CN112383137A - Transformer area monitoring system and method based on machine vision and thermal imaging technology - Google Patents
Transformer area monitoring system and method based on machine vision and thermal imaging technology Download PDFInfo
- Publication number
- CN112383137A CN112383137A CN202011075079.6A CN202011075079A CN112383137A CN 112383137 A CN112383137 A CN 112383137A CN 202011075079 A CN202011075079 A CN 202011075079A CN 112383137 A CN112383137 A CN 112383137A
- Authority
- CN
- China
- Prior art keywords
- thermal imaging
- data
- transformer
- module
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001931 thermography Methods 0.000 title claims abstract description 88
- 238000012544 monitoring process Methods 0.000 title claims abstract description 87
- 238000005516 engineering process Methods 0.000 title claims abstract description 41
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000002159 abnormal effect Effects 0.000 claims abstract description 67
- 230000001133 acceleration Effects 0.000 claims abstract description 30
- 238000013135 deep learning Methods 0.000 claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 26
- 238000004891 communication Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 22
- 230000008859 change Effects 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 17
- 238000003384 imaging method Methods 0.000 claims description 6
- 238000011144 upstream manufacturing Methods 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 10
- 238000004458 analytical method Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 8
- 230000007547 defect Effects 0.000 description 8
- 238000005286 illumination Methods 0.000 description 8
- 230000005540 biological transmission Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000005856 abnormality Effects 0.000 description 5
- 239000013589 supplement Substances 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000004880 explosion Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000009795 derivation Methods 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 238000010438 heat treatment Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 239000007769 metal material Substances 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00002—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by monitoring
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/70—Circuitry for compensating brightness variation in the scene
- H04N23/74—Circuitry for compensating brightness variation in the scene by influencing the scene brightness using illuminating means
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J2005/0077—Imaging
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a transformer platform area monitoring system and a method based on machine vision and thermal imaging technology, wherein the system comprises: the camera module is used for acquiring video data of a detected target in the transformer area; the infrared thermal imaging module is used for acquiring thermal imaging picture data of a detected target in the transformer area; the DMP module is used for acquiring the three-axis acceleration, the three-axis angular velocity and the azimuth information of the position of the transformer area monitoring system; the core control module is used for identifying the detected target by utilizing video data, thermal imaging picture data, three-axis acceleration, three-axis angular velocity and/or azimuth information, acquiring abnormal information of the detected target and carrying out deep learning on the abnormal information by utilizing a neural network so as to predict non-sudden faults in advance. The system and the method can realize the automatic detection and prediction of various faults of the transformer area, and can improve the detection accuracy of the system through autonomous learning.
Description
Technical Field
The invention belongs to the technical field of power system monitoring, and particularly relates to a transformer area monitoring system and method based on machine vision and thermal imaging technologies.
Background
At present, in a variable distribution system in China, due to the characteristics of large quantity, wide distribution area, dispersed positions, frequent and irregular changes (mainly including position changes, capacity changes or quantity increase and decrease), outdoor installation and the like of distribution transformers, the monitoring of the distribution transformers is often careless, the management level is lower, and a transformer area cannot be attended by people all the time, so that human resources are wasted, and the requirements on automatic management and monitoring of the transformer area are increasingly urgent, and the currently common monitoring method of the transformer area comprises the following steps: monitoring based on a private network online video system and monitoring based on a handheld thermal imaging system.
The monitoring of the online video system based on the private network has the following defects: the application scene limitation is large, and a monitored point needs to deploy a special video transmission network; the data flow is large, all video streams need to be transmitted to a centralized system background for centralized analysis, and the video streams generate huge communication flow; the central data processing server has high configuration requirements, all video streams need to be transmitted to the central server, and the central server analyzes the data streams, so that the required server configuration and calculation requirements are high, and when the server fails, all related nodes are monitored and failed; the abnormal event response is slow, and the video data needs to be transmitted to a central server through a communication network for analysis and calculation.
The monitoring based on the handheld thermal imaging system has the following defects: the system monitoring can not be carried out on line in 24 hours; the manual inspection is required to be carried out regularly, the labor is wasted, and the cost is high; the failure cannot be predicted in real time, and the accuracy of the judgment and prediction of the abnormality cannot be continuously improved.
In addition, the existing transformer area monitoring system is generally single in monitoring function and can only detect single or partial specific faults.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a transformer area monitoring system and method based on machine vision and thermal imaging technologies. The technical problem to be solved by the invention is realized by the following technical scheme:
one aspect of the present invention provides a transformer area monitoring system based on machine vision and thermal imaging technology, including:
the camera module is used for acquiring video data of a detected target in the transformer area;
the infrared thermal imaging module is used for acquiring thermal imaging picture data of a detected target in the transformer area;
the DMP module is used for acquiring the three-axis acceleration, the three-axis angular velocity and the azimuth information of the position of the transformer area monitoring system; and
the core control module is used for identifying the detected target by utilizing the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity and/or the azimuth information, acquiring abnormal information of the detected target and carrying out deep learning on the abnormal information by utilizing a neural network so as to predict non-sudden faults in advance.
In an embodiment of the present invention, the camera module includes a CMOS image sensor and an infrared fill-in light lamp, where the infrared fill-in light lamp is used to fill in light for the monitored target when ambient light is lower than minimum light required for imaging of the CMOS image sensor.
In one embodiment of the present invention, the core control module includes:
a data processing module, configured to decode or segment the video data, the thermal imaging picture data, the current triaxial acceleration, the current triaxial angular velocity, and the current orientation information to form a single-frame data sequence;
the deep learning operation module is used for identifying the data abnormal type by utilizing the trained neural network according to the single-frame data sequence and outputting a frame data sequence with fault identification information;
an anomaly classification module for classifying the types of the detected abnormal events according to the frame data sequence with the fault identification information,
the data processing module is further used for obtaining temperature information of the area where the detected target is located according to the thermal imaging picture data, judging the temperature information and generating temperature abnormal information of different levels according to the relationship between the temperature information and different thresholds.
In an embodiment of the present invention, the core control module is further configured to:
and continuously training the neural network in the deep learning operation module by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
In one embodiment of the invention, the transformer platform area monitoring system based on the machine vision and thermal imaging technology further comprises a power supply, and a power switch, a power indicator lamp and a status indicator lamp which are connected to the power supply, wherein the status indicator lamp is used for indicating the status of the detected object through different combinations of colored lamps and different flashing frequencies.
In one embodiment of the invention, the transformer platform area monitoring system based on machine vision and thermal imaging technology further comprises a communication module, wherein the communication module comprises an uplink communication unit and a downlink communication unit,
the uplink communication unit is used for electrically connecting the transformer area monitoring system to the TTU;
the downlink communication unit is used for electrically connecting the camera module and the infrared thermal imaging module to the core control module.
Another aspect of the present invention provides a transformer area monitoring method based on machine vision and thermal imaging technologies, including:
s1: collecting video data of a detected target in a transformer area;
s2: acquiring thermal imaging picture data of a detected target in the transformer area;
s3: acquiring three-axis acceleration, three-axis angular velocity and azimuth information of the transformer area monitoring system;
s4: and identifying the detected target by utilizing the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity and/or the azimuth information, acquiring abnormal information of the detected target, and deeply learning the abnormal information by utilizing a neural network so as to predict the non-sudden fault in advance.
In an embodiment of the present invention, the S4 includes:
s41: decoding or segmenting the video data, the thermal imaging picture data, the current triaxial acceleration, the current triaxial angular velocity, and the current orientation information to form a single frame data sequence;
s42: according to the single-frame data sequence, recognizing the data abnormal type by utilizing a trained neural network and outputting a frame data sequence with recognition information;
s43: and classifying the types of the detected abnormal events according to the frame data sequence with the identification information.
In an embodiment of the present invention, the S4 further includes:
and obtaining the temperature information of the area where the detected target is located according to the thermal imaging picture data, judging the temperature information, and generating temperature abnormal information of different levels according to the relationship between the temperature information and different thresholds.
In an embodiment of the present invention, the S4 further includes:
and continuously training the neural network by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
Compared with the prior art, the invention has the beneficial effects that:
1. the transformer area monitoring system based on machine vision and thermal imaging technology can realize automatic detection and automatic prediction of various faults of a transformer area through vision technology, thermal imaging technology and area acceleration angle speed technology, meanwhile, accuracy of prediction and identification can be improved by continuously iterating historical data and data change trend characteristics according to big data and deep learning technology, detection accuracy is superior to that of a traditional single threshold judgment method, the system can be effectively combined with the existing TTU, extra network configuration is not needed, operation pressure of a central server is not increased, and millisecond-level abnormal fault detection can be realized.
2. The transformer area monitoring system uses distributed analysis and calculation, namely intelligent analysis is carried out on a single transformer area monitoring system node, only the type of the generated abnormal event and the corresponding abnormal picture are transmitted to a TTU or a remote monitoring center for displaying and early warning, the transmitted data volume is far smaller than that of the prior art, the configuration requirement of a central server is very low or the central algorithm server is not needed, and the data monitoring of other nodes is not influenced when a single node breaks down.
The present invention will be described in further detail with reference to the accompanying drawings and examples.
Drawings
Fig. 1 is a block diagram of a transformer area monitoring system based on machine vision and thermal imaging technologies according to an embodiment of the present invention;
FIG. 2 is a block diagram of a core control module according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a data processing procedure of a core control module according to an embodiment of the present invention;
fig. 4 is a detailed structural diagram of a transformer area monitoring system based on machine vision and thermal imaging technologies according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a data processing procedure of a deep learning operation module according to an embodiment of the present invention;
fig. 6 is a flowchart of a transformer platform monitoring system based on machine vision and thermal imaging technologies according to an embodiment of the present invention.
Detailed Description
In order to further illustrate the technical means and effects of the present invention adopted to achieve the predetermined invention purpose, a transformer area monitoring system and method based on machine vision and thermal imaging technology according to the present invention will be described in detail below with reference to the accompanying drawings and the detailed description.
The foregoing and other technical matters, features and effects of the present invention will be apparent from the following detailed description of the embodiments, which is to be read in connection with the accompanying drawings. The technical means and effects of the present invention adopted to achieve the predetermined purpose can be more deeply and specifically understood through the description of the specific embodiments, however, the attached drawings are provided for reference and description only and are not used for limiting the technical scheme of the present invention.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that an article or device that comprises a list of elements does not include only those elements but may include other elements not expressly listed. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of additional like elements in the article or device comprising the element.
Example one
Referring to fig. 1, fig. 1 is a block diagram of a transformer area monitoring system based on machine vision and thermal imaging technologies according to an embodiment of the present invention. The transformer area monitoring system is arranged at the position of a detected target needing to be monitored in a transformer area, and comprises a camera module 1, an infrared thermal imaging module 2, a DMP module 3 and a core control module 4, wherein the camera module 1, the infrared thermal imaging module 2 and the DMP module 3 are respectively and electrically connected with the core control module 4. The camera module 1 is used for acquiring video data of a detected target in a transformer area; the infrared thermal imaging module 2 is used for acquiring thermal imaging picture data of a detected target in the transformer area; the DMP module 3 is used for acquiring the position of the transformer area monitoring system in real time, namely the three-axis acceleration, the three-axis angular velocity and the azimuth information of the detected target; the core control module 4 is configured to identify the detected target by using the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity, and/or the azimuth information, obtain abnormal information of the detected target, and perform deep learning on the abnormal information by using a neural network to predict a non-sudden failure in advance.
Specifically, the camera module 1 monitors a detected target (e.g., a fuse, a power transmission line, a transformer body, a distribution box, etc.) in the transformer area, acquires an original video stream of the detected target, encodes the original video stream into an H264/H265 video stream, and transmits the encoded stream to the core control module 4 through the ethernet port.
In this embodiment, the camera module (see IPC-RGB in fig. 4) 1 is connected to the core control module 4 through a standard POE network interface, data transmission and power supply are transmitted through an ethernet data line, and the core control module 4 can set the operating mode and parameters of the camera module 1.
In practical situations, if the transformer area is in a dark environment with weak light or no light, the signal-to-noise ratio of the video in the original video stream is greatly reduced, and the identification accuracy is seriously affected. To address this problem, the camera module 1 of this embodiment preferably includes a CMOS image sensor and an infrared fill-in light, where the infrared fill-in light is used to fill in light for the monitored target when ambient light is lower than the minimum required imaging light of the CMOS image sensor.
Specifically, the camera module 1 adopts the CMOS image sensor supporting low illumination, and can improve the signal-to-noise ratio in a low-illumination environment (for example, evening, dawn, and evening with moonlight), but imaging is still not ideal in a completely dark environment, and therefore, the camera module 1 may include a visible light fill light, and when ambient illumination is lower than the minimum required illumination for imaging of the CMOS image sensor, the visible light fill light of the camera module 1 is turned on to fill light to the transformer platform area to enhance illumination, so that a video stream acquired by the camera module 1 has a higher signal-to-noise ratio, and the false alarm rate is reduced.
However, the visible light may cause a certain light pollution, and therefore, the camera module 1 of the present embodiment preferably includes an infrared fill light in a use scenario where the light pollution is required. When the ambient illumination is lower than the minimum required illumination for imaging of the CMOS image sensor, the infrared light supplement lamp of the camera module 1 is turned on to supplement light to the transformer platform area to enhance illumination, so that the video stream acquired by the camera module 1 has a higher signal-to-noise ratio, and the false alarm rate is reduced.
It should be noted that, under the condition of light supplement by the infrared light supplement lamp, the CMOS image sensor may not collect color video streams of three primary colors of RGB, but may collect only a monochrome gray-scale video stream, and at this time, the mode is switched to the gray-scale video mode.
In addition, when the external light is strong, especially when the light is strongly irradiated to the top of the transformer, because the top of the transformer is made of metal material, part of the reflected light is generated, so that other areas in the picture are normal, and the top of the transformer is over-exploded, so that the image details at the top of the transformer are lost, and the final recognition result is affected. Aiming at the problem that the transformer area belongs to a static target background (namely, the detected target does not move frequently), the embodiment adopts a frame-by-frame exposure splicing mode to avoid the problem of image over-explosion. Specifically, a default exposure parameter A is used for detecting the current video stream, if an over-burst area C is detected, the exposure parameter is adjusted to be B, so that the over-burst area is normal (because the other areas are fully exposed, the underexposure problem occurs); respectively using the exposure parameter A and the exposure parameter B to carry out source data acquisition at intervals; splicing A, B parameter exposed interval frames (B parameter is an image outside an over-explosion area, A parameter is an over-explosion area C, and the two frames are combined into one frame); the final result is that the image frame rate is reduced by half to eliminate the problem of over-burst reflection.
Further, an infrared thermal imaging module (see IPC-ITC in fig. 4) 2 is connected to the core control module 4 through a standard POE interface, and power and data exchange are performed through a network cable. The infrared thermal imaging module 2 monitors a detected target (such as a fuse, a power transmission line, a transformer body, a distribution box and the like) of the transformer platform area, monitors the temperature of the power equipment of the transformer platform area in real time by using an infrared thermal imaging technology, finally generates thermal imaging picture data, and sends the thermal imaging picture data to the core control module 4 through an Ethernet port.
The DMP module 3 of the embodiment can acquire the current triaxial acceleration, triaxial angular velocity and current azimuth information in real time; the sampled original data are transmitted to the core control module 4, the core control module 4 can automatically learn the acceleration, the angular velocity and the direction change caused by the change of the vibration, the micro displacement and the posture generated by the impact of the transformer area, the historical data and the data change trend characteristics, and can judge and early warn the non-sudden abnormal state in advance, such as: other foreign matters of the vehicle impact the platform area, the tower of the platform area inclines, and whether the platform area is damaged by external force or not.
Referring to fig. 2, fig. 2 is a block diagram of a core control module according to an embodiment of the present invention. The core control module 4 of this embodiment includes a data processing module 41, a deep learning operation module 42, and an anomaly classification module 43, where the data processing module 41 is configured to decode or segment an original video stream, thermal imaging picture data, a current triaxial acceleration, a current triaxial angular velocity, and current orientation information to form a single-frame data sequence; the deep learning operation module 42 is used for identifying the data abnormal type by using the trained neural network according to the single-frame data sequence and outputting a frame data sequence with fault identification information; the anomaly classification module 43 is configured to classify the detected fault type according to the frame data sequence with the fault identification information.
In addition, the data processing module 41 is further configured to obtain temperature information of an area where the detected target is located according to the thermal imaging picture data, determine the temperature information, and generate different levels of temperature anomaly information according to a relationship between the temperature information and different thresholds. Specifically, after the core control module 4 acquires thermal imaging picture data of the transformer platform area, the data processing module 41 may determine the temperature of the entire area where the thermal imaging picture is located, when the absolute temperature of a certain hot spot is higher than a preset value or the relative temperature of a certain hot spot (that is, the temperature of the hot spot is higher than the ambient temperature or the temperature of an area near the hot spot), different levels of temperature anomaly information (emergency defect, serious defect, general defect, normal) or a rapid temperature rise is generated according to different thresholds or different anomaly types are generated according to historical data and data change trend characteristics, and the original video stream, the original thermal imaging picture data, and the anomaly types are sent to the remote monitoring device.
Referring to fig. 3, fig. 3 is a schematic diagram of a data processing process of a core control module according to an embodiment of the present invention. The method comprises the steps of inputting an original data stream (comprising an original video stream, thermal imaging picture data, triaxial acceleration, triaxial angular velocity and azimuth information data stream) into a data processing module 41, firstly, decoding or cutting the data stream by the data processing module 41 to divide the data stream into single-frame data sequences, then, inputting the single-frame data sequences into a deep learning operation module 42, identifying the data abnormal type by the deep learning operation module 42, outputting a frame data sequence with identification information, classifying detected abnormal events by an abnormal classification module 43, and distributing the abnormal events to corresponding service processing units (which can be displayed or stored according to the specific abnormal type or directly transmitted to a remote monitoring platform).
It should be noted that, the Transformer area monitoring system of this embodiment only transmits the generated abnormal event type and the image corresponding to the abnormality to a TTU (distribution Transformer monitoring Terminal) or a remote monitoring center, and the amount of transmitted data is much smaller than that of the prior art; in addition, the transformer area monitoring system uses distributed analysis and calculation, intelligent analysis is carried out on a single transformer area monitoring system node, the result is transmitted to a TTU or a remote monitoring center for display and early warning, the central server configuration requirement is low or a central algorithm server is not needed, and the data monitoring of other nodes is not influenced when a single node breaks down.
Further, please refer to fig. 5, wherein fig. 5 is a schematic diagram of a data processing process of a deep learning operation module according to an embodiment of the present invention. The data processing process of the deep learning operation module 42 is as follows:
model training: marking actual data generated in the actual monitoring operation process of the monitoring system, specific artificial fault data and original data streams (including original video streams, thermal imaging picture data and three-axis acceleration, three-axis angular velocity and orientation information data streams generated by the DMP module 31) to generate labels (labels); a neural network for deep learning is constructed, the neural network in the deep learning operation module 42 of the embodiment includes three layers, namely an input layer, a hidden layer and an output layer, wherein the number and the number of the neuron layers can be adjusted according to specific conditions; inputting the training data set into a neural network, performing forward propagation calculation through the neural network, and outputting a calculation result of a layer and an output result of the whole neural network through a series of processing; back propagation, namely calculating an error by using the output result and the label, and performing back propagation to update the characteristics of the neural network; and performing iterative updating to generate a final trained neural network model.
And (3) target result identification: firstly, loading neural network model data generated by the last training, then inputting an identification data set, namely the single-frame data sequence, into an input layer, processing the input layer through neurons of each layer, and outputting the final identification results of image identification, thermal imaging trend identification prediction, acceleration, angular velocity data element characteristics and the like.
By constructing different neural networks and labels, the deep learning operation module 432 of the present embodiment can effectively integrate the following types of fault detection:
1) equipment insulation damage caused by natural disasters such as thunder, rainstorm, freezing and the like, which causes partial discharge, short-circuit fault and the like; 2) the vehicle impact causes the inclination of a pole tower or the falling of a fuse, the falling of sundries such as branches and the like to a line to cause the abnormity caused by external force damage factors such as short circuit fault and the like; 3) abnormality caused by electrical equipment self factors such as abnormal heating problems caused by equipment aging, overload operation and the like; 4) the on-off state of switching equipment such as a drop-out fuse, an isolation knife and the like; 5) whether dangerous foreign matters fall onto the electrical equipment or not is judged; 6) abnormal heating is avoided at the positions of distribution transformers, wire connectors and the like; 7) whether the equipment is damaged by external force or not; 8) whether the arc light discharge phenomenon exists or not; 9) and monitoring illegal personnel invasion or site construction.
Further, the core control module 4 is further configured to: and continuously training the neural network in the deep learning operation module by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
Specifically, the core control module 4 continuously acquires the temperature information in the monitored area from the thermal imaging camera, and the deep learning operation module 42 may also acquire this information when the device generates abnormal information. The deep learning operation module 42 can continuously acquire the collected normal data and abnormal information and can realize early warning of abnormal events according to the historical data and the data change trend characteristics. The deep learning operation module 42 can perform autonomous learning through the acquired data, perform combined iteration and training, continuously improve the accuracy of the algorithm, and predict the non-sudden faults of the transformer area which are accumulated and gradually worsened for a long time in advance. For example, the transformer pile head is oxidized or loosened due to external reasons, so that internal resistance is increased to heat the connection point and finally cause burning, the process is not generated in a short time but is gradually heated until the connection point is burnt, the core control module 4 can effectively and timely give an early warning for the non-sudden abnormality, and then a worker can perform advanced treatment according to the early warning to reduce power failure loss caused by the burning damage of the transformer area.
It should be noted that the deep learning operation module 42 may implement data sharing. Specifically, the monitoring systems distributed in different areas can continuously generate accumulated deep learning original data, and two modes can be supported, so that the monitoring systems distributed in different areas share the acquired original data and the deep learning model, and the prediction and judgment capabilities of a single monitoring system are improved. The first method is as follows: training original data and training models generated by a single monitoring system are regularly and uniformly uploaded to a central server, and the data of the central server is regularly synchronized to different monitoring systems to realize the sharing of the training data and the training results before different devices; the second method comprises the following steps: for an offline or restrictive network, relevant data can be exported through a USB interface on a monitoring system, and after centralized processing, the relevant data is regularly updated to the monitoring system by using a USB flash disk, so that the sharing of previous training data and training results of different monitoring systems is realized.
Further, referring to fig. 4, the transformer platform area monitoring system further includes a power supply, and a power switch, a power indicator and a status indicator connected to the power supply, wherein the status indicator is used for indicating the status of the detected object through different combinations of colored lights and different flashing frequencies. In this embodiment, the status indicator light is a three-color light.
Further, the transformer station area monitoring system further comprises a communication module, wherein the communication module comprises an uplink communication unit and a downlink communication unit, the uplink communication unit is used for electrically connecting the transformer station area monitoring system to the TTU or other equipment, the uplink communication unit comprises a 1-path gigabit network, a path of RS485, a path of RS232 and a built-in 4G module, and communication requirements of different upper-layer equipment can be met. Most of the existing transformer areas are equipped with intelligent TTU terminals, the transformer area monitoring system of the embodiment can communicate with the TTU through a gigabit network or an RS485 interface, and the TTU can set and acquire the working mode and relevant parameters of the transformer area monitoring system. The transformer area monitoring system uploads the detected abnormal result to the TTU in real time, only needs to be physically connected with the TTU, and does not need to be provided with a dedicated communication network. The downlink communication unit comprises a standard POE interface for electrically connecting the camera module 1 and the infrared thermal imaging module 2 to the core control module 4. Specifically, the transformer platform district monitored control system of this embodiment designs 6 way standard POE interfaces, supports IEEE802.3af/at/bt standard protocol, can support power supply and communication, can external POE color camera, the POE thermal imaging camera etc. that support standard protocol, and accessible this interface supplies power and communicates for camera module 1 and infrared thermal imaging module 2, sets up and obtains real-time data of gathering to the camera parameter.
In addition, referring to fig. 4, the transformer area monitoring system of the present embodiment further includes an operation and maintenance interface, where the operation and maintenance interface includes a path of USB2.0, and is mainly used for system upgrade, derivation of generated learning data, derivation of system abnormal logs, and the like; and the gigabit network is used for daily operation and maintenance debugging, field camera debugging and alignment, can be connected with debugging equipment (such as a personal notebook computer), and supports online updating, rollback and system parameter configuration of the equipment.
In this embodiment, the deep learning operation module 42 analyzes various collected data information to obtain abnormal information, and sends an analysis result to the TTU (or other upper layer terminal) through the uplink communication module. In the prior art, all video streams are transmitted to a centralized system background for centralized analysis, the video streams generate huge communication flow, if 4G or other payment networks are used, the communication cost is greatly increased, however, the system of the embodiment only uploads the generated abnormal information, abnormal types and corresponding abnormal pictures, and the uploaded data volume is greatly reduced.
In summary, in the transformer area monitoring system based on machine vision and thermal imaging technology of this embodiment, through vision technology, thermal imaging technology, and area acceleration angle velocity technology, automatic detection and automatic prediction of multiple faults in a transformer area can be realized, meanwhile, according to big data and deep learning technology, historical data and data change trend characteristics can be continuously iterated to improve the accuracy of prediction and identification, the detection accuracy is superior to that of a traditional single threshold determination method, the system can be effectively combined with an existing TTU, no additional network configuration is required, the operational pressure of a central server is not increased, and millisecond-level abnormal fault detection can be realized. In addition, the transformer station area monitoring system uses distributed analysis and calculation, namely intelligent analysis is carried out on a single transformer station area monitoring system node, only the type of the generated abnormal event and the corresponding abnormal picture are transmitted to the TTU or a remote monitoring center for displaying and early warning, the transmitted data volume is far smaller than that of the prior art, the configuration requirement of a central server is low or the central algorithm server is not needed, and the data monitoring of other nodes is not influenced when a single node breaks down.
Example two
On the basis of the above embodiments, the present embodiment provides a transformer platform area monitoring method based on machine vision and thermal imaging technologies. Referring to fig. 6, fig. 6 is a flowchart of a transformer area monitoring system based on machine vision and thermal imaging technologies according to an embodiment of the present invention. The method of the embodiment comprises the following steps:
s1: and collecting video data of the detected target in the transformer area.
Specifically, a detected target (for example, a fuse, a transmission line, a transformer body, a distribution box, and the like) of a transformer area is monitored by using the camera module 1, an original video stream of the detected target is collected and encoded, and the original video stream is encoded into an H264/H265 video code stream.
S2: and acquiring thermal imaging picture data of the detected target in the transformer area.
Specifically, an infrared thermal imaging module is used for monitoring detected targets (such as fuses, transmission lines, transformer bodies, distribution boxes and the like) of the transformer area, an infrared thermal imaging technology is used for monitoring the temperature of electric equipment of the transformer area in real time, and thermal imaging picture data are generated finally.
S3: and acquiring the three-axis acceleration, the three-axis angular velocity and the azimuth information of the transformer area monitoring system.
Specifically, a DMP module is used for acquiring current triaxial acceleration, triaxial angular velocity and current azimuth information in real time; the sampled original data are transmitted to a core control module, the core control module can automatically learn the changes of acceleration, angular velocity and direction caused by the changes of vibration, micro displacement and posture generated by the impact of the transformer area and according to historical data and data change trend characteristics, and can judge and early warn the non-sudden abnormal state in advance, such as: other foreign matters of the vehicle impact the platform area, the tower of the platform area inclines, and whether the platform area is damaged by external force or not.
S4: and identifying the detected target by utilizing the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity and/or the azimuth information, acquiring abnormal information of the detected target, and deeply learning the abnormal information by utilizing a neural network so as to predict the non-sudden fault in advance.
In an embodiment of the present invention, the S4 includes:
s41: decoding or segmenting the video data, the thermal imaging picture data, the current triaxial acceleration, the current triaxial angular velocity, and the current orientation information to form a single frame data sequence;
s42: according to the single-frame data sequence, recognizing the data abnormal type by utilizing a trained neural network and outputting a frame data sequence with recognition information;
s43: and classifying the detected abnormal types according to the frame data sequence with the identification information.
Subsequently, the detected abnormal events are classified and distributed to corresponding service processing units, such as TTUs (which may be displayed or stored according to a specific abnormal type or directly transmitted to a remote monitoring platform). It should be noted that, in the Transformer area monitoring method of this embodiment, only the generated abnormal event type and the image corresponding to the abnormality are transmitted to a TTU (distribution Transformer supervisory Unit), and the amount of transmitted data is much smaller than that in the prior art; in addition, the transformer area monitoring method uses distributed analysis and calculation, intelligent analysis is carried out on a single transformer area monitoring system node, the result is transmitted to a TTU or a remote monitoring center for display and early warning, the central server configuration requirement is low or a central algorithm server is not needed, and the data monitoring of other nodes is not influenced when a single node fails.
Further, the S4 further includes:
and obtaining the temperature information of the area where the detected target is located according to the thermal imaging picture data, judging the temperature information, and generating temperature abnormal information of different levels according to the relationship between the temperature information and different thresholds.
Specifically, after acquiring thermal imaging picture data of a transformer platform area, the temperature of the whole area where the thermal imaging picture is located is judged, when the absolute temperature of a certain hot spot is higher than a preset value or the relative temperature of the certain hot spot (namely, the hot spot temperature is higher than the ambient temperature or the temperature of the area near the hot spot), different levels of temperature abnormal information (emergency defect, serious defect, general defect and normal) are generated according to different thresholds, rapid change is generated according to the temperature or abnormity is generated according to historical data and data change trend characteristics, and an original video stream, original thermal imaging picture data and abnormal types are sent to remote monitoring equipment.
In an embodiment of the present invention, the S4 further includes:
and continuously training the neural network by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
Specifically, the method of the embodiment includes: the collected normal data and abnormal information are continuously acquired, and according to historical data and data change trend characteristics, independent learning is performed, iteration and training are combined, so that the accuracy of the algorithm is continuously improved, and the non-sudden faults of the transformer area which are accumulated for a long time and gradually worsen can be predicted in advance. For example, the transformer pile head is oxidized or loosened due to external reasons, so that internal resistance is increased to heat a connecting point and finally cause burning, the process is not generated in a short time but gradually heated until the connecting point is burnt, non-sudden abnormity can be effectively and timely pre-warned, and then workers can perform pre-warning to reduce power failure loss caused by burning damage of a transformer area.
According to the transformer area monitoring system and method, automatic detection and automatic prediction of various faults of the transformer area can be achieved through a vision technology, a thermal imaging technology and an area acceleration angle speed technology, and meanwhile accuracy of prediction and recognition can be improved through continuous iteration according to a big data and deep learning technology.
In the embodiments provided in the present invention, it should be understood that the apparatus and method disclosed in the present invention can be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and in actual implementation, there may be other divisions, for example, multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not implemented.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware form, and can also be realized in a form of hardware and a software functional module.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (10)
1. A transformer platform district monitored control system based on machine vision and thermal imaging technique, its characterized in that includes:
the camera module (1) is used for acquiring video data of a detected target in a transformer area;
the infrared thermal imaging module (2) is used for acquiring thermal imaging picture data of a detected target in the transformer area;
the DMP module (3) is used for acquiring the three-axis acceleration, the three-axis angular velocity and the azimuth information of the position of the transformer area monitoring system; and
and the core control module (4) is used for identifying the detected target by utilizing the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity and/or the azimuth information, acquiring abnormal information of the detected target, and carrying out deep learning on the abnormal information by utilizing a neural network so as to predict non-sudden faults in advance.
2. The transformer platform area monitoring system based on machine vision and thermal imaging technology according to claim 1, wherein the camera module (1) comprises a CMOS image sensor and an infrared fill-in lamp, wherein the infrared fill-in lamp is used for filling in the monitored target when ambient light is lower than the minimum required imaging light of the CMOS image sensor.
3. The transformer station area monitoring system based on machine vision and thermal imaging technology according to claim 1, characterized in that the core control module (4) comprises:
a data processing module (41) for decoding or segmenting the video data, the thermal imaging picture data, the current triaxial acceleration, the current triaxial angular velocity and the current orientation information to form a single frame data sequence;
the deep learning operation module (42) is used for identifying the data abnormal type by utilizing the trained neural network according to the single-frame data sequence and outputting a frame data sequence with fault identification information;
an anomaly classification module (43) for classifying the detected type of the abnormal event according to the frame data sequence with the fault identification information,
the data processing module (41) is further configured to obtain temperature information of an area where the detected target is located according to the thermal imaging picture data, judge the temperature information, and generate temperature anomaly information of different levels according to the relationship between the temperature information and different thresholds.
4. The machine vision and thermal imaging technology based transformer bay monitoring system of claim 3, wherein the core control module (4) is further configured to:
and continuously training the neural network in the deep learning operation module (42) by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
5. The transformer platform monitoring system based on machine vision and thermal imaging technology according to claim 1, further comprising a power supply, and a power switch, a power indicator lamp and a status indicator lamp connected to the power supply, wherein the status indicator lamp is used for indicating the status of the detected object through different combinations of colored lamps and different flashing frequencies.
6. The transformer station area monitoring system based on machine vision and thermal imaging technology according to any one of claims 1 to 5, further comprising a communication module comprising an upstream communication unit and a downstream communication unit, wherein,
the uplink communication unit is used for electrically connecting the transformer area monitoring system to the TTU;
the downlink communication unit is used for electrically connecting the camera module and the infrared thermal imaging module to the core control module.
7. A transformer platform area monitoring method based on machine vision and thermal imaging technology is characterized by comprising the following steps:
s1: collecting video data of a detected target in a transformer area;
s2: acquiring thermal imaging picture data of a detected target in the transformer area;
s3: acquiring three-axis acceleration, three-axis angular velocity and azimuth information of the transformer area monitoring system;
s4: and identifying the detected target by utilizing the video data, the thermal imaging picture data, the three-axis acceleration, the three-axis angular velocity and/or the azimuth information, acquiring abnormal information of the detected target, and deeply learning the abnormal information by utilizing a neural network so as to predict the non-sudden fault in advance.
8. The transformer station area monitoring method based on machine vision and thermal imaging technology according to claim 7, wherein the S4 includes:
s41: decoding or segmenting the video data, the thermal imaging picture data, the current triaxial acceleration, the current triaxial angular velocity, and the current orientation information to form a single frame data sequence;
s42: according to the single-frame data sequence, recognizing the data abnormal type by utilizing a trained neural network and outputting a frame data sequence with recognition information;
s43: and classifying the types of the detected abnormal events according to the frame data sequence with the identification information.
9. The transformer station area monitoring method based on machine vision and thermal imaging technology according to claim 7, wherein the S4 further includes:
and obtaining the temperature information of the area where the detected target is located according to the thermal imaging picture data, judging the temperature information, and generating temperature abnormal information of different levels according to the relationship between the temperature information and different thresholds.
10. The transformer station area monitoring method based on machine vision and thermal imaging technology according to any one of claims 7 to 9, wherein the S4 further includes:
and continuously training the neural network by utilizing the collected or detected normal data and abnormal data as well as historical data and data change trend characteristics, and predicting the non-sudden fault of the transformer area in advance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011075079.6A CN112383137A (en) | 2020-10-09 | 2020-10-09 | Transformer area monitoring system and method based on machine vision and thermal imaging technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011075079.6A CN112383137A (en) | 2020-10-09 | 2020-10-09 | Transformer area monitoring system and method based on machine vision and thermal imaging technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112383137A true CN112383137A (en) | 2021-02-19 |
Family
ID=74581177
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011075079.6A Pending CN112383137A (en) | 2020-10-09 | 2020-10-09 | Transformer area monitoring system and method based on machine vision and thermal imaging technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112383137A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114228784A (en) * | 2021-12-31 | 2022-03-25 | 成都华动科技有限公司 | Locomotive vehicle electrical risk sensing system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108776273A (en) * | 2018-05-31 | 2018-11-09 | 广州供电局有限公司 | Distributing net platform region monitoring method, apparatus and system |
| CN109101906A (en) * | 2018-07-27 | 2018-12-28 | 中国南方电网有限责任公司超高压输电公司贵阳局 | A kind of converting station electric power equipment infrared image exception real-time detection method and device |
| CN111523660A (en) * | 2020-04-15 | 2020-08-11 | 南京清然能源科技有限公司 | Audio-visual-thermal integrated anomaly detection and alarm method based on artificial intelligence |
| EP3706267A1 (en) * | 2019-03-07 | 2020-09-09 | ABB Schweiz AG | Artificial intelligence monitoring system using infrared images to identify hotspots in a switchgear |
| CN112383698A (en) * | 2020-10-09 | 2021-02-19 | 西安星闪数创智能科技有限公司 | Transformer platform district terminal intelligent monitoring system |
-
2020
- 2020-10-09 CN CN202011075079.6A patent/CN112383137A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108776273A (en) * | 2018-05-31 | 2018-11-09 | 广州供电局有限公司 | Distributing net platform region monitoring method, apparatus and system |
| CN109101906A (en) * | 2018-07-27 | 2018-12-28 | 中国南方电网有限责任公司超高压输电公司贵阳局 | A kind of converting station electric power equipment infrared image exception real-time detection method and device |
| EP3706267A1 (en) * | 2019-03-07 | 2020-09-09 | ABB Schweiz AG | Artificial intelligence monitoring system using infrared images to identify hotspots in a switchgear |
| CN111523660A (en) * | 2020-04-15 | 2020-08-11 | 南京清然能源科技有限公司 | Audio-visual-thermal integrated anomaly detection and alarm method based on artificial intelligence |
| CN112383698A (en) * | 2020-10-09 | 2021-02-19 | 西安星闪数创智能科技有限公司 | Transformer platform district terminal intelligent monitoring system |
Non-Patent Citations (2)
| Title |
|---|
| 高兆丽 等: "变电站全站设备故障预警系统的研制", 《电气自动化》 * |
| 黄国良: "电力设备防盗智能系统技术研究与应用", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114228784A (en) * | 2021-12-31 | 2022-03-25 | 成都华动科技有限公司 | Locomotive vehicle electrical risk sensing system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110707823B (en) | Multi-dimensional intelligent remote inspection system for transformer substation | |
| US20090315722A1 (en) | Multi-wavelength video image fire detecting system | |
| CN112288984A (en) | Three-dimensional visual unattended substation intelligent linkage system based on video fusion | |
| CN105607617A (en) | Security fault diagnosis system and method based on Internet of Things | |
| CN111190065A (en) | Non-contact on-line monitoring system for power equipment | |
| CN103152601A (en) | Intelligent failure-reporting camera and network management client system thereof | |
| CN103927838A (en) | Smoke thermal imaging fire automatic positioning detection system and method | |
| CN110032093A (en) | A kind of engineering Electrical Safety management method of view-based access control model Internet of Things | |
| CN110941228B (en) | Intelligent railway traction substation management and control system | |
| CN208334991U (en) | Distribution room monitoring system | |
| CN110853287A (en) | Flame real-time monitoring system and method based on Internet of things distributed architecture | |
| CN116963357B (en) | Intelligent configuration control method, system and medium for lamp | |
| CN112634579A (en) | Transformer substation security early warning linkage system | |
| CN115297305A (en) | Multi-video-stream security risk early warning system and method based on edge calculation | |
| CN112383137A (en) | Transformer area monitoring system and method based on machine vision and thermal imaging technology | |
| CN112153344B (en) | Power distribution room equipment state online intelligent monitoring system and method based on embedded GPU platform and deep learning | |
| KR102648377B1 (en) | Intelligent integrated wiring board with network cable quality monitoring | |
| CN115664006B (en) | Intelligent management and control integrated platform for incremental power distribution network | |
| CN113095160A (en) | Power system personnel safety behavior identification method and system based on artificial intelligence and 5G | |
| CN108120475A (en) | The data monitoring method and device of a kind of piping lane | |
| CN208521417U (en) | A kind of electric fire disaster warning monitoring device | |
| KR102503532B1 (en) | Switchboard monitoring amn management system having image photographing device | |
| KR102361372B1 (en) | System for monitoring to network in smartcity coverage employing IoT sensor module | |
| CN115393340A (en) | AI vision product quality detection system based on 5G algorithm | |
| CN115980062A (en) | Industrial production line whole-process vision inspection method based on 5G |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210219 |
|
| RJ01 | Rejection of invention patent application after publication |