Search

CN-122016862-A - System and method for online detection of surface defects of petroleum pipe fitting by fusing multi-scale characteristics

CN122016862ACN 122016862 ACN122016862 ACN 122016862ACN-122016862-A

Abstract

The invention belongs to the technical field of defect detection, and discloses an online detection system and method for the surface defects of petroleum pipe fittings fused with multi-scale characteristics; the method comprises the steps of collecting multi-element detection data of the petroleum pipe fitting, carrying out data cleaning to obtain a multi-element original data set, carrying out oil film boundary identification, carrying out oil film influence inhibition, outputting an oil film inhibition data set, identifying a roughness area and an electromagnetic abnormal response area, carrying out space projection correction to generate a multi-channel space alignment data set, carrying out joint analysis, identifying boundary response characteristics of a welding seam structure area, outputting a welding seam characteristic subset, carrying out distribution comparison, outputting a rust boundary area, carrying out physical consistency analysis, outputting a trusted defect fusion block set, establishing a defect judgment rule to carry out defect detection, and sending a detection result to a preset online monitoring terminal, thereby improving the multi-physical quantity fusion sensing capability and multi-scale characteristic collaborative analysis precision of the online detection of the surface defects of the petroleum pipe fitting.

Inventors

  • LI MENGZHI
  • JIANG LEI
  • MA JI

Assignees

  • 江阴市南方管件制造有限公司

Dates

Publication Date
20260512
Application Date
20260323

Claims (10)

  1. 1. The on-line detection method for the surface defects of the petroleum pipe fitting integrating the multi-scale features is characterized by comprising the following steps: S1, acquiring multi-element detection data of a petroleum pipe fitting and performing data cleaning to obtain a multi-element original data set, wherein the multi-element original data set comprises optical characteristic image data, surface electrical response data and pipeline surface temperature data; S2, carrying out oil film boundary identification by combining the optical characteristic image data and the pipeline surface temperature data, carrying out oil film influence inhibition in an oil film boundary region, adjusting the numerical value of a multistage channel, and outputting an oil film inhibition data set; S3, extracting a suppression optical image of the oil film suppression data set, extracting rough image features and electromagnetic induction change features in the suppression optical image by combining surface electrical response data, and identifying a roughness region and an electromagnetic abnormal response region; S4, extracting data slices containing the welding seam structural region in the multichannel spatial alignment data set, carrying out joint analysis on the data slices based on temperature data and electromagnetic response data of the welding seam structural region, identifying boundary response characteristics of the welding seam structural region, and outputting a welding seam characteristic subset; S5, identifying image blocks containing rust areas in the multichannel spatial alignment data set, carrying out distribution comparison on the image blocks, and outputting rust boundary areas; s6, combining the weld joint feature subset, the credible defect fusion block set and the multichannel space alignment data set, establishing a defect judging rule to execute defect detection, and sending a detection result to a preset online monitoring terminal.
  2. 2. The method for online detection of surface defects of petroleum pipe fitting by fusing multi-scale features according to claim 1, wherein the mode of performing oil film boundary identification comprises the following steps: Extracting pixel point brightness values of each optical image in the optical characteristic image data, simultaneously acquiring a light source incident intensity parameter, and calculating the reflectivity of the surface of the pipe fitting based on the pixel point brightness values and the light source incident intensity parameter; extracting the radiation temperature of each coordinate sample point in the pipeline surface temperature data, comparing the radiation temperature with a second temperature reference, and extracting a sample point area with the radiation temperature of the pipeline surface lower than the second temperature reference as a low temperature area; Establishing a space coordinate corresponding relation between the low-reflectivity region and the low-temperature region, calculating the region overlapping area, and screening a pipeline surface region with the region overlapping area higher than a preset space overlapping threshold value as a joint response region; Constructing a multi-scale sliding window to perform joint calculation of local reflectivity gradient and thermal response gradient on the joint response region, outputting a joint gradient map of the pipeline surface, and screening a space region with gradient values reaching preset gradient strength in the joint gradient map as a brightness-temperature boundary transition region; and constructing a boundary contour curve based on the gradient direction of the combined gradient in the brightness-temperature boundary transition region, and outputting an oil film boundary region.
  3. 3. The method for online detection of surface defects of petroleum pipe fitting with multi-scale features fused according to claim 2, wherein the mode of oil film influence inhibition comprises: calculating the oil film surface reflectivity of each pixel point in the oil film boundary area, carrying out normalization processing in combination with the ambient brightness reference mean value, and outputting the corrected reflectivity of the corresponding pixel point; Calculating the difference between the corrected reflectivity and the standard reflectivity of the film-free surface, constructing a reflection compensation factor of the corresponding pixel point based on the difference, and adjusting the reflectivity value of the optical channel where the corresponding pixel point is positioned by using the reflection compensation factor to output the compensated reflectivity; Calculating the thermal diffusion gradient of each coordinate sample point in the oil film boundary area, and calculating the heat flow response offset of the corresponding coordinate sample point based on the thermal diffusion gradient; And jointly mapping the compensation reflectivity and the temperature correction value to corresponding coordinate positions, and integrating all the adjusted data to obtain an oil film inhibition data set.
  4. 4. The method for online detection of surface defects of petroleum pipe fitting by fusing multi-scale features as claimed in claim 3, wherein the manner of identifying the roughness region and the electromagnetic abnormal response region comprises: Dividing each suppressed optical image into grid areas, calculating the reflectivity variance of each grid area, and simultaneously calculating the microstructure direction change amplitude of adjacent grid areas; if the reflectivity variance is higher than the rough fluctuation threshold and the microstructure direction change amplitude is higher than the rough direction change threshold, taking a region formed by the corresponding grid region as a roughness region; And extracting an induced impedance change curve of each vortex space point in the surface electrical response data, calculating a response change slope, and taking a region formed by the continuous vortex space points as an electromagnetic abnormal response region if the response change slope of the continuous vortex space points is higher than an electromagnetic response slope threshold value and the number of the continuous vortex space points is higher than a preset number threshold value.
  5. 5. The method for online detection of surface defects of petroleum pipe fitting with multi-scale feature fusion according to claim 4, wherein the method for performing spatial projection correction comprises: Synchronously extracting the space center coordinates of each eddy current response subarea in the electromagnetic abnormal response area, recording the conductivity of eddy current space points in the eddy current response subarea, and calculating the conductivity mean value and the conductivity change gradient based on the conductivity; performing space adjacent pairing on the grid region and the vortex response sub-region, calculating the position of the central coordinate of each paired region and the space distance offset of the space central coordinate, and performing main direction projection on the space distance offset to obtain a channel response offset distance; Estimating the surface roughness grade of a corresponding area based on the reflectivity mean value of the grid area, and calculating the space drift compensation quantity of the electromagnetic response according to the corresponding relation between the roughness grade and the vortex lift-off effect; superposing the channel response offset distance and the space drift compensation quantity, and constructing a coordinate transformation function by combining the direction included angle between the reflectivity gradient direction and the direction included angle between the conductivity change gradient; calculating the optical matching coordinates of each eddy current space point in the electromagnetic abnormal response area by using the coordinate transformation function; and binding the reflectivity and the conductivity to uniform space positions based on the optical matching coordinates, calculating channel response correlation, determining that the space positions are effective when the correlation is higher than a preset correlation coefficient threshold value, and outputting a fused multichannel space alignment data set.
  6. 6. The method for online detection of surface defects of petroleum pipe fitting with multi-scale features according to claim 5, wherein the mode of performing joint analysis comprises: dividing vertical section lines for the welding line structure area, and extracting the radiation temperature and magnetic permeability of each sampling point on each section line; Sequencing the magnetic permeability on each section line based on the distance from the center of the welding seam, and calculating the magnetic permeability change rate of adjacent sampling points; Sequencing the radiation temperature on each section line according to the distance from the center of the welding line, calculating the temperature gradient of adjacent sampling points, identifying the change inflection point of the temperature gradient, and taking the position of the change inflection point as the change boundary of the thermal diffusivity; Calculating the difference value between the magnetic permeability and the corresponding magnetic permeability change standard, and outputting a magnetic permeability difference value; Calculating the radiation temperature average value of all sampling points in the welding line structure area, screening a continuous area formed by sampling points with the temperature difference value between the radiation temperature and the radiation temperature average value on each section line being greater than a preset temperature deviation threshold value, and marking the continuous area as a coarse-scale thermal anomaly area; Carrying out space overlapping on the fine-scale magnetic anomaly response points and the coarse-scale thermal anomaly regions, and multiplying the magnetic conductance difference values of the fine-scale magnetic anomaly response points and the temperature difference values of the corresponding positions to obtain defect response intensity if the space positions of the fine-scale magnetic anomaly response points are located in the coarse-scale thermal anomaly regions; And screening the fine-scale magnetic abnormal response points with the defect response intensity higher than the preset defect response threshold value, and integrating the spatial position coordinates, the magnetic conductance difference value, the temperature difference value and the defect response intensity to output a weld joint feature subset.
  7. 7. The method for online detection of surface defects of petroleum pipe fitting with multi-scale features according to claim 6, wherein the distribution comparison method comprises: and calculating the spatial gradient distribution of the color channels in the image block and the spatial difference distribution of the impedance amplitude values of the electrochemical channels corresponding to the spatial regions, extracting a response inconsistent region based on the variation trend difference of the spatial gradient distribution and the spatial difference distribution, and screening the coincident region as a corrosion boundary region.
  8. 8. The method for online detection of surface defects of petroleum pipe fitting with multi-scale features according to claim 7, wherein the manner of performing the physical consistency analysis comprises: extracting color components and electrochemical impedance of each rust sampling point in the rust boundary area, and respectively combining spatial gradient distribution and spatial difference distribution to construct a color gradient vector and an impedance vector; the method comprises the steps of identifying the main extension direction of a rust boundary area, setting a multi-scale analysis window along the main extension direction, fitting a color gradient vector set in each analysis window to generate a gradual trend curve, and fitting an impedance vector set to generate a step response curve; Calculating the deviation value of the color component of each corrosion sampling point and the corresponding position of the gradual change trend curve to be used as a color residual value, and simultaneously calculating the electrochemical impedance of the corrosion sampling point and the corresponding position deviation value of the step response curve to be used as an impedance residual value; judging whether the color residual value of each rust sampling point is consistent with the sign of the impedance residual value, and marking the rust sampling points with the same sign as the response homodromous points; the absolute value ratio of the color residual value and the impedance residual value of each response homodromous point is calculated to be used as a double-channel response ratio, response homodromous points with the double-channel response ratio in a preset proportion interval are screened to be used as physical response consistency points, and the area formed by all the physical response consistency points is integrated to be used as a trusted defect fusion block set.
  9. 9. The method for online detection of surface defects of petroleum pipe fitting with multi-scale features according to claim 8, wherein the manner of performing the defect detection comprises: Integrating the weld joint feature subset, the trusted defect fusion block set and the multichannel space alignment data set into a data set to be detected, taking a defect judgment rule as a basis, matching the data set to be detected with corresponding data dimensions in the defect judgment rule, and outputting a detection result.
  10. 10. An on-line detection system for surface defects of petroleum pipe fitting fusing multi-scale features, for realizing the on-line detection method for surface defects of petroleum pipe fitting fusing multi-scale features as defined in any one of claims 1 to 9, comprising: The data acquisition module acquires multi-element detection data of the petroleum pipe fitting and performs data cleaning to obtain a multi-element original data set, wherein the multi-element original data set comprises optical characteristic image data, surface electric response data and pipeline surface temperature data; The oil film suppression module is used for carrying out oil film boundary recognition by combining the optical characteristic image data and the pipeline surface temperature data, carrying out oil film influence suppression in an oil film boundary region, adjusting the numerical value of the multistage channel and outputting an oil film suppression data set; The spatial registration module extracts a suppression optical image of the oil film suppression data set, extracts rough image features and electromagnetic induction change features in the suppression optical image in combination with surface electrical response data, and identifies a roughness region and an electromagnetic abnormal response region; the welding seam abnormal decoupling module is used for extracting data slices containing welding seam structural areas in the multichannel spatial alignment data set, carrying out joint analysis on the data slices based on temperature data and electromagnetic response data of the welding seam structural areas, identifying boundary response characteristics of the welding seam structural areas and outputting a welding seam characteristic subset; The rust identification module is used for identifying image blocks containing rust areas in the multichannel space alignment data set, carrying out distribution comparison on the image blocks and outputting rust boundary areas; The joint detection module is used for establishing a defect judging rule to execute defect detection by combining the weld joint feature subset, the credible defect fusion block set and the multichannel space alignment data set, sending a detection result to a preset online monitoring terminal, and connecting the modules in a wired and/or wireless mode.

Description

System and method for online detection of surface defects of petroleum pipe fitting by fusing multi-scale characteristics Technical Field The invention relates to the technical field of defect detection, in particular to an on-line detection system and method for surface defects of petroleum pipe fittings fused with multi-scale features. Background The detection of surface defects is an important stage and step in the petroleum pipeline manufacturing and operation and maintenance process, the process is subjected to multi-physical-quantity fusion sensing to realize accurate distinction between tiny defects and pseudo-defect interference, and the service safety and quality control of the pipeline are ensured, however, the traditional defect detection method still faces a plurality of technical bottlenecks when facing more complex surface states and defects of the petroleum pipeline. In an actual scene, the oil film is often uneven due to residual oil film of a processing or protecting process, the existence of the oil film tends to easily cause distortion of infrared radiation rate and optical reflectivity, for example, microcracks at the same position are even in temperature in infrared detection under the coverage of a thick oil film, and the optical detection is in a low-reflectivity area, however, the traditional defect detection method ignores the multi-physical-field response distortion caused by the oil film, the oil film interference and true defect signals cannot be accurately stripped, the situation that defect omission detection or the oil film is misjudged as defects is easily caused, in addition, the roughness difference of the surface of the pipe can lead to dislocation of the optical detection and the electromagnetic detection to the spatial response of the same position, the boundary of the roughness mutation area in an optical image is clear, the position deviation is generated in the eddy current response due to the lift-off effect, however, the traditional defect detection method does not lead to dynamic compensation aiming at the physical correspondence relation between the roughness grade and the lift-off effect, the multichannel characteristic fusion is often failed, so that the position judgment is wrong, meanwhile, the gradient of a metallographic structure in the welding seam area is gradually changed, so that the magnetic permeability is abnormal, the situation that the gradient is easy to be attenuated, the situation of the actual defect signal is easily is misjudged, the gradient of the gradient is poor, the gradient of the gradient-phase transition is easily is detected, the gradient-phase transition stress is large, the gradient-phase transition region is the gradient-transition stress is difficult, and the gradient-phase transition region is difficult, and the gradient-transition phase transition, the gradient test is difficult, and the gradient test is poor, and the contrast of the physical transition phase transition region is poor, and the contrast test method is difficult, and the contrast test is on the contrast, and the contrast, on the contrast, and the problem, the defect detection in the area is always in a high false alarm or high false alarm state, so that the reliability of the detection result is difficult to guarantee under the complex working condition, and the overall quality inspection efficiency of the petroleum pipe fitting production line is affected. In view of the above, the present invention provides an on-line detection system and method for surface defects of petroleum pipe fitting that incorporates multi-scale features to solve the above-mentioned problems. Disclosure of Invention In order to overcome the defects in the prior art and achieve the purposes, the invention provides a technical scheme that the on-line detection method for the surface defects of the petroleum pipe fitting integrating the multi-scale characteristics comprises the following steps: S1, acquiring multi-element detection data of a petroleum pipe fitting and performing data cleaning to obtain a multi-element original data set, wherein the multi-element original data set comprises optical characteristic image data, surface electrical response data and pipeline surface temperature data; S2, carrying out oil film boundary identification by combining the optical characteristic image data and the pipeline surface temperature data, carrying out oil film influence inhibition in an oil film boundary region, adjusting the numerical value of a multistage channel, and outputting an oil film inhibition data set; S3, extracting a suppression optical image of the oil film suppression data set, extracting rough image features and electromagnetic induction change features in the suppression optical image by combining surface electrical response data, and identifying a roughness region and an electromagnetic abnormal response region; S4, extracting data slices containing the welding seam structura