CN-122023525-A - Machine vision-based process tube assembly method

CN122023525ACN 122023525 ACN122023525 ACN 122023525ACN-122023525-A

Abstract

The invention provides a process tube group alignment method based on machine vision. The method comprises the steps of respectively collecting end face images of a first pipeline element and end face images of a second pipeline element through at least two industrial cameras, calibrating the industrial cameras, establishing a mapping relation between pixel coordinates and measurement plane coordinates, executing preset known displacement operation under the condition that camera view fields are not overlapped, combining characteristic coordinate changes before and after displacement, determining a relative position relation among the cameras, establishing a unified coordinate system, extracting circular ring outlines from the end face images, performing circular fitting to obtain center coordinates of two ends, determining rotation angle parameters based on an extreme value of an inner diameter radiation distance, calculating centering translation quantity and group rotation quantity under the unified coordinate system, and outputting a motion control instruction to drive an executing mechanism to complete pairing. According to the technical scheme, automatic centering and angular pairing of the process pipe are realized, and the pairing precision and efficiency are improved.

Inventors

LIU SIYUAN
ZHAO YAN
CAO QIMING
HAN FANGFANG
GUO HAILIANG
LI ZHIWEI
CAO XIN
Yang Yedan

Assignees

中天智能装备(天津)有限公司
天津理工大学

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. A machine vision-based process tube assembly method, the method comprising: respectively acquiring images of the end face of the first pipeline element and the end face of the second pipeline element through at least two industrial cameras; calibrating each industrial camera, and establishing a mapping relation between pixel coordinates of an end face image and corresponding measurement plane coordinates; Under the condition that the fields of view of the at least two industrial cameras are not overlapped, converting the characteristic coordinates into measurement plane coordinates through the mapping relation based on preset known displacement operation and characteristic coordinate changes of images acquired before and after displacement, determining the relative position relation between the at least two industrial cameras, and establishing a unified coordinate system; Based on the images of the end surfaces of the first pipeline element and the second pipeline element, respectively extracting the outline characteristics of an end surface circular ring, respectively obtaining the circle center coordinates of the first pipeline element and the second pipeline element, and determining a rotation angle parameter based on the extreme value of the radial distance of the inner diameter of the circular ring outline; calculating centering translation quantity based on two circle center coordinates in a unified coordinate system, and calculating group centering rotation quantity based on the rotation angle parameter; And outputting a motion control instruction according to the centering translation amount and the pairing rotation amount, and performing centering translation and relative rotation on the first pipeline element and the second pipeline element by the pairing execution mechanism according to the motion control instruction to complete pairing.
2. The machine vision based process tube assembly alignment method of claim 1, wherein the centering translation is performed by a device table for carrying the second pipe element, the relative rotation is performed by a support mechanism for carrying the first pipe element, the device table performs centering translation driven by a translation mechanism, and the support mechanism performs relative rotation driven by a rotation mechanism.
3. The machine vision based process tube assembly method of claim 1, wherein the at least two industrial cameras are disposed at a first pipe element end face measurement location and a second pipe element end face measurement location, respectively, and acquire end face images toward the corresponding end faces, respectively.
4. The machine vision-based process tube alignment method of claim 1, wherein determining the relative positional relationship between the at least two industrial cameras comprises performing displacement operations in at least two non-collinear directions based on the preset known displacement operations, respectively, solving translational relationship parameters between the at least two industrial cameras based on feature coordinate changes of acquired images before and after each displacement operation, and establishing the unified coordinate system based on the translational relationship parameters.
5. The machine vision based process tube alignment method of claim 1, wherein the known displacement operation comprises: and controlling the first pipeline element and/or the second pipeline element to execute at least one preset displacement along the preset direction in the clamping state, and respectively acquiring end face images before and after each displacement.
6. The machine vision based process tube pairing method of claim 4, wherein the determining a relative positional relationship between the at least two industrial cameras comprises: And correlating the coordinate change with the displacement corresponding to the known displacement operation to solve the translation relation parameters among the industrial cameras, and establishing the unified coordinate system based on the translation relation parameters.
7. The machine vision based process tube alignment method of claim 1, wherein the extracting the end face ring profile features respectively, obtaining center coordinates of the first and second pipe elements respectively comprises: and performing distortion correction and preprocessing on the end face image, extracting an inner contour point set of the circular contour from the preprocessed end face image, performing circle fitting on the inner contour point set, and outputting circle center coordinates.
8. The machine vision based process tube pairing method of claim 1, wherein the determining a rotation angle parameter based on an inside diameter radial distance extremum of the annular ring profile comprises: And calculating the inner diameter radiation distance in a plurality of radiation angle directions by taking a preset radiation center as a reference, and determining the maximum value of the inner diameter radiation distance and the corresponding angle thereof as the rotation angle parameter, or determining the minimum value of the inner diameter radiation distance and the corresponding angle thereof as the rotation angle parameter.
9. The machine vision based process tube alignment method of claim 2, wherein, The centering translation amount is determined based on a coordinate difference between the center coordinates of the first pipeline element and the center coordinates of the second pipeline element under the unified coordinate system; The set of paired rotation amounts is determined based on an angle difference of the rotation angle parameter of the first pipe element and the rotation angle parameter of the second pipe element; the outputting of the motion control instruction includes outputting a displacement instruction corresponding to the centering translational amount to the translational mechanism, and outputting a rotation instruction corresponding to the set of centering rotational amounts to the rotation mechanism.
10. A machine vision-based process tube assembly system, the system comprising: a device table for carrying the second pipe element and being translatable; A support mechanism for carrying the first pipe element and being rotatable; The at least two industrial cameras are respectively used for acquiring images of the end face of the first pipeline element and the end face of the second pipeline element, and the fields of view of the at least two industrial cameras are not overlapped with each other; The pairing executing mechanism comprises a translation mechanism for driving the device table to translate and a rotating mechanism for driving the supporting mechanism to rotate; The motion controller is connected with the pairing executing mechanism and used for outputting a motion control instruction; Computer device comprising a processor and a memory, the memory having stored therein computer instructions for executing the computer instructions, which when executed by the processor, cause the system to implement the machine vision based process tube pairing method according to any one of claims 1-9.

Description

Machine vision-based process tube assembly method Technical Field The application relates to the field of image processing technology and intelligent machinery, in particular to a process tube assembly method based on machine vision. Background The process pipeline is widely applied to industries such as petrochemical industry, ship manufacturing, nuclear power, oil gas transmission and the like. In the pipeline construction process, the prefabricated assembly of the pipelines is one of key working procedures before welding. The process tube assembly generally refers to the process of positioning, aligning, fixing and adjusting gaps of two or more pipeline elements before welding, and the assembly quality (such as indexes of gap, misalignment, angle deviation and the like) directly influences the subsequent welding quality and the operation safety of the pipeline. In the prior art tube set pair operation, common technical paths mainly comprise an artificial pair, a mechanical/hydraulic pair device auxiliary pair, sensor auxiliary detection and the like, but the following defects still exist: The manual assembly mode depends on plumber experience and visual inspection, and is usually adjusted by means of a chain block, a jack, a spiral jackscrew, a wedge or a simple mouth-gag and other tools. For large-caliber or thick-wall pipelines, the workpiece weight is large, the tiny misalignment amount and the gap adjustment are laborious, the cooperation of multiple people is often needed, the time consumption is long, the labor intensity is high, the efficiency is low, meanwhile, the assembly accuracy is obviously influenced by experience of operators, the quality fluctuation is large, the requirements of high-standard application scenes are difficult to be met stably, and the safety risks such as extrusion, injury and the like exist in manual conveying and adjustment of heavy pipelines. In order to improve efficiency and quality, mechanical or hydraulic pairing machines are adopted in part of sites, and forced pairing is realized in a clamping, enclasping, rounding and other modes. In actual operation, operators still need to repeatedly stop the machine and measure by using a feeler gauge, a welded junction detection gauge or a caliper gauge and then finely adjust the machine to form an adjustment-measurement-readjustment cycle, the degree of automation is not high, and in addition, when the pipe has the conditions of ovality errors, uneven end face cutting and the like, the adaptability of the rigid clamp to complex shapes is limited, and the problems of exceeding of local misalignment amount and the like are easy to occur. With the development of automation, partial equipment is introduced into a contact sensor such as displacement to perform auxiliary centering detection, but the contact measurement is required to be in direct contact with the surface of a pipe orifice or a group of components, so that the contact measurement is easily influenced by rust, greasy dirt, splashing, burrs, surface roughness and the like to generate reading distortion or failure, interference factors such as dust, greasy dirt, vibration and the like further reduce measurement reliability and stability under complex environments such as a prefabricated workshop or outdoor construction, and meanwhile, secondary disturbance can be introduced into the contact process to cause micro displacement errors, scratch, pollution and the like are generated on the surface of a cleaned or sprayed anticorrosive layer, and the subsequent welding and anticorrosive performance are influenced. In conclusion, the existing process tube group has the problems of low efficiency, great influence on the alignment quality by manual experience, difficulty in stable quantitative control of alignment state, insufficient adaptability and reliability under complex working conditions and the like. Therefore, there is a need for a pairing method suitable for a process tube prefabricated pairing scene to improve pairing efficiency and pairing accuracy and enhance stability and safety of field operation. Disclosure of Invention The application provides a process tube alignment method based on machine vision, which comprises the steps of constructing a system comprising a translatable device table, a rotatable supporting mechanism, at least two industrial cameras with non-overlapping view fields, a motion controller and an alignment executing mechanism, introducing mapping calibration from pixel coordinates to measurement plane coordinates in end face measurement and solving a camera relative position relation based on preset known displacement operation to establish a unified coordinate system, further extracting outline characteristics of circular rings from images of two end faces, calculating circle center coordinates, determining rotation angle parameters based on an inner diameter radiation distance extremum, outputting centering translation quantity and alignmen