CN114331973B - Steel structure information extraction method suitable for oil-gas module manufacturing process - Google Patents
Steel structure information extraction method suitable for oil-gas module manufacturing process Download PDFInfo
- Publication number
- CN114331973B CN114331973B CN202111495100.2A CN202111495100A CN114331973B CN 114331973 B CN114331973 B CN 114331973B CN 202111495100 A CN202111495100 A CN 202111495100A CN 114331973 B CN114331973 B CN 114331973B
- Authority
- CN
- China
- Prior art keywords
- bounding box
- point
- steel structure
- image
- production
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 68
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 49
- 239000010959 steel Substances 0.000 title claims abstract description 49
- 238000000605 extraction Methods 0.000 title claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 15
- 230000000007 visual effect Effects 0.000 claims abstract description 9
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 2
- 238000003466 welding Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a steel structure information extraction method suitable for an oil-gas module manufacturing process, which comprises the following steps: the method comprises the steps of marking steel structure production serial numbers by laser, storing the steel structure production serial numbers in a database, shooting the steel structure surface production serial numbers by a CCD visual camera, leading a filtering algorithm into python to conduct noise reduction treatment on pictures, conducting image alignment treatment in python, conducting training of an image production serial number prediction model by using an MLP neural network algorithm, identifying on-site steel structure production serial number information, sending identification information to a server through a network, and displaying the information on a display screen. Compared with the traditional method, the method can effectively solve the problems of complicated information searching work, less information acquisition amount, poor information integrity and the like in the manufacturing process of the oil-gas module, and is efficient, reliable, high in universality and high in adaptability.
Description
Technical Field
The invention relates to an information extraction method, in particular to a steel structure information extraction method suitable for an oil-gas module manufacturing process.
Background
The oil-gas module is formed by welding various steel structures, the types of the steel structures are complex and numerous, and the accurate acquisition and correspondence of the information of each steel structure are very difficult. In many manufacturing fields, bar codes, RFID (radio frequency identification) marks and other means are adopted to facilitate information inquiry, but when an oil gas module is manufactured, common tags cannot be suitable for field operation environments due to extreme climates such as strong wind flow, welding high temperature and the like. In the current manufacturing process of the oil-gas module, the on-site engineer still adopts a paper drawing mode to search the information of the steel structure, the searching efficiency is low, the searching information is limited, the actual engineering requirements cannot be completely met, and an efficient, stable, convenient and easy information extraction method is needed in the manufacturing process of the oil-gas module.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide the steel structure information extraction method which can improve the extraction speed of basic information of a steel structure, improve the information accuracy of various parts in field operation, ensure the smooth proceeding of the manufacturing process of an oil-gas module, is suitable for a complex operation environment, has strong universality and simple searching mode and is suitable for the manufacturing process of the oil-gas module.
The invention discloses a steel structure information extraction method suitable for an oil-gas module manufacturing process, which comprises the following steps of:
When various steel structures of the oil-gas module are built, unique production serial numbers are engraved on the surfaces of the steel structures by utilizing lasers, and the production serial numbers and various basic information of the steel structures are stored in a database at a server side in a key value pair mode, wherein the production serial numbers are keys, and the various basic information are values;
Secondly, shooting a production sequence number on the surface of the steel structure by using a CCD visual camera, enabling the lens plane of the CCD visual camera to be parallel to the surface of the steel structure during shooting, and transmitting the shot picture to a computer through an industrial Ethernet;
Step three, opening a photo shot by a CCD visual camera in a python programming environment, and sequentially importing a median filtering algorithm and a wiener filtering algorithm to perform noise reduction treatment on the photo to obtain a noise reduction treatment image;
Step four, aligning the noise reduction image in a computer, wherein the specific process is as follows:
Firstly, introducing numpy modules into Python, performing discrete point coordinate extraction operation on the noise reduction processed image obtained in the second step, and obtaining a point A 1 with the largest abscissa, a point A 2 with the smallest abscissa, a point B 1 with the largest ordinate and a point B 2 with the smallest ordinate in all discrete points, creating a first bounding box and a second bounding box according to four points, wherein the two bounding boxes are rectangular, and establishing a linear equation of four sides of each bounding box; wherein the first and second sides of the first bounding box are parallel to line a 1A2, and the first side passes through point B 1, the second side passes through point B 2, the third side passes through point a 1, and the fourth side passes through point a 2; the first edge and the second edge of the second enclosure are parallel to the straight line B 1B2, and the first edge passes through the point A 1, the second edge passes through the point A 2, the third edge passes through the point B 1, and the fourth edge passes through the point B 2
Secondly, a K-means module is led into the Python, four side equations of the first bounding box and four side equations of the second bounding box obtained in the first step are respectively led in, the number of discrete points n 1 in the first bounding box and the number of discrete points n 2 in the second bounding box are calculated by using a clustering algorithm, n 1 and n 2 are compared, the bounding box with the largest number of discrete points is selected as a serial number bounding box, and the slope of the long side of the serial number bounding box is obtained;
Thirdly, calculating the inclination angle theta of the sequence number bounding box by combining the first step and the second step, and solving the center point coordinate of the sequence number bounding box, wherein theta refers to the included angle between the long side of the sequence number bounding box and the x-axis;
a fourth step of rotating the noise reduction processing image used in the first step clockwise around the central point of the sequence number bounding box according to an inclination angle theta to obtain an alignment processing image;
training an image production sequence number prediction model by using an MLP neural network algorithm, wherein the specific process is as follows:
The first step, repeating the second step to the fourth step for a plurality of times to obtain a large number of alignment treatment images, wherein the alignment treatment images are taken as image samples, and the production serial numbers corresponding to the image samples are manually determined;
Secondly, introducing a neural network module into python, extracting an image sample obtained in the first step, and carrying out production sequence number identification by using an MPL neural network algorithm to obtain an identification result of the image sample;
Thirdly, judging whether the identification result is matched with the manually identified production serial number result, if the identification result is failed to be matched, returning to the second step to reuse the MPL neural network algorithm to identify the production serial number of the image sample; if the matching is successful, returning to the second step to extract the next image sample for production serial number identification until all the image samples can be successfully identified, and obtaining an image production serial number identification model;
Step six, selecting a steel structure body in the actual operation of the oil-gas module, sequentially carrying out the processes of step two, step three and step four, and calculating the alignment treatment image obtained in step four in python by using the image production sequence number prediction model obtained in step five to obtain the production sequence number of the steel structure;
And step seven, the computer sends a request to the server through a network, a steel structure corresponding to the production sequence in the step six is found in a database of the server, and basic information of the steel structure is displayed on a display screen of a working site through network transmission, so that the working on site is facilitated.
The invention has the advantages that: the server remote database is used for storing basic information of various steel structures, so that the integrity and the comprehensiveness of the information are ensured, and the multi-terminal use on an operation site is also ensured; the production serial number is added on the steel structure in a laser code mode, so that the production serial number can resist the environment of a high-temperature manufacturing site, and corrosion and damage during service are resisted; the neural network is used for obtaining the image character prediction model to finish the identification work of the production serial number, so that the identification efficiency and accuracy can be greatly improved; the main calculation and data processing are concentrated before the operation, so that the calculated amount is less during information searching, and the efficiency of information searching is improved; the method can effectively solve the problems of complicated information searching work, small information acquisition amount, poor information integrity and the like in the manufacturing process of the oil-gas module, and is efficient, reliable, high in universality and high in adaptability.
Drawings
FIG. 1 is a flow chart of a method of extracting steel structure information suitable for use in the manufacturing process of an oil and gas module;
FIG. 2 is a schematic diagram of a steel structure information extraction method data transmission suitable for use in the oil and gas module manufacturing process;
FIG. 3 is a schematic diagram of a steel structure information extraction method bounding box setup suitable for use in the oil and gas module manufacturing process.
Detailed Description
The invention will now be described in detail with reference to the drawings and to specific embodiments.
The invention relates to a steel structure information extraction method suitable for an oil-gas module manufacturing process, which is described in the accompanying drawings and comprises the following steps:
When various steel structures of the oil-gas module are built, unique production serial numbers are engraved on the surfaces of the steel structures by utilizing lasers, and the production serial numbers and various basic information (such as length information) of the steel structures are stored in a database at a server end in a key value pair mode, wherein the production serial numbers are keys, and the various basic information are values.
And secondly, photographing the production serial number on the surface of the steel structure by using a CCD visual camera, enabling the lens plane of the CCD visual camera to be parallel to the surface of the steel structure during photographing, and transmitting the photographed picture to a computer through an industrial Ethernet.
And thirdly, opening a photo shot by the CCD vision camera in a python programming environment, and sequentially introducing a median filtering algorithm and a wiener filtering algorithm to perform noise reduction treatment on the photo to obtain a noise reduction treatment image.
Step four, aligning the noise reduction image in a computer, wherein the specific process is as follows:
Firstly, introducing numpy modules into Python, performing discrete point coordinate extraction operation on the noise reduction processed image obtained in the second step, and obtaining a point A 1 with the largest abscissa, a point A 2 with the smallest abscissa, a point B 1 with the largest ordinate and a point B 2 with the smallest ordinate in all discrete points, creating a first bounding box and a second bounding box according to four points, wherein the two bounding boxes are rectangular, and establishing a linear equation of four sides of each bounding box. Wherein the first and second sides of the first bounding box are parallel to line a 1A2, and the first side passes through point B 1, the second side passes through point B 2, the third side passes through point a 1, and the fourth side passes through point a 2. The second enclosure has a first side and a second side parallel to line B 1B2, with the first side passing through point a 1, the second side passing through point a 2, the third side passing through point B 1, and the fourth side passing through point B 2. From this, the first, second, third and fourth side equations of the first bounding box are respectively:
the first, second, third and fourth side equations of the second enclosure are respectively:
Wherein x 1、y1 represents the abscissa and the ordinate of the A 1 point respectively, x 2、y2 represents the abscissa and the ordinate of the A 2 point respectively, x 3、y3 represents the abscissa and the ordinate of the B 1 point respectively, and x 4、y4 represents the abscissa and the ordinate of the B 2 point respectively; k iA denotes the ith slope of the first bounding box. k iB denotes the ith slope of the second bounding box. b iA represents the first bounding box ith intercept and b iB represents the second bounding box ith intercept.
Secondly, a K-means module is led into the Python, four side equations of the first bounding box and four side equations of the second bounding box obtained in the first step are respectively led in, the number of discrete points n 1 in the first bounding box and the number of discrete points n 2 in the second bounding box are calculated by using a clustering algorithm, n 1 and n 2 are compared, the bounding box with the largest number of discrete points is selected as a serial number bounding box, and the slope of the long side of the serial number bounding box is obtained, and the process is as follows:
The equation of the ith edge of the numbered bounding box is
y=kix+bi
K i=kiA or k iB,bi=biA or b iB are known.
The intersection point of the first side and the third side of the sequence number bounding box is C 13, the intersection point of the first side and the fourth side is C 14, the coordinate of the intersection point of the second side and the third side is C 23.C13, the coordinate can be obtained by the combination of the first side and the third side equation, and the coordinate of the intersection point of the second side and the third side is C 14,C23. Comparing the length of the line segment C 13C23 with the length of the line segment C 13C14, and recording the slope of the longer line segment as k
And thirdly, calculating the inclination angle theta of the sequence number bounding box by combining the first step and the second step, and solving the center point coordinate of the sequence number bounding box, wherein theta refers to the included angle between the long side of the sequence number bounding box and the x-axis.
θ=arctank
If the coordinates of the central point of the sequence number bounding box are (x 0,y0), then there is
Where x 14、y14 represents the abscissa and ordinate of point C 14 and x 23、y23 represents the abscissa and ordinate of point C 23;
and fourthly, rotating the noise reduction processing image used in the first step clockwise around the central point (x 0,y0) of the sequence number bounding box according to the inclination angle theta to obtain an alignment processing image.
Training an image production sequence number prediction model by using an MLP neural network algorithm, wherein the specific process is as follows:
and step one, repeating the step two to the step four for a plurality of times to obtain a large number of alignment processing images, wherein the alignment processing images are taken as image samples, and the production serial numbers corresponding to the image samples are manually determined.
Secondly, introducing a neural network module into python, extracting an image sample obtained in the first step, and carrying out production sequence number identification by using an MPL neural network algorithm to obtain an identification result of the image sample;
And thirdly, judging whether the identification result is matched with the manually identified production serial number result, if the identification result is failed to be matched, returning to the second step to reuse the MPL neural network algorithm to carry out production serial number identification on the image sample (automatically strengthening a model in the training process of the neural network to enable the next identification to be more accurate). And if the matching is successful, returning to the second step to extract the next image sample for production serial number identification until all the image samples can be successfully identified, and obtaining an image production serial number identification model.
Step six, selecting a steel structure body in the actual operation of the oil-gas module, sequentially passing through the step two, the step three and the step four, and calculating the alignment processing image obtained in the step four in python by using the image production sequence number prediction model obtained in the step five to obtain the production sequence number of the steel structure.
And step seven, the computer sends a request to the server through a network, a steel structure corresponding to the production sequence in the step six is found in a database of the server, and basic information of the steel structure is displayed on a display screen of a working site through network transmission, so that the working on site is facilitated.
Claims (1)
1. The steel structure information extraction method suitable for the manufacturing process of the oil and gas module is characterized by comprising the following steps of:
When various steel structures of the oil-gas module are built, unique production serial numbers are engraved on the surfaces of the steel structures by utilizing lasers, and the production serial numbers and various basic information of the steel structures are stored in a database at a server side in a key value pair mode, wherein the production serial numbers are keys, and the various basic information are values;
Secondly, shooting a production sequence number on the surface of the steel structure by using a CCD visual camera, enabling the lens plane of the CCD visual camera to be parallel to the surface of the steel structure during shooting, and transmitting the shot picture to a computer through an industrial Ethernet;
Step three, opening a photo shot by a CCD visual camera in a python programming environment, and sequentially importing a median filtering algorithm and a wiener filtering algorithm to perform noise reduction treatment on the photo to obtain a noise reduction treatment image;
Step four, aligning the noise reduction image in a computer, wherein the specific process is as follows:
Firstly, introducing numpy modules into Python, performing discrete point coordinate extraction operation on the noise reduction processed image obtained in the second step, and obtaining a point A 1 with the largest abscissa, a point A 2 with the smallest abscissa, a point B 1 with the largest ordinate and a point B 2 with the smallest ordinate in all discrete points, creating a first bounding box and a second bounding box according to four points, wherein the two bounding boxes are rectangular, and establishing a linear equation of four sides of each bounding box; wherein the first and second sides of the first bounding box are parallel to line a 1A2, and the first side passes through point B 1, the second side passes through point B 2, the third side passes through point a 1, and the fourth side passes through point a 2; the first edge and the second edge of the second enclosure are parallel to the straight line B 1B2, and the first edge passes through the point A 1, the second edge passes through the point A 2, the third edge passes through the point B 1, and the fourth edge passes through the point B 2
Secondly, a K-means module is led into the Python, four side equations of the first bounding box and four side equations of the second bounding box obtained in the first step are respectively led in, the number of discrete points n 1 in the first bounding box and the number of discrete points n 2 in the second bounding box are calculated by using a clustering algorithm, n 1 and n 2 are compared, the bounding box with the largest number of discrete points is selected as a serial number bounding box, and the slope of the long side of the serial number bounding box is obtained;
Thirdly, calculating the inclination angle theta of the sequence number bounding box by combining the first step and the second step, and solving the center point coordinate of the sequence number bounding box, wherein theta refers to the included angle between the long side of the sequence number bounding box and the x-axis;
a fourth step of rotating the noise reduction processing image used in the first step clockwise around the central point of the sequence number bounding box according to an inclination angle theta to obtain an alignment processing image;
training an image production sequence number prediction model by using an MLP neural network algorithm, wherein the specific process is as follows:
The first step, repeating the second step to the fourth step for a plurality of times to obtain a large number of alignment treatment images, wherein the alignment treatment images are taken as image samples, and the production serial numbers corresponding to the image samples are manually determined;
Secondly, introducing a neural network module into python, extracting an image sample obtained in the first step, and carrying out production sequence number identification by using an MPL neural network algorithm to obtain an identification result of the image sample;
Thirdly, judging whether the identification result is matched with the manually identified production serial number result, if the identification result is failed to be matched, returning to the second step to reuse the MPL neural network algorithm to identify the production serial number of the image sample; if the matching is successful, returning to the second step to extract the next image sample for production serial number identification until all the image samples can be successfully identified, and obtaining an image production serial number identification model;
Step six, selecting a steel structure body in the actual operation of the oil-gas module, sequentially carrying out the processes of step two, step three and step four, and calculating the alignment treatment image obtained in step four in python by using the image production sequence number prediction model obtained in step five to obtain the production sequence number of the steel structure;
And step seven, the computer sends a request to the server through a network, a steel structure corresponding to the production sequence in the step six is found in a database of the server, and basic information of the steel structure is displayed on a display screen of a working site through network transmission, so that the working on site is facilitated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111495100.2A CN114331973B (en) | 2021-12-08 | 2021-12-08 | Steel structure information extraction method suitable for oil-gas module manufacturing process |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111495100.2A CN114331973B (en) | 2021-12-08 | 2021-12-08 | Steel structure information extraction method suitable for oil-gas module manufacturing process |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114331973A CN114331973A (en) | 2022-04-12 |
CN114331973B true CN114331973B (en) | 2024-06-04 |
Family
ID=81051165
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111495100.2A Active CN114331973B (en) | 2021-12-08 | 2021-12-08 | Steel structure information extraction method suitable for oil-gas module manufacturing process |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114331973B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104574502A (en) * | 2014-12-22 | 2015-04-29 | 博迈科海洋工程股份有限公司 | Laser cross section feature identification method based on steel structure model |
CN108416355A (en) * | 2018-03-09 | 2018-08-17 | 浙江大学 | A kind of acquisition method of the industry spot creation data based on machine vision |
CN112749502A (en) * | 2021-01-27 | 2021-05-04 | 天津博迈科海洋工程有限公司 | Regional virtual assembly lightweight method for oil-gas platform module |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR102256181B1 (en) * | 2019-12-30 | 2021-05-27 | 한국과학기술원 | Method of inspecting and evaluating coating state of steel structure and system for the same |
-
2021
- 2021-12-08 CN CN202111495100.2A patent/CN114331973B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104574502A (en) * | 2014-12-22 | 2015-04-29 | 博迈科海洋工程股份有限公司 | Laser cross section feature identification method based on steel structure model |
CN108416355A (en) * | 2018-03-09 | 2018-08-17 | 浙江大学 | A kind of acquisition method of the industry spot creation data based on machine vision |
CN112749502A (en) * | 2021-01-27 | 2021-05-04 | 天津博迈科海洋工程有限公司 | Regional virtual assembly lightweight method for oil-gas platform module |
Also Published As
Publication number | Publication date |
---|---|
CN114331973A (en) | 2022-04-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110543878B (en) | Pointer instrument reading identification method based on neural network | |
CN108510476B (en) | Mobile phone screen circuit detection method based on machine vision | |
WO2017067321A1 (en) | Pcb card matching method and device based on outer contour | |
CN105740899A (en) | Machine vision image characteristic point detection and matching combination optimization method | |
CN110246168A (en) | A kind of feature matching method of mobile crusing robot binocular image splicing | |
CN107977656A (en) | A kind of pedestrian recognition methods and system again | |
CN111401358B (en) | Instrument dial correction method based on neural network | |
CN109034237B (en) | Loop detection method based on convolutional neural network signposts and sequence search | |
CN108694716B (en) | Workpiece detection method, model training method and equipment | |
CN111144401A (en) | Touch screen control operation method for deep learning and visual servo of power plant centralized control room | |
CN111167731B (en) | Product sorting method, product sorting system and intelligent sorting robot | |
CN114092478B (en) | Anomaly detection method | |
CN111783672A (en) | Image feature identification method for improving bridge dynamic displacement precision | |
CN114331973B (en) | Steel structure information extraction method suitable for oil-gas module manufacturing process | |
CN114235815A (en) | Method for detecting surface defects of outdoor electrical equipment of converter station based on scene filtering | |
CN114219753A (en) | Power equipment surface defect detection method based on deep learning and terminal | |
CN113705564A (en) | Pointer type instrument identification reading method | |
CN117152727A (en) | Automatic reading method of pointer instrument for inspection robot | |
CN107193965A (en) | A kind of quick indoor orientation method based on BoVW algorithms | |
CN112365399B (en) | Deep learning-based panoramic stitching method and system for fan blade images | |
US20230360262A1 (en) | Object pose recognition method based on triangulation and probability weighted ransac algorithm | |
CN106778915A (en) | A kind of target matching method towards multiple-camera | |
CN113947763A (en) | Pavement arrow recognition method and device based on template self-supervision | |
CN112149671A (en) | Target area accurate positioning method based on normalized cross-correlation template matching | |
CN110070110A (en) | A kind of adaptive threshold image matching method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |