CN113184017A - ATP video monitoring and intelligent recognition analysis system - Google Patents
ATP video monitoring and intelligent recognition analysis system Download PDFInfo
- Publication number
- CN113184017A CN113184017A CN202110409928.5A CN202110409928A CN113184017A CN 113184017 A CN113184017 A CN 113184017A CN 202110409928 A CN202110409928 A CN 202110409928A CN 113184017 A CN113184017 A CN 113184017A
- Authority
- CN
- China
- Prior art keywords
- state
- vehicle
- atp
- image
- video
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 51
- 238000004458 analytical method Methods 0.000 title claims abstract description 23
- 238000003745 diagnosis Methods 0.000 claims abstract description 20
- 238000005516 engineering process Methods 0.000 claims abstract description 10
- 230000005540 biological transmission Effects 0.000 claims abstract description 8
- 238000013473 artificial intelligence Methods 0.000 claims abstract description 4
- 238000001514 detection method Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 18
- 238000013135 deep learning Methods 0.000 claims description 16
- 238000010801 machine learning Methods 0.000 claims description 13
- 238000000034 method Methods 0.000 claims description 13
- 238000001816 cooling Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 238000013139 quantization Methods 0.000 claims description 4
- 229910000838 Al alloy Inorganic materials 0.000 claims description 3
- 238000010191 image analysis Methods 0.000 claims description 3
- 238000003780 insertion Methods 0.000 claims description 3
- 230000037431 insertion Effects 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims description 2
- 238000009833 condensation Methods 0.000 claims description 2
- 230000005494 condensation Effects 0.000 claims description 2
- 230000001502 supplementing effect Effects 0.000 claims 1
- 230000005856 abnormality Effects 0.000 abstract description 2
- 238000013527 convolutional neural network Methods 0.000 description 8
- 238000012423 maintenance Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000011897 real-time detection Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L15/00—Indicators provided on the vehicle or train for signalling purposes
- B61L15/0081—On-board diagnosis or maintenance
-
- 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
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The invention discloses an ATP video monitoring and intelligent recognition analysis system, which comprises a vehicle-mounted video acquisition device and an expert intelligent diagnosis system, the vehicle-mounted video acquisition device is connected with a vehicle-mounted server through a network cable, the vehicle-mounted server is connected with an expert intelligent diagnosis system through the network cable, the expert intelligent diagnosis system is connected with a central server through a transmission platform, the video information of each board card of the CTCS2-200H type ATP equipment is collected and gathered through a multi-channel parallel acquisition technology, the on-off and stroboscopic changes of the indicator light, the reading of the dialer and the position of the shifting fork are monitored and can be called at any time in a section, the vehicle-mounted server actively sends an alarm to the server in the section after finding the abnormality of the equipment through an artificial intelligence recognition algorithm, and the real-time working state of the vehicle-mounted ATP equipment is restored in the server software in the section, so that the real-time monitoring and management of the vehicle-mounted ATP equipment faults are realized.
Description
Technical Field
The invention relates to an ATP video monitoring and intelligent recognition analysis system.
Background
The train overspeed protection system, called vehicle-mounted ATP for short, is the core safety equipment of a train operation control system, and once a fault occurs, an electric service section cannot carry out online fault removal or make targeted maintenance preparation because the equipment is not provided with a real-time fault alarm function and is installed in a closed space of a train. The real-time monitoring of the working state of the equipment is realized, the online running state of the electric control equipment can be known at the first time in the electric service section, and after a fault occurs, the fault type can be judged at the first time so as to conveniently scientifically decide a coping mode and reduce the consequence caused by the fault to the minimum.
At present, maintenance personnel mainly judge the fault state of the vehicle-mounted ATP cabinet through the on-off and flashing of signal lamps in each control unit and carry out maintenance operation according to the description of a use and maintenance specification. The device is expected to collect and restore the signal lamp condition of the equipment, and carry out fault analysis by applying a video analysis technology to realize online monitoring.
CN201911008812.X discloses an ATP host front panel video monitoring system, which comprises a micro camera matrix, a vehicle-mounted video host, a central server and a client terminal, wherein in the running process of a motor train unit, the system adopts the technologies of video shooting, splicing, compression and the like to shoot the whole condition (including the conditions of on and off of various lamp positions, the state of equipment and the like) of the front panel of the vehicle-mounted ATP host in real time, intelligently identifies the flashing and abnormal conditions of the lamp positions according to ATP of different types, transmits the flashing and abnormal conditions to a ground server through a mobile network in real time, and displays the information such as videos, identification results, fault alarm and the like on the terminal.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides an ATP video monitoring and intelligent recognition analysis system which has the advantages of multi-channel parallel acquisition, on-off and stroboscopic change of a monitoring indicator lamp, reading of a dialer, monitoring of a shifting fork position, random calling, artificial intelligent recognition algorithm, active alarm sending, real-time monitoring and management of vehicle-mounted ATP equipment and accurate recognition result.
The purpose of the invention is realized by the following technical scheme.
An ATP video monitoring and intelligent identification analysis system comprises a vehicle-mounted video acquisition device and an expert intelligent diagnosis system, wherein the vehicle-mounted video acquisition device is connected with a vehicle-mounted server through a network cable, the vehicle-mounted server is connected with the expert intelligent diagnosis system through the network cable, the expert intelligent diagnosis system is connected with a central server through a transmission platform, the expert intelligent diagnosis system comprises an image acquisition module, the image acquisition module is connected with an indicator lamp, a switch, a dialer and a row fork through cables, the indicator lamp, the switch, the dialer and the row fork are connected with an image processing module through cables, the image processing module is connected with an indicator lamp state, a switch state, a dialer wheel diameter value and a row fork state through cables, the indicator lamp state, the switch state, the dialer wheel diameter value and the row fork state are connected with collected monitoring information through cables and compared, and the collected monitoring information is compared and then is connected with a normal running state through a cable to be displayed in real time and give a fault alarm.
Furthermore, the vehicle-mounted video acquisition device adopts a plurality of paths of cameras to acquire all target states of ATP (automatic train protection) such as video information of a monitored object, outputs a standard video format to the vehicle-mounted server and provides a real-time image for the intelligent analysis system; the vehicle-mounted server is used for processing video information acquisition transmitted to the vehicle-mounted server through a network cable, carrying out image analysis on the vehicle-mounted server, judging fault conditions and converting a judgment result into a digital signal; the expert intelligent diagnosis system is connected with a railway internal transmission platform through a network cable, and data are transmitted to the central server through the platform; the central server is responsible for signal receiving, state restoring work, presenting results, and carrying out digital processing, sorting and storing on fault information; the image processing module acquires an image of a scene through the image acquisition module, stores the image as an independent two-dimensional signal array, quantizes the image by adopting 256 quantization levels, wherein the value range of each pixel is 0-255, performs color characterization by adopting an RGB color model, expresses any color by using red, green and blue color components, each pixel point comprises the saturation of the pixel point under the three color components, and can acquire the color information of the image through the three color components of RGB; the image acquisition module acquires an image of the state of the indicator lamp by using the camera, acquires the current color of the indicator lamp and the instrument state corresponding to the color through an artificial intelligence cloud image processing technology, thereby realizing the instant monitoring of equipment, and no light source is arranged in the cabinet.
Furthermore, the image processing module is connected with the machine learning deep learning module through a cable, and the machine learning deep learning module is connected with the state of the indicator light, the state of the switch, the wheel diameter value of the code dialing device and the state of the line fork through the cable, so that the current state of the instrument can be rapidly converted into the information of the state of the instrument through the processing method after the image is acquired.
Further, the real-time display of the normal operation state comprises the following contents: a. the states of all the indicator lights on the panel are 187, including on-off and stroboscopic, wherein the number of the indicator lights is VC144, VC 244, DRU45, RLU16, STN 18 and BTM 20; b. the row forks CN3, CN4, CN5, CN6, CN7 and CN8 are provided with plug insertion slots when the system is powered on; VC1, VC2, DRU module dialer setting value reading; d, whether the two dial wheel diameter value codes of the VC1 module are consistent or not and whether the two dial wheel diameter values of the VC2 module are consistent or not.
Further, the fault alarm condition includes: a. the state of the indicator light changes, and the module has a fault; b. when the system is electrified, the slot position has no corresponding row fork; c. displaying all the dialer readings; and d, the two dialers of VC1 and VC2 show that the wheel diameter value codes are inconsistent.
Furthermore, the central server sends a calling request to the vehicle-mounted server, the vehicle-mounted server responds and transmits real-time information, and the real-time working state of the ATP cabinet is displayed on an interface through the central server.
Further, the technical index requirements of the vehicle-mounted system are as follows: the power of the vehicle-mounted video acquisition device is 150W, the power of the vehicle-mounted server is not more than 300W, the total power is not more than 450W, the voltage is DC 110V, the temperature of a working environment ranges from minus 25 ℃ to plus 70 ℃, the protection level is IP20, active cooling is not needed, the atmospheric pressure is that the altitude does not exceed 2500m, the impact vibration GB/T25119-: the voltage is AC 220V, the power is not more than 1250W, the working environment temperature is 5-35 ℃, active cooling is needed, the altitude of atmospheric pressure is not more than 2500m, the relative humidity is not more than 90%, and the non-condensation is realized.
Further, on-vehicle video acquisition device includes 1 group of aluminum alloy frame, camera module and quick detach list 18 groups, 1 set of video decoding system, 1 set of light filling system.
Further, an ATP device state image recognition SDK is arranged in the expert intelligent diagnosis system, the ATP device state image recognition SDK includes functions of device panel detection area positioning, indicator light state recognition, dialer digital recognition, and aviation plug connection state recognition, and the ATP device state image recognition SDK implementation steps include image preprocessing: the image frame collected by the camera is subjected to image preprocessing to obtain an image with higher definition; detection area marking: according to the serial number of the camera, automatically selecting a monitoring area marking strategy, finishing marking of a detection area and extracting a detection area image; area detection object identification: the state of the indicating lamp, the number of the dialing device and the connection state of the aviation plug are mainly identified, and finally, the identification result is collected and output.
Furthermore, the machine learning deep learning module performs convolution characteristic training on manually marked picture data based on a fault detection and identification method of a deep learning network fast R-CNN to obtain a detection identification model for three devices, namely an indicator light, a wheel diameter code dialing device and an aviation plug, in the scene, sample data is derived from an acquired ATP device state monitoring video, frame images are intercepted at intervals, 6000 images related to states of the ATP device indicator light, the wheel diameter code dialing device and the aviation plug are respectively selected in a manual mode and serve as a training sample set; by configuring the operating environment of the Tensorflow platform, a Faster R-CNN algorithm is built on the platform, data labeling and storage are carried out on the screened sample set according to the standard of a PASCAL VOC 2007 data set, and parameter fine adjustment is carried out on the basis of a VGG16 pre-training model, so that the final model is suitable for fault detection and identification of ATP equipment state monitoring.
Compared with the prior art, the invention has the advantages that: the ATP video monitoring and intelligent recognition analysis system collects and gathers video information of each board card of CTCS2-200H type ATP equipment through a multichannel parallel acquisition technology, monitors the on-off and stroboscopic change of an indicator light, the reading of a dialer and the shifting position, can be called at any time in a section, and a vehicle-mounted server actively sends an alarm to the server in the section after discovering the equipment is abnormal through an artificial intelligent recognition algorithm and restores the real-time working state of the vehicle-mounted ATP equipment in the server software in the section, so that the real-time monitoring and management of the fault of the vehicle-mounted ATP equipment in a region by a road bureau and an electric service section are realized.
The present invention has: the accuracy is as follows: acquiring and labeling a large amount of image data sets, training by adopting a deep learning fast R-CNN network method to obtain a high-precision model, and realizing accurate equipment state identification;
real-time performance: optimizing the training process to enable the model to achieve real-time detection performance, and realizing real-time detection and identification of the monitoring video;
robustness: the images under the complex environment are used as a training set, so that the method can adapt to the equipment state recognition under various complex environments;
convenience: the scheme of local storage and cloud backup is adopted, the video material can be stored for 30 days by local cache, and whether the video material is synchronized to the cloud end or not can be selected for calling at any time. When a fault occurs, videos or images before and after the fault occurs can be recorded on a rear interface, so that electric service maintenance personnel can analyze the reason of the fault in time.
Drawings
Fig. 1 is a schematic diagram of the overall system structure of the present invention.
Fig. 2 is a schematic flow chart of the system function implementation of the present invention.
FIG. 3 is a diagram of the status image recognition SDK distribution of the ATP device of the present invention.
FIG. 4 is a flow chart of the ATP device status image recognition SDK of the present invention.
FIG. 5 is a flowchart of a learning deep learning module according to the present invention.
Detailed Description
The invention is described in detail below with reference to the drawings and specific examples.
As shown in FIG. 1-2, the ATP video monitoring and intelligent recognition analysis system comprises a vehicle-mounted video acquisition device, a vehicle-mounted server, an expert intelligent diagnosis system, a central server, an image acquisition device, an indicator light, a switch, a dialer, a line fork, an image processing device, a machine learning deep learning device, an indicator light state, a switch state, a dialer wheel diameter value, a line fork state, a summary monitoring information and comparison, a normal state real-time display and a fault alarm, wherein the vehicle-mounted video acquisition device is connected with the vehicle-mounted server through a network cable, the vehicle-mounted server is connected with the expert intelligent diagnosis system through the network cable, the expert intelligent diagnosis system is connected with the central server through a transmission platform, the image acquisition device is connected with the indicator light, the switch, the dialer and the line fork through cables, the indicator light, the switch, the dialer and the line fork are connected with the image processing device through the cables, the image processing device is connected with the machine learning deep learning through the cables, the machine learning deep learning collects monitoring information and compares the monitoring information through cable connection, collects the monitoring information and compares the monitoring information through cable connection normal state real-time display and fault alarm, the vehicle-mounted video collecting device adopts a multi-path camera to collect all target states of ATP such as video information of a monitored object, outputs a standard video format to a vehicle-mounted server to provide real-time images for an intelligent analysis system, the vehicle-mounted server is responsible for processing video information collection transmitted to the vehicle-mounted server through a network cable, performs image analysis on the vehicle-mounted server to judge fault conditions and convert the judgment result into digital signals, and the expert intelligent diagnosis system is connected with a railway internal transmission platform through the network cable, the data are transmitted to a central server through a platform, the central server is responsible for signal receiving and state restoration work, results are presented, fault information is digitally processed, sorted and stored, images of scenes are acquired through image acquisition equipment in the image processing, and the images are stored into an independent two-dimensional signal array. Then, quantizing an image by adopting 256 quantization levels, wherein the value range of each pixel is 0-255, then performing color representation by adopting an RGB color model, expressing any color by using red, green and blue color components, each pixel point comprises the saturation of the pixel point under the three color components, the color information of the image can be obtained through the three color components of RGB, the image is obtained by utilizing a camera to obtain the image of the state of an indicator light, the current color of the indicator light and the instrument state corresponding to the color are obtained through an artificial intelligent cloud image processing technology, so that the equipment is monitored immediately, no light source is arranged in a cabinet, for equipment without the indicator light, the method for collecting the image is a synchronously controlled LED flash lamp, the image collection is performed when the flash lamp is on, in order to not interfere with the image collection of other cameras around, the LED is specially designed, the machine learning deep learning method can be used for learning the state of the equipment by a machine learning method to obtain the states of the equipment corresponding to different pictures, an image processing module and a machine learning program are added in image acquisition, so that the current state of the instrument can be quickly converted into the information of the state of the instrument by the processing method after the image is acquired, and the normal running state can be displayed in real time and comprises the following contents: a. the states of all the indicator lights on the panel are 187, including on-off and stroboscopic, wherein the number of the indicator lights is VC144, VC 244, DRU45, RLU16, STN 18 and BTM 20; b. the row forks CN3, CN4, CN5, CN6, CN7 and CN8 are provided with plug insertion slots when the system is powered on; VC1, VC2, DRU module dialer setting value reading; whether two dial wheel diameter value codes of the VC1 module are consistent or not and whether two dial wheel diameter values of the VC2 module are consistent or not, wherein the fault alarm condition comprises the following steps: a. the state of the indicator light changes, and the module has a fault; b. when the system is electrified, the slot position has no corresponding row fork; c. displaying all the dialer readings; the two dialing devices of VC1 and VC2 display that the wheel diameter value codes are not consistent, the central server sends a calling request to the vehicle-mounted server, the vehicle-mounted server responds and transmits real-time information, and displays the real-time working state of the ATP cabinet on an interface through the central server, and the technical index requirements of the vehicle-mounted system are as follows: the power of the vehicle-mounted video acquisition device is 150W, the power of the vehicle-mounted server is not more than 300W, the total power is not more than 450W, the voltage is DC 110V (+ 25%/-30%), the temperature of a working environment is-25 to +70 ℃, the protection level is IP20, no active cooling is needed, the atmospheric pressure is the altitude of no more than 2500m, the impact vibration GB/T25119-: the voltage is AC 220V, the power is not more than 1250W, the temperature of the working environment is 5-35 ℃, active cooling is needed, the altitude of the atmospheric pressure is not more than 2500m, the relative humidity is not more than 90 percent (25 ℃), the vehicle-mounted video acquisition device is not condensed and comprises the following parts: 1 group of aluminum alloy frame, camera module and quick detach list 18 group, 1 set of video decoding system, 1 set of light filling system, in order to ensure that device intensity satisfies the mobile unit requirement, high definition video in can shooting in the vibration environment in order to satisfy camera system simultaneously, the vibration factor has fully been considered to the device to confirm through the repeated vibration test.
As shown in fig. 3, the expert intelligent diagnosis system is provided with an ATP device status image recognition SDK, where the ATP device status image recognition SDK includes functions of device panel detection area positioning, indicator light status recognition, dialer number recognition, and aviation plug connection status recognition, and as shown in fig. 4, the ATP device status image recognition SDK implementation step includes image preprocessing: the image frame collected by the camera is subjected to image preprocessing to obtain an image with higher definition; detection area marking: according to the serial number of the camera, automatically selecting a monitoring area marking strategy, finishing marking of a detection area and extracting a detection area image; area detection object identification: the state of the indicating lamp, the number of the dialing device and the connection state of the aviation plug are mainly identified, and finally, the identification result is collected and output.
SDK specification packet size: 110M
Digital image identification size: subcode stream 620px 480px (main stream 1920px 1080px)
Detection area positioning speed: 80 to 130ms (620px 480px)
Detection area positioning accuracy: 97.68% (10W frame video image, inaccurate positioning or positioning failure rate 2.32%)
Speed of state recognition of indicator lights: 1 to 5ms (620px 480px)
And indicating lamp state identification precision: 99.0% (192 lamps, except 2 lamp failure, can be identified accurately)
Number of dials identification speed: 30 to 60ms (620px 480px)
Digital recognition model accuracy: 93.6% (based on MNIST hand-written digital data set, 6W training images, 1W testing images)
Aviation plug connection state recognition speed: 20 to 60ms (620px 480px)
Aviation plug connection state identification precision: 91.2% (video image test result)
Remarking: the indexes are operated on real equipment according to the latest version of the SDK and are measured by a real data set, and the algorithm performance is influenced by the conditions of operating equipment, a test data set and the like.
Aiming at the complex environment scenes of narrow, dark and dispersed targets in a vehicle-mounted ATP equipment cabinet, the machine learning deep learning module performs convolution characteristic training on manually marked picture data based on a fault detection and identification method of a deep learning network Faster R-CNN to obtain a detection and identification model for three devices of an indicator light, a wheel diameter code dialing device and an aviation plug under the scene, wherein sample data is derived from an acquired ATP equipment state monitoring video, frame images are intercepted at intervals, 6000 images related to states of an ATP equipment indicator light, a wheel diameter code dialing device and an aviation plug are respectively selected in a manual mode and are used as a training sample set; by configuring the operating environment of the Tensorflow platform, a Faster R-CNN algorithm is built on the platform, data labeling and storage are carried out on the screened sample set according to the standard of a PASCAL VOC 2007 data set, and parameter fine adjustment is carried out on the basis of a VGG16 pre-training model, so that the final model is suitable for fault detection and identification of ATP equipment state monitoring.
During the model training process, the characteristics of the pictures are extracted by adopting a 16-layer VGG16 convolutional neural network, and a mode of cross training of an RPN network and a Fast R-CNN network is used. The convolution characteristics are shared during training, so that the parameter quantity during training is reduced, and the training efficiency is improved, so that the model achieves real-time detection performance.
Practical application shows that the Faster R-CNN algorithm can achieve ideal effect on fault detection of the operation state of the ATP equipment under different shooting angles, shelters and different illumination conditions, and can achieve the rate of real-time monitoring.
The working principle is as follows: the invention relates to an ATP video monitoring and intelligent recognition analysis system, which acquires images of scenes through a vehicle-mounted video acquisition device (image acquisition equipment), stores the images as an independent two-dimensional signal array, quantizes the images by adopting 256 quantization levels, the value range of each pixel is 0-255, performs color representation by adopting an RGB color model, expresses any color by using red, green and blue color components, each pixel point comprises the saturation of the pixel point under the three color components, can acquire the color information of the images through the three color components of RGB, acquires the images of the states of indicator lights by using a camera, acquires the current color of the indicator lights and the instrument states corresponding to the color through an artificial intelligent cloud image processing technology, thereby realizing the instant monitoring of the equipment, and can realize the machine learning method for the equipment without the indicator lights, the equipment state is learned to obtain the states of the equipment corresponding to different pictures, no light source is arranged in the cabinet, for equipment without an indicator light, the method for acquiring the image is the LED flash lamp which is synchronously controlled, and the image acquisition is carried out when the flash lamp is on. In order not to interfere with image acquisition of other cameras around, the LED is specially designed to just illuminate the area where the air switch is located, and the main process comprises the following steps: the system is connected with a railway internal transmission platform through a network cable, data are transmitted to a central server through the platform, the central server is responsible for signal receiving, state recovery work and result presentation, the central server sends a calling request to the vehicle-mounted server, the vehicle-mounted server responds and transmits real-time information, the real-time working state of an ATP cabinet is displayed on an interface through the central server, the central server carries out digital processing on the fault information, the fault information is sorted and stored and an alarm is sent out in time, the real-time image is uploaded, and maintenance personnel intervene; the main key realization technology is as follows: the method comprises the steps that video information of each board card of CTCS2-200H type ATP equipment is collected and gathered through a multichannel parallel acquisition technology, and the on-off and stroboscopic change of an indicator light, the reading of a dialer and the shifting position are monitored and can be called at any time in a section; the vehicle-mounted server actively sends an alarm to the in-segment server after discovering the equipment abnormality through an artificial intelligence recognition algorithm, and restores the real-time working state of the vehicle-mounted ATP equipment in the in-segment server software, so that the real-time monitoring and management of the road bureau and the electric service segment on the faults of the vehicle-mounted ATP equipment in the area are realized.
Claims (10)
1. The utility model provides an ATP video monitoring and intelligent recognition analytic system which characterized in that: comprises a vehicle-mounted video acquisition device and an expert intelligent diagnosis system, wherein the vehicle-mounted video acquisition device is connected with a vehicle-mounted server through a network cable, the vehicle-mounted server is connected with an expert intelligent diagnosis system through a network cable, the expert intelligent diagnosis system is connected with a central server through a transmission platform, the expert intelligent diagnosis system comprises an image acquisition module, the image acquisition module is connected with an indicator light, a switch, a dialer and a line fork through cables, the indicator light, the switch, the code dialing device and the line fork are connected with the image processing module through cables, the image processing module is connected with the state of the indicator light, the state of the switch, the wheel diameter value of the code dialing device and the state of the line fork through cables, the indicator light state, the switch state, the dial indicator wheel diameter value and the row fork state are connected through cables to collect monitoring information and compared, and the collected monitoring information is compared and then is connected with a normal running state through a cable to be displayed in real time and give a fault alarm.
2. The ATP video monitoring and intelligent recognition analysis system according to claim 1, wherein the vehicle-mounted video acquisition device adopts multiple cameras to acquire all ATP target states such as video information of a monitored object, outputs a standard video format to a vehicle-mounted server, and provides a real-time image for the intelligent analysis system;
the vehicle-mounted server is used for processing video information acquisition transmitted to the vehicle-mounted server through a network cable, carrying out image analysis on the vehicle-mounted server, judging fault conditions and converting a judgment result into a digital signal;
the expert intelligent diagnosis system is connected with a railway internal transmission platform through a network cable, and data are transmitted to the central server through the platform; the central server is responsible for signal receiving, state restoring work, presenting results, and carrying out digital processing, sorting and storing on fault information;
the image processing module acquires an image of a scene through the image acquisition module, stores the image as an independent two-dimensional signal array, quantizes the image by adopting 256 quantization levels, wherein the value range of each pixel is 0-255, performs color characterization by adopting an RGB color model, expresses any color by using red, green and blue color components, each pixel point comprises the saturation of the pixel point under the three color components, and can acquire the color information of the image through the three color components of RGB;
the image acquisition module acquires an image of the state of the indicator lamp by using the camera, acquires the current color of the indicator lamp and the instrument state corresponding to the color through an artificial intelligence cloud image processing technology, thereby realizing the instant monitoring of equipment, and no light source is arranged in the cabinet.
3. The ATP video monitoring and intelligent recognition analysis system of claim 1 or 2, wherein: the image processing module is connected with the machine learning deep learning module through a cable, and the machine learning deep learning module is connected with the state of the indicator light, the state of the switch, the wheel diameter value of the code dialing device and the state of the line fork through the cable, so that the current state of the instrument can be rapidly converted into the information of the state of the instrument through the processing method after the image is acquired.
4. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the real-time display of the normal operation state comprises the following contents: a. the states of all the indicator lights on the panel are 187, including on-off and stroboscopic, wherein the number of the indicator lights is VC144, VC 244, DRU45, RLU16, STN 18 and BTM 20; b. the row forks CN3, CN4, CN5, CN6, CN7 and CN8 are provided with plug insertion slots when the system is powered on; VC1, VC2, DRU module dialer setting value reading; d, whether the two dial wheel diameter value codes of the VC1 module are consistent or not and whether the two dial wheel diameter values of the VC2 module are consistent or not.
5. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the fault alarm condition includes: a. the state of the indicator light changes, and the module has a fault; b. when the system is electrified, the slot position has no corresponding row fork; c. displaying all the dialer readings; and d, the two dialers of VC1 and VC2 show that the wheel diameter value codes are inconsistent.
6. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the central server sends a calling request to the vehicle-mounted server, the vehicle-mounted server responds and transmits real-time information, and the real-time working state of the ATP cabinet is displayed on an interface through the central server.
7. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the technical index requirements of the vehicle-mounted system are as follows: the power of the vehicle-mounted video acquisition device is 150W, the power of the vehicle-mounted server is not more than 300W, the total power is not more than 450W, the voltage is DC 110V, the temperature of a working environment ranges from minus 25 ℃ to plus 70 ℃, the protection level is IP20, active cooling is not needed, the atmospheric pressure is that the altitude does not exceed 2500m, the impact vibration GB/T25119-: the voltage is AC 220V, the power is not more than 1250W, the working environment temperature is 5-35 ℃, active cooling is needed, the altitude of atmospheric pressure is not more than 2500m, the relative humidity is not more than 90%, and the non-condensation is realized.
8. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the vehicle-mounted video acquisition device comprises 1 aluminum alloy frame group, 18 camera modules and quick-release units, 1 video decoding system and 1 light supplementing system.
9. The ATP video monitoring and intelligent recognition analysis system of claim 1, wherein: the expert intelligent diagnosis system is provided with an ATP equipment state image recognition SDK, the ATP equipment state image recognition SDK comprises functions of equipment panel detection area positioning, indicator lamp state recognition, dialer digital recognition and aviation plug connection state recognition, and the ATP equipment state image recognition SDK implementation steps comprise image preprocessing: the image frame collected by the camera is subjected to image preprocessing to obtain an image with higher definition; detection area marking: according to the serial number of the camera, automatically selecting a monitoring area marking strategy, finishing marking of a detection area and extracting a detection area image; area detection object identification: the state of the indicating lamp, the number of the dialing device and the connection state of the aviation plug are mainly identified, and finally, the identification result is collected and output.
10. The ATP video monitoring and intelligent recognition analysis system of claim 3, wherein: the machine learning deep learning module is based on a fault detection and identification method of a deep learning network fast R-CNN, a detection identification model for three devices, namely an indicator light, a wheel diameter dialer and an aviation plug, in the scene is obtained by performing convolution characteristic training on manually marked picture data, sample data is derived from an acquired ATP device state monitoring video, frame images are intercepted at intervals, 6000 images related to states of the ATP device indicator light, the wheel diameter dialer and the aviation plug are respectively selected in a manual mode and are used as a training sample set; by configuring the operating environment of the Tensorflow platform, a Faster R-CNN algorithm is built on the platform, data labeling and storage are carried out on the screened sample set according to the standard of a PASCAL VOC 2007 data set, and parameter fine adjustment is carried out on the basis of a VGG16 pre-training model, so that the final model is suitable for fault detection and identification of ATP equipment state monitoring.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110409928.5A CN113184017A (en) | 2021-04-16 | 2021-04-16 | ATP video monitoring and intelligent recognition analysis system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110409928.5A CN113184017A (en) | 2021-04-16 | 2021-04-16 | ATP video monitoring and intelligent recognition analysis system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113184017A true CN113184017A (en) | 2021-07-30 |
Family
ID=76977033
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110409928.5A Pending CN113184017A (en) | 2021-04-16 | 2021-04-16 | ATP video monitoring and intelligent recognition analysis system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113184017A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114979562A (en) * | 2022-04-08 | 2022-08-30 | 大连海事大学 | Transformer substation indicator lamp state monitoring system and method based on image recognition |
| CN115331155A (en) * | 2022-10-14 | 2022-11-11 | 智慧齐鲁(山东)大数据科技有限公司 | Mass video monitoring point location graph state detection method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104333736A (en) * | 2014-10-28 | 2015-02-04 | 山东大学 | Intelligent recognition monitoring system and method for unmanned substation |
| CN107730017A (en) * | 2017-11-29 | 2018-02-23 | 马鞍山市科泰电气科技有限公司 | Intelligent metallurgical equipment operational diagnostics system |
| CN107786850A (en) * | 2017-11-02 | 2018-03-09 | 上海达实联欣科技发展有限公司 | A kind of master control room intelligent monitor system and method |
| CN109784156A (en) * | 2018-12-10 | 2019-05-21 | 国网天津市电力公司客户服务中心 | A kind of full Intelligent unattended business hall safety monitoring system based on artificial intelligence |
| CN110626389A (en) * | 2019-10-23 | 2019-12-31 | 河南蓝信科技有限责任公司 | ATP host front panel video monitoring system |
| CN111815912A (en) * | 2020-06-17 | 2020-10-23 | 珠海格力电器股份有限公司 | Equipment state detection method, system, device and readable storage medium |
| CN112287816A (en) * | 2020-10-28 | 2021-01-29 | 西安交通大学 | Dangerous working area accident automatic detection and alarm method based on deep learning |
-
2021
- 2021-04-16 CN CN202110409928.5A patent/CN113184017A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104333736A (en) * | 2014-10-28 | 2015-02-04 | 山东大学 | Intelligent recognition monitoring system and method for unmanned substation |
| CN107786850A (en) * | 2017-11-02 | 2018-03-09 | 上海达实联欣科技发展有限公司 | A kind of master control room intelligent monitor system and method |
| CN107730017A (en) * | 2017-11-29 | 2018-02-23 | 马鞍山市科泰电气科技有限公司 | Intelligent metallurgical equipment operational diagnostics system |
| CN109784156A (en) * | 2018-12-10 | 2019-05-21 | 国网天津市电力公司客户服务中心 | A kind of full Intelligent unattended business hall safety monitoring system based on artificial intelligence |
| CN110626389A (en) * | 2019-10-23 | 2019-12-31 | 河南蓝信科技有限责任公司 | ATP host front panel video monitoring system |
| CN111815912A (en) * | 2020-06-17 | 2020-10-23 | 珠海格力电器股份有限公司 | Equipment state detection method, system, device and readable storage medium |
| CN112287816A (en) * | 2020-10-28 | 2021-01-29 | 西安交通大学 | Dangerous working area accident automatic detection and alarm method based on deep learning |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114979562A (en) * | 2022-04-08 | 2022-08-30 | 大连海事大学 | Transformer substation indicator lamp state monitoring system and method based on image recognition |
| CN115331155A (en) * | 2022-10-14 | 2022-11-11 | 智慧齐鲁(山东)大数据科技有限公司 | Mass video monitoring point location graph state detection method and system |
| CN115331155B (en) * | 2022-10-14 | 2023-02-03 | 智慧齐鲁(山东)大数据科技有限公司 | Mass video monitoring point location graph state detection method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113184017A (en) | ATP video monitoring and intelligent recognition analysis system | |
| CN112396999B (en) | Abnormal display block detection method, display screen fault detection method and equipment thereof | |
| CN111832398B (en) | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method | |
| CN105242232B (en) | Outdoor electrical energy meter fault self-verifying method | |
| CN109784138B (en) | Image datamation inspection method for unmanned transformer substation | |
| CN109326125B (en) | Picture quality diagnosis system and method based on embedded system | |
| CN110008937B (en) | Switch cabinet running state management monitoring system and method and computing equipment | |
| CN115290696A (en) | Infrared thermal imaging defect detection method and device for transformer substation insulator | |
| CN114627388B (en) | Foreign matter detection equipment and foreign matter detection method for power transmission line | |
| CN111948501A (en) | Automatic inspection equipment for power grid operation | |
| CN110626389A (en) | ATP host front panel video monitoring system | |
| CN115936672A (en) | Smart power grid online safety operation and maintenance management method and system | |
| CN114894791A (en) | Rice growth analysis system and method based on matching factors | |
| CN112966552B (en) | Routine inspection method and system based on intelligent identification | |
| CN210822296U (en) | ATP host front panel video monitoring system | |
| CN109104591B (en) | Image acquisition device for power equipment | |
| CN112697050A (en) | Night side slope displacement monitoring system based on luminous body | |
| CN113065540A (en) | Power transmission line element failure identification method and system based on machine patrol image | |
| CN111950512A (en) | Substation circuit breaker indicator lamp on-off state identification method | |
| CN114167245B (en) | Intelligent detection method for partial discharge on surface of power transmission and transformation equipment and unmanned aerial vehicle fusion ultraviolet system | |
| CN112985823B (en) | Testing device for simulating camera sensor | |
| CN115733957A (en) | Navigation equipment alarm processing method based on image recognition | |
| CN209929831U (en) | Switch cabinet with image recognition position | |
| CN113762185A (en) | Intelligent system for achieving state acquisition of power equipment based on computer vision | |
| CN112383137A (en) | Transformer area monitoring system and method based on machine vision and thermal imaging technology |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 230000 office building of No. 039 electric service section, Fengyang Road, Yaohai District, Hefei City, Anhui Province Applicant after: Hefei Electric Service Section of China Railway Shanghai Bureau Group Co.,Ltd. Applicant after: Jiangsu ningtie Technology Co.,Ltd. Address before: 230000 office building of No. 039 electric service section, Fengyang Road, Yaohai District, Hefei City, Anhui Province Applicant before: Hefei Electric Service Section of China Railway Shanghai Bureau Group Co.,Ltd. Applicant before: Nanjing ningtie NDT Technology Research Institute Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210730 |