CN105014196A - Underwater weld seam tracking device based on neural network recognition - Google Patents

Underwater weld seam tracking device based on neural network recognition Download PDF

Info

Publication number
CN105014196A
CN105014196A CN201510471484.2A CN201510471484A CN105014196A CN 105014196 A CN105014196 A CN 105014196A CN 201510471484 A CN201510471484 A CN 201510471484A CN 105014196 A CN105014196 A CN 105014196A
Authority
CN
China
Prior art keywords
underwater
weld
subset
human agent
welding
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
Application number
CN201510471484.2A
Other languages
Chinese (zh)
Inventor
梁彦云
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to CN201510471484.2A priority Critical patent/CN105014196A/en
Publication of CN105014196A publication Critical patent/CN105014196A/en
Pending legal-status Critical Current

Links

Classifications

    • BPERFORMING OPERATIONS; TRANSPORTING
    • B23MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
    • B23KSOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
    • B23K9/00Arc welding or cutting
    • B23K9/0061Underwater arc welding
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B23MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
    • B23KSOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
    • B23K9/00Arc welding or cutting
    • B23K9/12Automatic feeding or moving of electrodes or work for spot or seam welding or cutting
    • B23K9/127Means for tracking lines during arc welding or cutting
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B23MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
    • B23KSOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
    • B23K9/00Arc welding or cutting
    • B23K9/24Features related to electrodes
    • B23K9/28Supporting devices for electrodes
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B23MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
    • B23KSOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
    • B23K9/00Arc welding or cutting
    • B23K9/32Accessories

Abstract

The invention relates to an underwater weld seam tracking device based on neural network recognition. The underwater weld seam tracking device comprises an underwater robot body, a feature extraction device and a weld seam type recognition device. The feature extraction device and the weld seam type recognition device are located on the underwater robot body and used for extracting the features of underwater weld seams and determining the types of the weld seams respectively, and when it is determined that the types of the weld seams are non-welding types, related alarm information is sent out. By means of the underwater weld seam tracking device, the positions of the underwater weld seams can be accurately determined, and convenience can be brought to underwater welding.

Description

Based on the underwater weld tracker of neural network recognization
Technical field
The present invention relates to field of image recognition, particularly relate to a kind of underwater weld tracker based on neural network recognization.
Background technology
Along with expanding economy and scientific and technological progress, there is multiple underwater installation to need to carry out Underwater Welding, first welding will search out position while welding, which dictates that the accuracy of welding and the quality of welding, for the weld seam that cannot weld, need and alarm, simultaneously, when performing Underwater Welding, due to this special working environment under water, very high to the requirement of welding personnel.
The general location adopting the mode of welding personnel range estimation to carry out weld seam in prior art, but artificial visually examine's mode performs difficulty under water, and tack weld is unsatisfactory, meanwhile, welding personnel cannot perform sustainable welding under water.
For this reason, need the technical scheme that a kind of execution underwater weld of mechanization is located, adopt the mode of underwater robot welding to substitute Underwater Welding personnel welding, and can automatically provide position while welding accurately, to facilitate underwater robot to perform welding operation according to weld seam, improve welding quality.
Summary of the invention
In order to solve the technical problem that prior art exists, the invention provides a kind of underwater weld tracker based on neural network recognization, the Preprocessing Technique butt welded seam type of neural network recognization technology and multiple filter and position while welding is utilized to identify, transform the concrete structure of underwater welding equipment and the concrete structure of transformation underwater robot simultaneously, make it be suitable for Underwater Welding.
According to an aspect of the present invention, provide a kind of underwater weld tracker based on neural network recognization, described tracker comprises underwater human agent, feature extracting device and welding type identification equipment, described feature extracting device and described welding type identification equipment are all positioned on described underwater human agent, for extracting underwater weld characteristic sum determination welding type respectively, and determining that welding type is cannot send relative alarm information during welds types.
More specifically, also comprise: electrode holders based in the underwater weld tracker of neural network recognization described, for fixing welding rod, described welding rod is wet method coated electrode, and material is mild steel, safety switch, its cathode conductor is connected to described electrode holders, earth clip, is fixed on workpiece to be welded, electric welding machine, negative pole is connected to described electrode holders, plus earth, electrode holders driving arrangement, is connected with described electrode holders, for driving described electrode holders to go to the position while welding of workpiece to be welded according to electrode holders drive singal, electrode holders drive motors is a direct current generator, for the driving of described electrode holders driving arrangement to described electrode holders provides power, described underwater human agent comprises driving arrangement and robotic's framework, described driving arrangement comprises driving governor, larboard DC speed-reducing, starboard DC speed-reducing, sink-float motor, right-handed screw oar, counterpropeller, subsidiary screw and shaft coupling, described larboard DC speed-reducing, described starboard DC speed-reducing and described sink-float motor are respectively by described shaft coupling and described right-handed screw oar, described counterpropeller is connected with described subsidiary screw, described driving governor receives the drive control signal that described treatment facility waterborne sends, with larboard DC speed-reducing according to the content driven of described drive control signal and described starboard DC speed-reducing to control described right-handed screw oar and described counterpropeller respectively, underwater human agent is driven to realize advancing, retreat and left and right action, also for the motor that rises and falls according to the content driven of described drive control signal to control described subsidiary screw, underwater human agent is driven to realize rising and dive action, underwater camera equipment, be arranged on described underwater human agent, comprise hemispherical watertight transparent cover, floor light subset and CMOS camera, described hemispherical watertight transparent cover is for holding described floor light subset and described CMOS camera, described floor light subset provides floor light for the underwater photograph technical of described CMOS camera, described CMOS camera to objects ahead shooting to obtain the underwater picture comprising objects ahead, pre-processing device, is arranged on described underwater human agent, is connected with described CMOS camera, comprises medium filtering subset, LPF subset and homomorphic filtering subset, described medium filtering subset is connected with described CMOS camera, for performing medium filtering to described underwater picture, with the spot noise in underwater picture described in filtering, obtains the first filtering image, described LPF subset is connected with described medium filtering subset, for removing the random noise in described first filtering image, obtains the second filtering image, described homomorphic filtering subset is connected with described LPF subset, for performing image enhaucament, to obtain enhancing underwater picture to described second filtering image, feature extracting device, be arranged on described underwater human agent, be connected with described pre-processing device, comprise Iamge Segmentation subset and eigenvector recognition subset, the weld seam target in described enhancing underwater picture identifies to obtain Underwater weld seam image based on weld image gray threshold scope by described Iamge Segmentation subset, described eigenvector recognition subset is connected with described Iamge Segmentation subset, 8 geometric properties based on described Underwater weld seam image determination underwater weld target: Euler's hole count, circularity, angle point number, convex-concave degree, smoothness, draw ratio, tight ness rating and main shaft angle, and by described 8 geometric properties composition characteristics vector, welding type identification equipment, be arranged on described underwater human agent, be connected with described feature extracting device, adopt single hidden layer BP neutral net that 8 inputs 2 export, using 8 of underwater weld target geometric properties as input layer, output layer is underwater weld type, and described underwater weld type comprises normal welding type and cannot welds types, buoy waterborne, is arranged on the top water surface of described underwater human agent, power supply unit, be arranged on described buoy waterborne, comprise waterproof sealing cover, solar powered device, battery, change-over switch and electric pressure converter, described waterproof sealing cover is for holding described solar powered device, described battery, described change-over switch and described electric pressure converter, described change-over switch is connected respectively with described solar powered device and described battery, dump energy according to battery determines whether be switched to described solar powered device to be powered by described solar powered device, described electric pressure converter is connected with described change-over switch, with the 5V voltage transitions will inputted by change-over switch for 3.3V voltage, the Big Dipper positioning equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for receiving the Big Dipper locator data that the Big Dipper satellite sends, sonar detection equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for detecting the relative distance of described submersible main body to described buoy waterborne, and to adjust the distance output as first-phase, ultrasonic ranging equipment, is arranged on described underwater human agent, for measuring the distance of described underwater human agent distance front weld seam target, and to adjust the distance output as second-phase, single-chip microcomputer, be arranged on described underwater human agent, with described welding type identification equipment, described the Big Dipper positioning equipment, described sonar detection equipment is connected respectively with described ultrasonic ranging equipment, based on described the Big Dipper locator data, described first-phase is adjusted the distance and described second-phase is adjusted the distance calculates weld and HAZ data, and when receive described underwater weld type for cannot welds types time, send alarm signal, when to receive described underwater weld type be normal welding type, according to Underwater weld seam image at the relative position determination electrode holders drive singal strengthening underwater picture.
More specifically, described based in the underwater weld tracker of neural network recognization, described tracker also comprises: submerged cable, for being connected with soldering center waterborne by described single-chip microcomputer, described alarm signal and described weld and HAZ data is sent to described soldering center waterborne.
More specifically, described based in the underwater weld tracker of neural network recognization: described single-chip microcomputer is AT89C51.
More specifically, described based in the underwater weld tracker of neural network recognization: described robotic's framework comprises support, left pressure gram transparent tube, right pressure gram transparent tube, connecting hoop, storage platform, mechanical arm, manipulator and water proof sealing drum.
Accompanying drawing explanation
Below with reference to accompanying drawing, embodiment of the present invention are described, wherein:
Fig. 1 is the block diagram of the underwater weld tracker based on neural network recognization illustrated according to an embodiment of the present invention.
Reference numeral: 1 underwater human agent; 2 underwater weld identification equipments; 3 electrode holders driving arrangements
Detailed description of the invention
Below with reference to accompanying drawings the embodiment of the underwater weld tracker based on neural network recognization of the present invention is described in detail.
Underwater Welding is the necessary operation of fixing underwater installation, its welding quality quality directly determines the firmness of underwater installation, but, the intelligent equipment that can adapt to underwater operation environment, automatically can complete weld and HAZ, welding operation is still there is not in prior art, for the welding type that cannot weld, also cannot carry out and alarm.
In order to overcome above-mentioned deficiency, the present invention has built a kind of underwater weld tracker based on neural network recognization, with the artificial operating platform of underwater, introduce the various fixation and recognition equipment that can adapt to underwater operation environment, thus underwater weld type can be realized and carry out underwater installation welding adaptively.
Fig. 1 is the block diagram of the underwater weld tracker based on neural network recognization illustrated according to an embodiment of the present invention, described tracker comprises underwater human agent, feature extracting device and welding type identification equipment, described feature extracting device and described welding type identification equipment are all positioned on described underwater human agent, for extracting underwater weld characteristic sum determination welding type respectively, and determining that welding type is cannot send relative alarm information during welds types.
Then, continue to be further detailed the concrete structure of the underwater weld tracker based on neural network recognization of the present invention.
Described tracker also comprises: electrode holders, and for fixing welding rod, described welding rod is wet method coated electrode, and material is mild steel; Safety switch, its cathode conductor is connected to described electrode holders; Earth clip, is fixed on workpiece to be welded; Electric welding machine, negative pole is connected to described electrode holders, plus earth; Electrode holders driving arrangement, is connected with described electrode holders, for driving described electrode holders to go to the position while welding of workpiece to be welded according to electrode holders drive singal; Electrode holders drive motors is a direct current generator, for the driving of described electrode holders driving arrangement to described electrode holders provides power.
Described underwater human agent comprises driving arrangement and robotic's framework, described driving arrangement comprises driving governor, larboard DC speed-reducing, starboard DC speed-reducing, sink-float motor, right-handed screw oar, counterpropeller, subsidiary screw and shaft coupling, described larboard DC speed-reducing, described starboard DC speed-reducing and described sink-float motor are respectively by described shaft coupling and described right-handed screw oar, described counterpropeller is connected with described subsidiary screw, described driving governor receives the drive control signal that described treatment facility waterborne sends, with larboard DC speed-reducing according to the content driven of described drive control signal and described starboard DC speed-reducing to control described right-handed screw oar and described counterpropeller respectively, underwater human agent is driven to realize advancing, retreat and left and right action, also for the motor that rises and falls according to the content driven of described drive control signal to control described subsidiary screw, underwater human agent is driven to realize rising and dive action.
Described tracker also comprises: underwater camera equipment, be arranged on described underwater human agent, comprise hemispherical watertight transparent cover, floor light subset and CMOS camera, described hemispherical watertight transparent cover is for holding described floor light subset and described CMOS camera, described floor light subset provides floor light for the underwater photograph technical of described CMOS camera, described CMOS camera to objects ahead shooting to obtain the underwater picture comprising objects ahead.
Described tracker also comprises: pre-processing device, is arranged on described underwater human agent, is connected with described CMOS camera, comprises medium filtering subset, LPF subset and homomorphic filtering subset; Described medium filtering subset is connected with described CMOS camera, for performing medium filtering to described underwater picture, with the spot noise in underwater picture described in filtering, obtains the first filtering image; Described LPF subset is connected with described medium filtering subset, for removing the random noise in described first filtering image, obtains the second filtering image; Described homomorphic filtering subset is connected with described LPF subset, for performing image enhaucament, to obtain enhancing underwater picture to described second filtering image.
Described tracker also comprises: feature extracting device, be arranged on described underwater human agent, be connected with described pre-processing device, comprise Iamge Segmentation subset and eigenvector recognition subset, the weld seam target in described enhancing underwater picture identifies to obtain Underwater weld seam image based on weld image gray threshold scope by described Iamge Segmentation subset; Described eigenvector recognition subset is connected with described Iamge Segmentation subset, 8 geometric properties based on described Underwater weld seam image determination underwater weld target: Euler's hole count, circularity, angle point number, convex-concave degree, smoothness, draw ratio, tight ness rating and main shaft angle, and by described 8 geometric properties composition characteristics vector.
Described tracker also comprises: welding type identification equipment, be arranged on described underwater human agent, be connected with described feature extracting device, adopt single hidden layer BP neutral net that 8 inputs 2 export, using 8 of underwater weld target geometric properties as input layer, output layer is underwater weld type, and described underwater weld type comprises normal welding type and cannot welds types.
Described tracker also comprises: buoy waterborne, is arranged on the top water surface of described underwater human agent.
Described tracker also comprises: power supply unit, be arranged on described buoy waterborne, comprise waterproof sealing cover, solar powered device, battery, change-over switch and electric pressure converter, described waterproof sealing cover is for holding described solar powered device, described battery, described change-over switch and described electric pressure converter, described change-over switch is connected respectively with described solar powered device and described battery, dump energy according to battery determines whether be switched to described solar powered device to be powered by described solar powered device, described electric pressure converter is connected with described change-over switch, with the 5V voltage transitions will inputted by change-over switch for 3.3V voltage.
Described tracker also comprises: the Big Dipper positioning equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for receiving the Big Dipper locator data that the Big Dipper satellite sends; Sonar detection equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for detecting the relative distance of described submersible main body to described buoy waterborne, and to adjust the distance output as first-phase.
Described tracker also comprises: ultrasonic ranging equipment, is arranged on described underwater human agent, for measuring the distance of described underwater human agent distance front weld seam target, and to adjust the distance output as second-phase.
Described tracker also comprises: single-chip microcomputer, be arranged on described underwater human agent, with described welding type identification equipment, described the Big Dipper positioning equipment, described sonar detection equipment is connected respectively with described ultrasonic ranging equipment, based on described the Big Dipper locator data, described first-phase is adjusted the distance and described second-phase is adjusted the distance calculates weld and HAZ data, and when receive described underwater weld type for cannot welds types time, send alarm signal, when to receive described underwater weld type be normal welding type, according to Underwater weld seam image at the relative position determination electrode holders drive singal strengthening underwater picture.
Alternatively, described based in the underwater weld tracker of neural network recognization, described tracker also comprises: submerged cable, for being connected with soldering center waterborne by described single-chip microcomputer, described alarm signal and described weld and HAZ data is sent to described soldering center waterborne; Described single-chip microcomputer is AT89C51; Described robotic's framework comprises support, left pressure gram transparent tube, right pressure gram transparent tube, connecting hoop, storage platform, mechanical arm, manipulator and water proof sealing drum.
Alternatively, medium filtering subset, LPF subset and homomorphic filtering subset adopt different fpga chips to realize respectively.
In addition, FPGA (Field-Programmable Gate Array), i.e. field programmable gate array, he is the product further developed on the basis of the programming devices such as PAL, GAL, CPLD.He occurs as a kind of semi-custom circuit in special IC (ASIC) field, has both solved the deficiency of custom circuit, overcomes again the shortcoming that original programming device gate circuit number is limited.
With the circuit design that hardware description language (Verilog or VHDL) completes, can through simple comprehensive and layout, being burned onto fast on FPGA and testing, is the technology main flow of modern IC designs checking.These can be edited element and can be used to realize some basic logic gates (such as AND, OR, XOR, NOT) or more more complex combination function such as decoder or mathematical equation.Inside most FPGA, in these editable elements, also comprise memory cell such as trigger (Flip-flop) or other more complete block of memory.System designer can be coupled together the logical block of FPGA inside by editable connection as required, just looks like that a breadboard has been placed in a chip.One dispatch from the factory after the logical block of finished product FPGA can change according to designer with being connected, so FPGA can complete required logic function.
FPGA is in general slow than the speed of ASIC (special IC), realizes same function ratio ASIC circuit area and wants large.But they also have a lot of advantages such as can finished product fast, can be modified the mistake in correction program and more cheap cost.Manufacturer also may provide the FPGA of cheap still edit capability difference.Because these chips have poor can edit capability, so exploitations of these designs complete on common FPGA, then design is transferred to one and is similar on the chip of ASIC.Another method is with CPLD (Complex Programmable LogicDevice, CPLD).The exploitation of FPGA has a great difference relative to the exploitation of conventional P C, single-chip microcomputer.FPGA, based on concurrent operation, realizes with hardware description language; Very large difference is had compared to the operation in tandem of PC or single-chip microcomputer (no matter being von Neumann structure or Harvard structure).
As far back as 1980 mid-nineties 90s, FPGA takes root in PLD equipment.CPLD and FPGA includes the Programmadle logic unit of some relatively large amount.The density of CPLD gate is between several thousand to several ten thousand logical blocks, and FPGA normally arrives millions of several ten thousand.The main distinction of CPLD and FPGA is their system architecture.CPLD is a somewhat restrictive structure.This structure is arranged by the logical groups of one or more editable result sum and forms with the register of the locking of some relatively small amounts.Such result lacks editor's flexibility, but but have the time delay and logical block that can estimate to the advantage of linkage unit height ratio.And FPGA has a lot of linkage units, although allow him edit more flexibly like this, structure is complicated many.
Adopt the underwater weld tracker based on neural network recognization of the present invention, for the technical problem of human weld's difficulty in prior art, by transforming the concrete structure of underwater robot and underwater welding equipment, improve the automaticity of welding, simultaneously with the image pre-processor filtering of multiple filter all kinds of interference under water, with neural network recognization technology identification underwater weld type, with underwater position fixing technique identification weld seam particular location.
Be understandable that, although the present invention with preferred embodiment disclose as above, but above-described embodiment and be not used to limit the present invention.For any those of ordinary skill in the art, do not departing under technical solution of the present invention ambit, the technology contents of above-mentioned announcement all can be utilized to make many possible variations and modification to technical solution of the present invention, or be revised as the Equivalent embodiments of equivalent variations.Therefore, every content not departing from technical solution of the present invention, according to technical spirit of the present invention to any simple modification made for any of the above embodiments, equivalent variations and modification, all still belongs in the scope of technical solution of the present invention protection.

Claims (5)

1. the underwater weld tracker based on neural network recognization, described tracker comprises underwater human agent, feature extracting device and welding type identification equipment, described feature extracting device and described welding type identification equipment are all positioned on described underwater human agent, for extracting underwater weld characteristic sum determination welding type respectively, and determining that welding type is cannot send relative alarm information during welds types.
2., as claimed in claim 1 based on the underwater weld tracker of neural network recognization, it is characterized in that, described tracker also comprises:
Electrode holders, for fixing welding rod, described welding rod is wet method coated electrode, and material is mild steel;
Safety switch, its cathode conductor is connected to described electrode holders;
Earth clip, is fixed on workpiece to be welded;
Electric welding machine, negative pole is connected to described electrode holders, plus earth;
Electrode holders driving arrangement, is connected with described electrode holders, for driving described electrode holders to go to the position while welding of workpiece to be welded according to electrode holders drive singal;
Electrode holders drive motors is a direct current generator, for the driving of described electrode holders driving arrangement to described electrode holders provides power;
Described underwater human agent comprises driving arrangement and robotic's framework, described driving arrangement comprises driving governor, larboard DC speed-reducing, starboard DC speed-reducing, sink-float motor, right-handed screw oar, counterpropeller, subsidiary screw and shaft coupling, described larboard DC speed-reducing, described starboard DC speed-reducing and described sink-float motor are respectively by described shaft coupling and described right-handed screw oar, described counterpropeller is connected with described subsidiary screw, described driving governor receives the drive control signal that described treatment facility waterborne sends, with larboard DC speed-reducing according to the content driven of described drive control signal and described starboard DC speed-reducing to control described right-handed screw oar and described counterpropeller respectively, underwater human agent is driven to realize advancing, retreat and left and right action, also for the motor that rises and falls according to the content driven of described drive control signal to control described subsidiary screw, underwater human agent is driven to realize rising and dive action,
Underwater camera equipment, be arranged on described underwater human agent, comprise hemispherical watertight transparent cover, floor light subset and CMOS camera, described hemispherical watertight transparent cover is for holding described floor light subset and described CMOS camera, described floor light subset provides floor light for the underwater photograph technical of described CMOS camera, described CMOS camera to objects ahead shooting to obtain the underwater picture comprising objects ahead;
Pre-processing device, is arranged on described underwater human agent, is connected with described CMOS camera, comprises medium filtering subset, LPF subset and homomorphic filtering subset; Described medium filtering subset is connected with described CMOS camera, for performing medium filtering to described underwater picture, with the spot noise in underwater picture described in filtering, obtains the first filtering image; Described LPF subset is connected with described medium filtering subset, for removing the random noise in described first filtering image, obtains the second filtering image; Described homomorphic filtering subset is connected with described LPF subset, for performing image enhaucament, to obtain enhancing underwater picture to described second filtering image;
Feature extracting device, be arranged on described underwater human agent, be connected with described pre-processing device, comprise Iamge Segmentation subset and eigenvector recognition subset, the weld seam target in described enhancing underwater picture identifies to obtain Underwater weld seam image based on weld image gray threshold scope by described Iamge Segmentation subset; Described eigenvector recognition subset is connected with described Iamge Segmentation subset, 8 geometric properties based on described Underwater weld seam image determination underwater weld target: Euler's hole count, circularity, angle point number, convex-concave degree, smoothness, draw ratio, tight ness rating and main shaft angle, and by described 8 geometric properties composition characteristics vector;
Welding type identification equipment, be arranged on described underwater human agent, be connected with described feature extracting device, adopt single hidden layer BP neutral net that 8 inputs 2 export, using 8 of underwater weld target geometric properties as input layer, output layer is underwater weld type, and described underwater weld type comprises normal welding type and cannot welds types;
Buoy waterborne, is arranged on the top water surface of described underwater human agent;
Power supply unit, be arranged on described buoy waterborne, comprise waterproof sealing cover, solar powered device, battery, change-over switch and electric pressure converter, described waterproof sealing cover is for holding described solar powered device, described battery, described change-over switch and described electric pressure converter, described change-over switch is connected respectively with described solar powered device and described battery, dump energy according to battery determines whether be switched to described solar powered device to be powered by described solar powered device, described electric pressure converter is connected with described change-over switch, with the 5V voltage transitions will inputted by change-over switch for 3.3V voltage,
The Big Dipper positioning equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for receiving the Big Dipper locator data that the Big Dipper satellite sends;
Sonar detection equipment, is arranged on described buoy waterborne, provides supply of electric power by described power supply unit, for detecting the relative distance of described submersible main body to described buoy waterborne, and to adjust the distance output as first-phase;
Ultrasonic ranging equipment, is arranged on described underwater human agent, for measuring the distance of described underwater human agent distance front weld seam target, and to adjust the distance output as second-phase;
Single-chip microcomputer, be arranged on described underwater human agent, with described welding type identification equipment, described the Big Dipper positioning equipment, described sonar detection equipment is connected respectively with described ultrasonic ranging equipment, based on described the Big Dipper locator data, described first-phase is adjusted the distance and described second-phase is adjusted the distance calculates weld and HAZ data, and when receive described underwater weld type for cannot welds types time, send alarm signal, when to receive described underwater weld type be normal welding type, according to Underwater weld seam image at the relative position determination electrode holders drive singal strengthening underwater picture.
3., as claimed in claim 2 based on the underwater weld tracker of neural network recognization, it is characterized in that, described tracker also comprises:
Submerged cable, for being connected with soldering center waterborne by described single-chip microcomputer, is sent to described soldering center waterborne by described alarm signal and described weld and HAZ data.
4., as claimed in claim 2 based on the underwater weld tracker of neural network recognization, it is characterized in that:
Described single-chip microcomputer is AT89C51.
5., as claimed in claim 2 based on the underwater weld tracker of neural network recognization, it is characterized in that:
Described robotic's framework comprises support, left pressure gram transparent tube, right pressure gram transparent tube, connecting hoop, storage platform, mechanical arm, manipulator and water proof sealing drum.
CN201510471484.2A 2015-08-04 2015-08-04 Underwater weld seam tracking device based on neural network recognition Pending CN105014196A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510471484.2A CN105014196A (en) 2015-08-04 2015-08-04 Underwater weld seam tracking device based on neural network recognition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510471484.2A CN105014196A (en) 2015-08-04 2015-08-04 Underwater weld seam tracking device based on neural network recognition

Publications (1)

Publication Number Publication Date
CN105014196A true CN105014196A (en) 2015-11-04

Family

ID=54404692

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510471484.2A Pending CN105014196A (en) 2015-08-04 2015-08-04 Underwater weld seam tracking device based on neural network recognition

Country Status (1)

Country Link
CN (1) CN105014196A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106182038A (en) * 2016-09-08 2016-12-07 肇庆市小凡人科技有限公司 One automatic welding machine people under water
CN108732970A (en) * 2018-05-28 2018-11-02 深圳臻迪信息技术有限公司 Paddle robot self-rescue method and the robot that paddles with self-rescue function

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100902142B1 (en) * 2007-12-12 2009-06-10 에스티엑스조선주식회사 The welds seam tracking device of welding robot and the control method
KR20100035501A (en) * 2008-09-26 2010-04-05 삼성중공업 주식회사 Seam tracking robot
CN103586565A (en) * 2013-11-04 2014-02-19 哈尔滨工业大学(威海) Underwater wet welding tracking system based on active vision
CN104588833A (en) * 2015-01-24 2015-05-06 无锡桑尼安科技有限公司 Underwater welding robot based on image identification
CN104625320A (en) * 2015-01-24 2015-05-20 无锡桑尼安科技有限公司 Underwater welding method
CN204397131U (en) * 2015-01-24 2015-06-17 无锡桑尼安科技有限公司 Based on the Underwater Welding robot of image recognition

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100902142B1 (en) * 2007-12-12 2009-06-10 에스티엑스조선주식회사 The welds seam tracking device of welding robot and the control method
KR20100035501A (en) * 2008-09-26 2010-04-05 삼성중공업 주식회사 Seam tracking robot
CN103586565A (en) * 2013-11-04 2014-02-19 哈尔滨工业大学(威海) Underwater wet welding tracking system based on active vision
CN104588833A (en) * 2015-01-24 2015-05-06 无锡桑尼安科技有限公司 Underwater welding robot based on image identification
CN104625320A (en) * 2015-01-24 2015-05-20 无锡桑尼安科技有限公司 Underwater welding method
CN204397131U (en) * 2015-01-24 2015-06-17 无锡桑尼安科技有限公司 Based on the Underwater Welding robot of image recognition

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
庄志坚: "基于BP神经网络方法的焊缝图像识别系统", 《科技创新导报》 *
胡震: "基于自主作业的AUV视觉系统", 《机器人》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106182038A (en) * 2016-09-08 2016-12-07 肇庆市小凡人科技有限公司 One automatic welding machine people under water
CN108732970A (en) * 2018-05-28 2018-11-02 深圳臻迪信息技术有限公司 Paddle robot self-rescue method and the robot that paddles with self-rescue function

Similar Documents

Publication Publication Date Title
CN104985290A (en) Neural-network-identification-based underwater weld joint tracking method
CN105129052A (en) Underwater human body searching method
CN104967833A (en) Fishing method based on data communication
CN104999164A (en) Underwater robot based on multiple filtering processing
CN105136808A (en) Underwater robot underwater equipment damage detection platform
CN105014196A (en) Underwater weld seam tracking device based on neural network recognition
CN105160638A (en) Image processing based underwater welding line tracking system
CN104999200B (en) Diver portable system
CN105123636A (en) Fish body identifying system based on laser image detection
Joshi et al. Underwater exploration and mapping
Shi et al. An underwater pipeline tracking system for amphibious spherical robots
CN105010276A (en) Automatic fish catching method
CN105096325A (en) Underwater equipment detection system based on laser image
CN105665877A (en) Method for performing underwater automatic welding by using underwater robot
CN105109646A (en) Underwater human body detecting method based on neural network identification
CN104991259A (en) Laser imaging detection method
CN105059506A (en) Intelligent underwater robot based on laser imaging detection
CN105058430A (en) Underwater robot operation method based on image processing
CN105138007B (en) Underwater robot based on image procossing avoids platform
CN104999201B (en) A kind of method using underwater robot to carry out automatic welding under water
CN105173036A (en) Underwater human body detector based on neural network identification
CN105728975A (en) Method for performing underwater automatic welding by using underwater robot
CN105035282A (en) Submerged reef detection system based on laser imaging
CN105160640A (en) Method for implementing scattered light elimination and welding line identification on underwater image
CN105057848A (en) Underwater welding joint search platform capable of improving image processing accuracy

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20151104