RGB image and carrying out target recognition on the divided subgraph, and the problems of high network state requirements and high network communication expense are solved by only uploading pictures of early warning objects to the cloud server 200. The cloud server 200 is used for carrying out target detection on the state picture of the transmission line, which is possibly invaded by foreign matters, transmitted by the side equipment, so that misjudgment and missed judgment on the foreign matters of the transmission line, which are caused by insufficient calculation power and simple network model, of the side equipment are solved, and the accuracy of the foreign matter invasion detection task of the transmission line is further improved. By means of simultaneous detection by the RGB camera 310 and the infrared camera 320, accuracy of target detection is improved.
Based on the same inventive concept, the embodiment of the application also provides a foreign matter detection method of a foreign matter detection device of a power transmission line, wherein the foreign matter detection device of the power transmission line comprises an image acquisition device 300, a processor 100 and a cloud server 200, the image acquisition device 300 is connected with the processor 100, the processor 100 is communicated with the cloud server 200, and the image acquisition device 300 and the processor 100 are both arranged on the power transmission line; referring to fig. 6, the foreign matter detection method of the foreign matter detection device for a power transmission line is applied to the processor 100, and the method includes:
Step S110: and acquiring a state image of the power transmission line.
Specifically, the processor 100 is connected to the image acquisition device 300, and can acquire a status image of the power transmission line through the image acquisition device 300. The type of the image capturing device 300 is not unique, for example, the image capturing device may be an optical camera or an infrared camera 320, and the optical camera may capture a visible light image of the power transmission line as a status image, which is simple to implement and visual to image. The infrared camera 320 can collect an infrared image of the power transmission line as a status image, and because the temperature of the power transmission line is generally inconsistent with the temperature of the foreign object, the infrared image of the power transmission line can also be used as a basis for detecting whether the foreign object exists later.
Step S120: and performing target detection on the state image.
After the processor 100 acquires the status image, it determines whether or not foreign objects exist on the power transmission line by performing image analysis on the status image. The image analysis process is not unique, and may include gray scale processing and contour analysis, and by analyzing the gray scale and the object contour of each target object of the state image, it is determined whether or not foreign objects exist on the power transmission line.
Step S130: after detecting that the state image with the foreign matters exists, sending a processing signal to the cloud server, and enabling the cloud server to judge whether to issue an alarm according to the processing signal.
When the processor 100 analyzes that the foreign matter exists in the state image, it is considered that the foreign matter invasion of the transmission line is detected for the first time, and a processing signal is sent to the cloud server 200, so that the cloud server 200 determines whether to issue an alarm according to the processing signal. When the image capturing device 300 includes an RGB camera, the processor 100 analyzes a status image from the RGB camera, and after detecting that a status image with a foreign object exists, sends the status image with the foreign object to the cloud server 200, and the processing signal at this time is the status image with the foreign object.
Or, when the image capturing device 300 includes the RGB camera and the infrared camera 320, after the processor 100 analyzes the status image from the RGB camera and analyzes the status image from the infrared camera 320, and detects that the status image has foreign objects, the processor 100 sends an alarm signal to the cloud server 200, the processing signal at this time is the alarm signal, and the cloud server may not analyze the image after receiving the alarm signal, and directly alarm, so as to save the control flow. It can be understood that when the image capturing device 300 includes an RGB camera and an infrared camera 320, when the processor 100 analyzes the status images from the RGB camera and the status images from the infrared camera 320, and detects that only the status images from one camera have foreign matters, the status images with foreign matters are sent to the cloud server 200 for target review, and the cloud server 200 further determines whether the transmission line has foreign matter invasion. Thus, the influence of the working error of the image acquisition device 300 on the detection result can be reduced, so that the working reliability of the foreign matter detection equipment of the power transmission line can be improved.
The mode of issuing the alarm is not unique, for example, sending a short message or making a call to an operation and maintenance person, informing the operation and maintenance person that the transmission line has foreign matter invasion, and the foreign matter needs to be processed in time. Alternatively, the cloud server 200 may also send an alarm signal to a connected alarm device, which may be an audible and visual alarm or a display screen, etc., to inform the foreign matter invasion in time.
In one embodiment, referring to fig. 7, step S120 includes step S121.
Step S121: after the state image is segmented, object detection is performed on the segmented sub-images respectively.
When the processor 100 performs target detection on the state image, the state image is divided to obtain more than two sub-images, then the divided sub-images are detected, if foreign matters exist in the sub-images, the foreign matters are considered to be detected in the state image where the sub-images are located, and the state image where the sub-images are located is sent to the cloud server 200 for further processing. By adopting the mode of firstly dividing the image and then analyzing and processing the sub-images, the small target in the image can be detected more easily, which is beneficial to reducing the omission ratio.
Based on the same inventive concept, the embodiment of the application also provides a foreign matter detection method of a foreign matter detection device of a power transmission line, the foreign matter detection device of the power transmission line includes an image acquisition device 300, a processor 100 and a cloud server 200, the image acquisition device 300 is connected with the processor 100, the processor 100 communicates with the cloud server 200, the image acquisition device 300 and the processor 100 are both arranged on the power transmission line, the foreign matter detection method of the foreign matter detection device of the power transmission line is applied to the cloud server 200, please refer to fig. 8, the method includes:
step S210: a processed signal is received.
The processing signal is detected by the processor 100, and the processing signal is transmitted after the detection of the presence of the foreign object in the status image, and the status image is acquired by the image acquisition device 300 through the power transmission line. The type of the processing signal is not unique, and when the image capturing apparatus 300 includes an RGB camera, the processor 100 analyzes the status image from the RGB camera, and after detecting that the status image has a foreign object, sends the status image having the foreign object to the cloud server 200, where the processing signal is the status image having the foreign object. Alternatively, when the image capturing device 300 includes an RGB camera and an infrared camera 320, after the processor 100 analyzes the status image from the RGB camera and the status image from the infrared camera 320, and detects that a status image with a foreign object exists, the processor 100 sends an alarm signal to the cloud server 200, and the processing signal at this time is the alarm signal. When the image capturing device 300 includes an RGB camera and an infrared camera 320, when the processor 100 analyzes a status image from the RGB camera and analyzes a status image from the infrared camera 320, and detects that only a status image from one camera has a foreign object, the status image having the foreign object is sent to the cloud server 200 for target review, and a processing signal at this time is the status image having the foreign object.
Step S220: and judging whether to issue an alarm according to the processing signal.
The cloud server 200 is also called a cloud server, and a more advanced algorithm can be stored in the cloud server, so that the accuracy of analysis and detection results can be improved when the images are analyzed. In this embodiment, after receiving the processing signal, the cloud server 200 determines whether to issue an alarm according to the processing signal. The process of alarm judgment by the cloud server 200 is different according to the type of the processing signal. For example, when the image capturing apparatus 300 includes an RGB camera, the processor 100 analyzes a status image from the RGB camera, and after detecting that a status image of a foreign object exists, sends the status image of the foreign object to the cloud server 200, and the processing signal at this time is the status image of the foreign object. The cloud server 200 performs target review on the state image with the foreign object to determine whether the foreign object is actually present in the state image with the foreign object. If foreign matter exists, an alarm is issued. Specifically, the mode of performing the target review is not unique, for example, the target review may be performed by using a common image processing process, including image gray level processing, contour extraction, analysis, and the like, or may be implemented by using an artificial neural network, for example, a convolutional neural network, which may be specifically selected according to actual needs, so long as those skilled in the art consider that the target review may be implemented. Generally, an algorithm used by the cloud server 200, such as an artificial neural network, is complex, the target review precision is high, the types of objects which can be identified are more, and the accuracy of the foreign matter detection result can be further ensured.
Or, when the image capturing device 300 includes the RGB camera and the infrared camera 320, after the processor 100 analyzes the status image from the RGB camera and analyzes the status image from the infrared camera 320, and detects that the status image has foreign objects, the processor 100 sends an alarm signal to the cloud server 200, the processing signal at this time is the alarm signal, and the cloud server may not analyze the image after receiving the alarm signal, and directly issue an alarm, so as to save a control flow. It can be understood that when the image capturing device 300 includes an RGB camera and an infrared camera 320, when the processor 100 analyzes a status image from the RGB camera and a status image from the infrared camera 320, and detects that only a status image from one camera has a foreign object, the status image with the foreign object is sent to the cloud server 200 for target review, and at this time, the processing signal is the status image with the foreign object, the cloud server 200 further determines whether the transmission line does have a foreign object invasion condition, and if it does, issues an alarm. Thus, the influence of the working error of the image acquisition device 300 on the detection result can be reduced, so that the working reliability of the foreign matter detection equipment of the power transmission line can be improved.
The mode of issuing the alarm is not unique, for example, sending a short message or making a call to an operation and maintenance person, informing the operation and maintenance person that the transmission line has foreign matter invasion, and the foreign matter needs to be processed in time. Alternatively, the cloud server 200 may also send an alarm signal to a connected alarm device, which may be an audible and visual alarm or a display screen, etc., to inform the foreign matter invasion in time.
In one embodiment, the image capturing device includes an RGB camera 310 and an infrared camera 320, please refer to fig. 9, and step S220 includes step S221.
Step S221: when the processing signal includes a first detection result obtained after the processor performs the target detection on the state image from the RGB camera and includes a second detection result obtained after the processor performs the target detection on the state image from the infrared camera 320, an alarm is issued.
It will be appreciated that the first detection result here is that a foreign object is detected in the status image from the RGB camera 310, and the second detection result is that a foreign object is detected in the status image from the infrared camera 320. When the image capturing device 300 includes the RGB camera and the infrared camera 320, when the processor 100 analyzes the status image from the RGB camera and analyzes the status image from the infrared camera 320, after detecting that the status image has foreign objects, the processor 100 sends an alarm signal to the cloud server 200, the processing signal at this time is the alarm signal, and the cloud server may not analyze the image after receiving the alarm signal, and directly issue an alarm, so as to save a control flow.
It can be appreciated that, in other embodiments, when the image capturing device 300 includes an RGB camera and an infrared camera 320, if the processor 100 analyzes a status image from the RGB camera and analyzes a status image from the infrared camera 320, and detects that only a foreign object exists in the status image from one camera, the status image with the foreign object is sent to the cloud server 200 for target review, and the processing signal at this time is the status image with the foreign object, the cloud server 200 performs target review on the status image with the foreign object, further determines whether the foreign object invasion situation exists in the transmission line, and issues an alarm if it is determined that the foreign object invasion situation does exist. Thus, the influence of the working error of the image acquisition device 300 on the detection result can be reduced, so that the working reliability of the foreign matter detection equipment of the power transmission line can be improved.
When the cloud server 200 performs target rechecks on the state image with the foreign matters, the state image with the foreign matters is divided to obtain more than two sub-images, then the divided sub-images are detected, if the foreign matters are detected in the sub-images, the state image with the foreign matters in which the sub-images are located is considered to be truly provided with the foreign matters, and an alarm is issued. By adopting the mode of firstly dividing the image and then analyzing and processing the sub-images, the small target in the image can be detected more easily, which is beneficial to reducing the omission ratio.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.
Claims (5)
1. The power transmission line foreign matter detection equipment is characterized by comprising an image acquisition device, a processor and a cloud server, wherein the image acquisition device is connected with the processor, the processor is communicated with the cloud server, and the image acquisition device and the processor are both arranged on a power transmission line;
the image acquisition device is used for acquiring a state image of the power transmission line and sending the state image to the processor, the processor is used for carrying out target detection on the state image, after detecting that the state image with foreign matters exists, a processing signal is sent to the cloud server, and the cloud server is used for judging whether to issue an alarm according to the processing signal;
The image acquisition device comprises an RGB camera and an infrared camera, and the RGB camera and the infrared camera are both connected with the processor; the number of the RGB cameras is more than two, the number of the infrared cameras is more than two, the RGB cameras and the infrared cameras are arranged at different positions, and the RGB cameras and the infrared cameras have the same shooting angle;
the processor is used for dividing the state image from the RGB camera, respectively carrying out target detection on the divided sub-images, and if foreign matters exist in the sub-images, sending the state image where the sub-images are located to the cloud server as a processing signal; the processor is used for dividing the state image from the RGB camera by adopting a halving or octaving method;
the processor directly detects the target without performing image segmentation when detecting the target of the state image from the infrared camera;
when the processor analyzes the state image from the RGB camera and the state image from the infrared camera, and detects the state image with foreign matters, the processor sends an alarm signal to the cloud server, and the cloud server issues an alarm after receiving the alarm signal;
When the processor analyzes the state images from the RGB cameras and analyzes the state images from the infrared cameras, and when detecting that only the state images from one camera have foreign matters, the state images with the foreign matters are sent to the cloud server for target rechecking, the cloud server further judges whether the transmission line has the foreign matter invasion situation or not, and if the transmission line has the foreign matter invasion situation, an alarm is issued; the processor is used for carrying out target detection on the state image based on an artificial neural network model with a first weight, the cloud server is used for carrying out target rechecking on the received state image with foreign matters based on an artificial neural network model with a second weight, and parameters of the first weight are smaller than those of the second weight.
2. The transmission line foreign matter detection apparatus according to claim 1, wherein the processor includes a microcontroller unit and an embedded neural network processing unit, both of which are connected to the microcontroller unit, the microcontroller unit being in communication with the cloud server;
The microcontroller unit is used for receiving the state image, transmitting the state image to the embedded neural network processing unit, performing target detection on the state image by the embedded neural network processing unit, and transmitting a processing signal to the cloud server through the microcontroller unit after detecting the state image with foreign matters.
3. The transmission line foreign matter detection apparatus according to claim 1, wherein the processor detects with an artificial neural network model having different weights for the RGB image and the infrared image.
4. The foreign matter detection method of the foreign matter detection equipment of the power transmission line is characterized in that the foreign matter detection equipment of the power transmission line comprises an image acquisition device, a processor and a cloud server, wherein the image acquisition device is connected with the processor, the processor is communicated with the cloud server, and the image acquisition device and the processor are both arranged on the power transmission line; the image acquisition device comprises an RGB camera and an infrared camera, and the RGB camera and the infrared camera are both connected with the processor; the number of the RGB cameras is more than two, the number of the infrared cameras is more than two, the RGB cameras and the infrared cameras are arranged at different positions, and the RGB cameras and the infrared cameras have the same shooting angle;
The foreign matter detection method of the transmission line foreign matter detection device is applied to the processor, and comprises the following steps:
acquiring a state image of the power transmission line;
performing target detection on the state image;
after detecting a state image with foreign matters, sending a processing signal to the cloud server, and judging whether to issue an alarm or not by the cloud server according to the processing signal;
after the state image from the RGB camera is segmented, respectively carrying out target detection on the segmented sub-images, and if foreign matters exist in the sub-images, sending the state image where the sub-images are located to the cloud server as a processing signal; the processor is used for dividing the state image from the RGB camera by adopting a halving or octaving method; when the state image from the infrared camera is subjected to target detection, the step of image segmentation is not performed, and the target detection is directly performed;
when the state images from the RGB camera and the state images from the infrared camera are analyzed, after the state images with foreign matters are detected, an alarm signal is sent to the cloud server, and the cloud server issues an alarm after receiving the alarm signal;
When the state images from the RGB cameras are analyzed and the state images from the infrared cameras are analyzed, and when the state images from only one camera are detected to have foreign matters, the state images with the foreign matters are sent to the cloud server for target rechecking, the cloud server further judges whether the transmission line has the foreign matter invasion situation or not, and if yes, an alarm is issued; the processor is used for carrying out target detection on the state image based on an artificial neural network model with a first weight, the cloud server is used for carrying out target rechecking on the received state image with foreign matters based on an artificial neural network model with a second weight, and parameters of the first weight are smaller than those of the second weight.
5. The foreign matter detection method of the foreign matter detection equipment of the power transmission line is characterized in that the foreign matter detection equipment of the power transmission line comprises an image acquisition device, a processor and a cloud server, wherein the image acquisition device is connected with the processor, the processor is communicated with the cloud server, and the image acquisition device and the processor are both arranged on the power transmission line; the image acquisition device comprises an RGB camera and an infrared camera, and the RGB camera and the infrared camera are both connected with the processor; the number of the RGB cameras is more than two, the number of the infrared cameras is more than two, the RGB cameras and the infrared cameras are arranged at different positions, and the RGB cameras and the infrared cameras have the same shooting angle;
The foreign matter detection method of the transmission line foreign matter detection equipment is applied to the cloud server, and the method comprises the following steps:
receiving a processing signal; the processing signal is subjected to target detection on a state image by the processor based on an artificial neural network model with a first weight, and is sent after the state image with foreign matters is detected, and the state image is acquired by the image acquisition device and is obtained by acquiring the power transmission line;
judging whether to issue an alarm according to the processing signal;
the image acquisition device comprises an RGB camera and an infrared camera, judges whether to issue an alarm according to the processing signal, and comprises:
after the state image from the RGB camera is segmented, respectively carrying out target detection on the segmented sub-images, and if foreign matters exist in the sub-images, sending the state image where the sub-images are located to the cloud server as a processing signal; dividing a state image from the RGB camera by adopting a bisection or octaion method; when the state image from the infrared camera is subjected to target detection, the step of image segmentation is not performed, and the target detection is directly performed;
When the processing signal comprises a first detection result obtained after the processor performs target detection on the state image from the RGB camera and a second detection result obtained after the processor performs target detection on the state image from the infrared camera, an alarm is issued; the first detection result is that foreign matters are detected in a state image from the RGB camera, and the second detection result is that foreign matters are detected in a state image from the infrared camera;
when the processing signal comprises the first detection result obtained after the processor performs target detection on the state image from the RGB camera or comprises the second detection result obtained after the processor performs target detection on the state image from the infrared camera, performing target recheck on the state image with the foreign matters based on an artificial neural network model with second weight, further judging whether the foreign matter invasion condition exists in the power transmission line, and if so, issuing an alarm; the first weight has a smaller parameter than the second weight.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210288689.7A CN114627388B (en) | 2022-03-23 | 2022-03-23 | Foreign matter detection equipment and foreign matter detection method for power transmission line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210288689.7A CN114627388B (en) | 2022-03-23 | 2022-03-23 | Foreign matter detection equipment and foreign matter detection method for power transmission line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114627388A CN114627388A (en) | 2022-06-14 |
| CN114627388B true CN114627388B (en) | 2024-03-29 |
Family
ID=81903594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210288689.7A Active CN114627388B (en) | 2022-03-23 | 2022-03-23 | Foreign matter detection equipment and foreign matter detection method for power transmission line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114627388B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115601707B (en) * | 2022-11-03 | 2024-01-23 | 国网湖北省电力有限公司荆州供电公司 | On-line monitoring method and system for power transmission line of power system |
| CN115855166B (en) * | 2023-02-21 | 2023-05-05 | 中国市政工程西南设计研究总院有限公司 | Abnormal supervision and management platform of intelligent communication machine room |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20100059135A (en) * | 2008-11-26 | 2010-06-04 | 한전케이피에스 주식회사 | Perception system for power-tranmission line |
| WO2016078081A1 (en) * | 2014-11-21 | 2016-05-26 | 河南送变电工程公司 | Multi-rotor-wing inspection aircraft and power transmission line inspection system |
| CN107424373A (en) * | 2017-08-02 | 2017-12-01 | 国网北京市电力公司 | External force intrusion detection method and apparatus and system |
| CN109631848A (en) * | 2018-12-14 | 2019-04-16 | 山东鲁能软件技术有限公司 | Electric line foreign matter intruding detection system and detection method |
| WO2020062088A1 (en) * | 2018-09-28 | 2020-04-02 | 安徽继远软件有限公司 | Image identification method and device, storage medium, and processor |
| CN111600383A (en) * | 2020-05-12 | 2020-08-28 | 合肥中科类脑智能技术有限公司 | Intelligent integrated inspection device for power transmission line |
| CN112491982A (en) * | 2020-11-13 | 2021-03-12 | 国网天津市电力公司 | Refined sensing method based on cloud edge cooperative power transmission line |
| CN113160150A (en) * | 2021-04-01 | 2021-07-23 | 西安科技大学 | AI (Artificial intelligence) detection method and device for invasion of foreign matters in wire network based on multi-type sample fusion and multi-complex network |
-
2022
- 2022-03-23 CN CN202210288689.7A patent/CN114627388B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20100059135A (en) * | 2008-11-26 | 2010-06-04 | 한전케이피에스 주식회사 | Perception system for power-tranmission line |
| WO2016078081A1 (en) * | 2014-11-21 | 2016-05-26 | 河南送变电工程公司 | Multi-rotor-wing inspection aircraft and power transmission line inspection system |
| CN107424373A (en) * | 2017-08-02 | 2017-12-01 | 国网北京市电力公司 | External force intrusion detection method and apparatus and system |
| WO2020062088A1 (en) * | 2018-09-28 | 2020-04-02 | 安徽继远软件有限公司 | Image identification method and device, storage medium, and processor |
| CN109631848A (en) * | 2018-12-14 | 2019-04-16 | 山东鲁能软件技术有限公司 | Electric line foreign matter intruding detection system and detection method |
| CN111600383A (en) * | 2020-05-12 | 2020-08-28 | 合肥中科类脑智能技术有限公司 | Intelligent integrated inspection device for power transmission line |
| CN112491982A (en) * | 2020-11-13 | 2021-03-12 | 国网天津市电力公司 | Refined sensing method based on cloud edge cooperative power transmission line |
| CN113160150A (en) * | 2021-04-01 | 2021-07-23 | 西安科技大学 | AI (Artificial intelligence) detection method and device for invasion of foreign matters in wire network based on multi-type sample fusion and multi-complex network |
Non-Patent Citations (3)
| Title |
|---|
| Jinguo Zhu,et al.A Deep Learning Method to Detect Foreign Objects for Inspecting Power Transmission Lines.《IEEE Access》.2020,94065 - 94075页. * |
| 赵晓鹏.输电线路异物检测研究.《中国优秀硕士学位论文全文数据库》.2021,第2021年卷(第04期),C042-461. * |
| 输电线路异物检测研究;赵晓鹏;《中国优秀硕士学位论文全文数据库》;第2021年卷(第04期);C042-461 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114627388A (en) | 2022-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109068099B (en) | Virtual electronic fence monitoring method and system based on video monitoring | |
| CN114627388B (en) | Foreign matter detection equipment and foreign matter detection method for power transmission line | |
| US9965835B2 (en) | Defogging images and video | |
| CN110516522B (en) | Inspection method and system | |
| CN113052029A (en) | Abnormal behavior supervision method and device based on action recognition and storage medium | |
| CN113095114B (en) | Method, device and equipment for detecting oil leakage of hydraulic system | |
| CN112733690A (en) | High-altitude parabolic detection method and device and electronic equipment | |
| CN113096158A (en) | Moving object identification method and device, electronic equipment and readable storage medium | |
| CN113313006B (en) | Urban illegal building supervision method, system and storage medium based on unmanned aerial vehicle | |
| CN113887412A (en) | Detection method, detection terminal, monitoring system and storage medium for pollution emission | |
| CN110648489A (en) | Storehouse early warning system based on infrared thermal imaging ignition point identification | |
| CN116797977A (en) | Method and device for identifying dynamic target of inspection robot and measuring temperature and storage medium | |
| CN115082813A (en) | Detection method, unmanned aerial vehicle, detection system and medium | |
| CN110728212B (en) | Road well lid monitoring device and monitoring method based on computer vision | |
| CN114878005A (en) | Temperature detection method, device, system, electronic equipment and storage medium | |
| CN110827317B (en) | Four-eye moving object detection and identification equipment and method based on FPGA | |
| CN117253120A (en) | Fire disaster identification method, device and storage medium | |
| CN116567402A (en) | Light source controller trigger delay control method and system | |
| CN111145219B (en) | Efficient video moving target detection method based on Codebook principle | |
| CN113629574A (en) | Strong wind sand area transmission conductor galloping early warning system based on video monitoring technology | |
| CN111079682A (en) | Cabin supervision method and system based on artificial intelligence | |
| CN113496136A (en) | Unmanned aerial vehicle and image identification method thereof | |
| CN113311301B (en) | Method and system for testing insulation performance of electrical equipment | |
| CN110930362A (en) | Screw safety detection method, device and system | |
| CN113705502B (en) | Ship target behavior understanding system integrating target detection and target tracking |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: Room 86, room 406, No.1, Yichuang street, Zhongxin Guangzhou Knowledge City, Huangpu District, Guangzhou City, Guangdong Province Patentee after: Southern Power Grid Digital Grid Research Institute Co.,Ltd. Country or region after: China Address before: Room 86, room 406, No.1, Yichuang street, Zhongxin Guangzhou Knowledge City, Huangpu District, Guangzhou City, Guangdong Province Patentee before: Southern Power Grid Digital Grid Research Institute Co.,Ltd. Country or region before: China |