Intelligent stainless steel pipe defect detection system and method

Abstract

The invention relates to the technical field of defect detection, in particular to an intelligent stainless steel pipe defect detection system and method, and the system comprises a surface image acquisition module, a surface texture analysis module, a defect position calibration module, a defect type judgment module and a defect detail confirmation module. According to the method, a texture difference extraction and region boundary labeling method of crack and corrosion regions is adopted, high-precision detection of tiny defects is achieved, the probability of missed detection is remarkably reduced, and the defect detection accuracy is improved through center point positioning, size calculation and form description of the defects, defect feature recording and introduction of a classification mechanism for matching image features and a database. The method can realize accurate identification of different types of defects and accurate judgment of severity, dynamically evaluate defect risks and development possibilities, enhance prediction capability of detecting potential risks, improve quality control perspectiveness, realize automatic closed loop from surface feature analysis to defect classification and risk evaluation, reduce manual intervention, and improve detection efficiency. And the detection efficiency and accuracy are obviously improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of defect detection, and in particular to an intelligent stainless steel pipe fitting defect detection system and method. BACKGROUND

[0002] The technical field of defect detection involves the use of automated tools and systems to identify defects and irregularities in materials, components, or structures. The field utilizes various sensor technologies, image processing algorithms, and machine learning methods to analyze test objects in order to efficiently identify potential defects. Common techniques include X-ray imaging, infrared and thermal imaging, ultrasonic testing, and vision-based detection systems. It is widely used in the manufacturing, construction, aerospace, and automotive industries, among others, with the goal of ensuring product quality and operational safety, reducing the workload of manual inspections, and improving the accuracy and efficiency of detection.

[0003] Among them, the intelligent stainless steel pipe fitting defect detection system is a technology specially designed for defect recognition in the production and inspection process of stainless steel pipe fittings. The system integrates high-resolution cameras, image processing software, and machine learning algorithms to automatically identify and classify various defects on the surface or inside the structure of the pipe fitting, such as cracks, holes, corrosion, etc. The main uses include improving the production quality of stainless steel pipe fittings, reducing the scrap rate, and ensuring the reliability and safety of the pipe fittings used in various industrial applications, providing manufacturers with an efficient and accurate quality control tool that can significantly reduce rework and losses caused by defects.

[0004] The existing technology has weak detection capability for micro-defects, and cannot effectively distinguish subtle defects with similar texture characteristics, resulting in a high risk of missed detection. The description of defect location and size usually relies on rough calibration, making it difficult to provide accurate data support for subsequent processing, and there are limitations in analyzing defect morphology and distribution. In the defect classification process, the existing method lacks comprehensive comparison with the standard database, which can easily lead to inaccurate classification results and insufficient differentiation of complex defect types. Risk assessment relies more on static analysis, making it difficult to dynamically predict the potential development trend of defects, and the overall cognitive ability of defects is insufficient. Defects restrict the detection efficiency and accuracy, affecting the comprehensiveness and safety of quality control, resulting in unreliable detection results and increasing manufacturing and operation risks. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art and to propose an intelligent stainless steel pipe fitting defect detection system and method.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: an intelligent stainless steel pipe fitting defect detection system comprises: The surface image acquisition module captures the surface image of the stainless steel pipe fitting by a camera, adjusts the image contrast and brightness, highlights the surface features of the stainless steel pipe fitting, adjusts the image detail visibility, and generates a surface optimized image of the stainless steel pipe fitting; The surface texture analysis module analyzes the crack and corrosion area according to the surface optimized image of the stainless steel pipe fitting, extracts the texture difference in the crack and corrosion area, labels the area boundary, distinguishes the difference texture area, and generates a preliminary defect analysis report; The defect position calibration module determines the center point of the defect, calculates the defect size, records the position of each defect, and generates defect position information according to the preliminary defect analysis report; The defect type judgment module classifies each defect type, determines the defect category and severity, and generates a defect identification result based on the defect position information, and compares and analyzes the defect image with the known defect database; The defect detail confirmation module analyzes the defect characteristics, evaluates the potential risk, calculates the possibility of defect development, and generates a defect feature confirmation record according to the defect identification result.

[0007] As a further scheme of the application, the surface optimized image of the stainless steel pipe fitting includes optimized image contrast, optimized image brightness and surface feature details; the preliminary defect analysis report includes crack area features, corrosion area features, area boundary information and difference texture features; the defect position information includes defect center point coordinates, defect size parameters, defect morphology description and defect distribution position; the defect identification result includes defect category classification and defect severity level; and the defect feature confirmation record includes defect characteristic description, potential risk evaluation result and defect development possibility prediction result.

[0008] As a further scheme of the application, the surface image acquisition module includes an image shooting sub-module, an image processing sub-module and an image optimization sub-module. The image shooting sub-module captures the surface image of the stainless steel pipe fitting by a camera, adjusts the focal length and exposure time of the camera to match the difference light conditions, and generates an original surface image; The image processing sub-module adjusts the contrast and brightness based on the original surface image, enhances the image edges, highlights the surface features of the stainless steel pipe fitting, and obtains a feature enhanced image; The image optimization sub-module adjusts the detail level of the image based on the feature enhanced image, adjusts the color saturation, optimizes the image texture, and obtains a surface optimized image of the stainless steel pipe fitting.

[0009] As a further scheme of the application, the surface texture analysis module includes a crack recognition sub-module, a corrosion analysis sub-module and a texture labeling sub-module. The crack identification submodule searches and identifies the crack features in the image based on the optimized image of the surface of the stainless steel pipe fitting, subdivides the crack contour, adjusts the image parameters to distinguish the crack from the normal material, and obtains a crack feature map; The corrosion analysis submodule identifies the corrosion area by detecting the pixel intensity based on the crack feature map, discriminates and analyzes the texture difference of the corrosion area, and generates a corrosion area detail map; The texture labeling submodule automatically labels and draws the texture feature boundary based on the corrosion area detail map, distinguishes the boundary of the different texture features, and generates a preliminary defect analysis report.

[0010] As a further scheme of the present application, the defect position calibration module comprises a center point positioning submodule, a size calculation submodule, and a position recording submodule. The center point positioning submodule performs edge detection on each defect based on the preliminary defect analysis report, identifies the geometric center point, segments each defect area, and generates a defect center positioning map. The size calculation submodule performs image scale conversion based on the defect center positioning map, measures the length and width of each calibrated defect, converts the pixels to actual sizes, calculates the actual size of the defect, and obtains defect size data. The position recording submodule uses the defect size data to perform coordinate positioning, records and labels the actual position of each defect, and obtains defect position information.

[0011] As a further scheme of the present application, the defect type judgment module comprises a feature extraction submodule, a matching analysis submodule, and a severity assessment submodule. The feature extraction submodule refines the visual features of each defect based on the defect position information, extracts the shape and texture parameters of the defect, and outputs a defect feature vector. The matching analysis submodule compares the defect feature vector with a known defect type database, identifies similar features, determines the category of each defect, and obtains a matching analysis result. The severity assessment submodule analyzes the potential impact of each defect based on the matching analysis result, calculates the severity score of the defect area on the structural stability, and generates a defect recognition result.

[0012] As a further scheme of the present application, the severity score is calculated according to the formula: ; to generate the severity evaluation result of the defect, wherein, represents the severity score, represents the area of the i-th defect, represents the area of the i-th defect, represents the area of the i-th defect, a depth of a defect, representing a total number of defects identified.

[0013] As a further scheme of the present application, the defect detail confirmation module comprises a characteristic depth analysis submodule, a risk level evaluation submodule, and a development trend prediction submodule. The characteristic depth analysis submodule analyzes the morphological characteristics of each defect based on the defect identification result, reveals the geometric structure and surface characteristics of the defect, and generates a characteristic analysis report. The risk level evaluation submodule analyzes the influence of the defect on the mechanical strength and functional stability of the stainless steel pipe based on the characteristic analysis report, determines the risk level of the defect, and obtains a risk evaluation result. The development trend prediction submodule calculates the development score of the defect under future operating conditions based on the risk evaluation result, analyzes the expansion speed of the defect, and generates a defect feature confirmation record.

[0014] As a further scheme of the present application, the development score is calculated according to the formula: ; wherein, represents a probability value, represents an absolute value of the defect depth predicted by the th decision tree, represents the product of the operating conditions and environmental factors corresponding to the th decision tree under the square root, represents the total number of decision trees in the model.

[0015] An intelligent stainless steel pipe defect detection method, comprising the following steps: S1: capturing the surface image of the stainless steel pipe using a camera, adjusting the focal length and exposure time of the camera to match the different light conditions, adjusting the contrast, brightness and color saturation to obtain an optimized surface image of the stainless steel pipe; S2: based on the optimized surface image of the stainless steel pipe, searching and identifying the crack features in the image, identifying the corrosion area, automatically labeling and drawing the boundaries distinguishing the different texture features, and generating a preliminary defect analysis report; S3: based on the preliminary defect analysis report, performing edge detection on each defect, identifying the geometric center point, segmenting each defect area, measuring the length and width of each calibrated defect, calculating the actual size of the defect, and obtaining defect size data; S4: using the defect size data, recording and labeling the actual position of each defect, refining the visual features of each defect, extracting the morphological and texture parameters of the defect, and outputting a defect feature vector; S5: Comparing with the known defect type database through the defect feature vector, the category of each defect is determined, the severity score of defect area to structure stability is calculated, and the defect identification result is generated; S6: Based on the defect identification result, the morphological characteristics of each defect are analyzed, the risk level of the defect is determined, the development score of the defect under future operation conditions is calculated, and the defect feature confirmation record is generated.

[0016] Compared with the prior art, the advantages and positive effects of the present application are: In the present application, the texture difference extraction and region boundary labeling method of cracks and corrosion areas is adopted to realize high-precision detection of micro defects, significantly reduce the missed detection probability, record the defect features through the center point positioning, size calculation and morphological description of the defect, introduce the classification mechanism of image feature and database matching, and can realize accurate identification and precise judgment of the severity of different categories of defects, dynamically evaluate the defect risk and development possibility, enhance the prediction ability of potential risks, improve the forward-looking of quality control, realize the automation of the closed loop from surface feature analysis to defect classification and risk assessment, reduce manual intervention, and significantly improve the detection efficiency and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a system flowchart of the present application; Figure 2 is a system module diagram of the present application; Figure 3 is a surface image acquisition module flowchart of the present application; Figure 4 is a surface texture analysis module flowchart of the present application; Figure 5 is a defect position calibration module flowchart of the present application; Figure 6 is a defect type judgment module flowchart of the present application; Figure 7 is a defect detail confirmation module flowchart of the present application; Figure 8 is a method step flowchart of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the present application more clear and explicit, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0019] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0020] Referring to Figure 1 An intelligent stainless steel pipe fitting defect detection system comprises: The surface image acquisition module captures the surface image of the stainless steel pipe fitting through the camera, adjusts the image contrast and brightness, highlights the surface features of the stainless steel pipe fitting, adjusts the image detail visibility, and generates the stainless steel pipe fitting surface optimization image; The surface texture analysis module analyzes the crack and corrosion area according to the stainless steel pipe fitting surface optimization image, extracts the texture difference in the crack and corrosion area, labels the area boundary, distinguishes the difference texture area, and generates the preliminary defect analysis report; The defect position calibration module determines the center point of the defect, calculates the defect size, records the position of each defect, and generates the defect position information according to the preliminary defect analysis report; The defect type judgment module performs feature matching on the defect image based on the defect position information, compares and analyzes with the known defect database, classifies each defect type, determines the defect category and severity, and generates the defect recognition result; The defect detail confirmation module analyzes the defect characteristics, evaluates the potential risk, calculates the possibility of defect development, and generates the defect feature confirmation record according to the defect recognition result.

[0021] The stainless steel pipe fitting surface optimization image includes optimized image contrast, optimized image brightness and surface feature details; the preliminary defect analysis report includes crack area features, corrosion area features, area boundary information and difference texture features; the defect position information includes defect center point coordinates, defect size parameters, defect morphology description and defect distribution position; the defect recognition result includes defect category classification and defect severity level; and the defect feature confirmation record includes defect characteristic description, potential risk evaluation result and defect development possibility prediction result.

[0022] Referring to Figure 2 And Figure 3 The surface image acquisition module comprises an image shooting sub-module, an image processing sub-module and an image optimization sub-module; The image capturing sub-module employs a camera to capture the surface image of the stainless steel pipe fitting, adjusts the focal length and exposure time of the camera to match different light conditions, and generates the original surface image; The image capturing sub-module employs a high-resolution camera system, carefully adjusts the focal length and exposure time, and adapts to changing light conditions to ensure the capture of high-quality original surface images. During this process, appropriate camera settings are selected based on the reflective properties of the stainless steel pipe fitting, such as adjusting the ISO sensitivity to adapt to low-light environments or modifying the shutter speed to reduce the risk of overexposure. By monitoring the intensity of ambient light in real-time, camera parameters are dynamically adjusted to optimize image capture effects. In addition, the automatic exposure lock function is applied to maintain consistent exposure settings and avoid fluctuations in image quality caused by changes in light, thereby obtaining clear and representative images of the stainless steel pipe fitting.

[0023] The image processing sub-module adjusts the contrast and brightness based on the original surface image, enhances image edges, and highlights the surface features of the stainless steel pipe fitting to obtain a feature-enhanced image; The image processing sub-module uses advanced image editing software to carefully adjust the contrast and brightness of the original surface image, and enhances key image edges to better highlight the surface features of the stainless steel pipe fitting. An edge enhancement algorithm is used to improve the clarity of edges in the image, and the image's depth and level of detail are enhanced by adjusting the grayscale threshold and contrast. At the same time, adaptive brightness enhancement technology is applied to ensure good visual effects on different display devices. Through these processing steps, the image's recognition is significantly improved, making small scratches or defects clearly visible in the image, facilitating subsequent quality analysis and evaluation.

[0024] The image optimization sub-module adjusts the detail level and color saturation based on the feature-enhanced image, optimizes the image quality, and obtains a surface-optimized image of the stainless steel pipe fitting; The image optimization sub-module further optimizes the feature-enhanced image by adjusting the detail level and color saturation to improve the overall visual effect of the image. The technician uses professional image processing tools to finely adjust the color saturation, making the image colors more vivid and natural. In addition, by enhancing the detail level of the image, the texture performance is improved, making the surface details of the stainless steel pipe fitting more precise and accurate. This stage of processing not only optimizes the image's aesthetics, but also improves its practicality for further analysis, making the image more attractive when displayed and more conducive to in-depth quality inspection and evaluation by professionals.

[0025] Please refer to Figure 2 and Figure 4 The surface texture analysis module includes a crack identification sub-module, a corrosion analysis sub-module, and a texture labeling sub-module; The crack identification submodule is based on the optimized image of the stainless steel pipe surface, searches and identifies the crack features in the image, subdivides the crack profile, adjusts the image parameters to distinguish the crack from the normal material, and obtains a crack feature map; In the crack identification submodule, advanced image analysis techniques are used to search and identify cracks from the optimized stainless steel pipe surface image. By using advanced image processing algorithms such as edge detection and texture analysis, the crack profile is subdivided and clearly distinguished from the undamaged material. It involves precise adjustment of image parameters such as contrast and brightness to better highlight the crack features, making the cracks more obvious in the image. In addition, the width and depth of the cracks are quantitatively evaluated to generate a detailed crack feature map, which serves as the basis for subsequent analysis and provides key data for evaluating the integrity of the structure.

[0026] The corrosion analysis submodule detects pixel intensity based on the crack feature map to identify corrosion areas, discriminates and analyzes the texture differences of the corrosion areas, and generates a corrosion area detail map; The corrosion analysis submodule identifies detailed corrosion areas based on the crack feature map. By analyzing the pixel intensity in the image, it identifies the corroded areas and discriminates the texture differences between normal and damaged materials using texture analysis techniques. By adjusting the threshold and parameters of image analysis, the accuracy and detail of corrosion identification are ensured, and a detailed corrosion area map is further generated to reveal the extent and scope of corrosion, providing a basis for repair work and future corrosion prevention measures, enhancing the safety of pipe use and extending its service life.

[0027] The texture labeling submodule automatically labels and draws texture feature boundaries based on the corrosion area detail map, distinguishes the boundaries of different texture features, and generates a preliminary defect analysis report; The texture labeling submodule focuses on automatic labeling from the corrosion area detail map and draws the precise boundaries of each texture feature. This process includes using machine learning techniques to distinguish different texture features and automatically label the location and range of various defects. Through precise image segmentation techniques, this module can accurately depict the boundaries of corrosion and other material defects, providing important information for developing repair measures and improving material design. Labeling helps generate a preliminary defect analysis report, which details the types and distribution of defects on the pipe surface, providing detailed reference materials for quality control and subsequent processing.

[0028] Please refer to Figure 2 and Figure 5 , the defect position calibration module includes a center point positioning submodule, a size calculation submodule, and a position recording submodule; The center point positioning submodule is based on the preliminary defect analysis report, performs edge detection on each defect, identifies the geometric center point, segments each defect area, and generates a defect center positioning map. The center point positioning sub-module, based on the preliminary defect analysis report, uses edge detection techniques to accurately identify the geometric center point of each defect. This process involves advanced image processing algorithms such as Canny edge detection or Sobel operator to clearly segment the outline of each defect area. By analyzing the geometric shape and boundaries of each defect, high-precision center point positioning is achieved. Subsequently, these data are converted into a defect center positioning map, which details the center positions of each defect, providing necessary visual references for subsequent size calculation and position labeling.

[0029] The size calculation sub-module, based on the defect center positioning map, performs image scale conversion to measure the length and width of each calibrated defect. By converting pixels to actual sizes, the actual size of the defect is calculated, resulting in defect size data. The size calculation sub-module measures the size of defects based on the defect center positioning map. This module applies image scale conversion techniques to convert pixel sizes in the image to actual physical sizes. By setting a scale and calibration factor, the length and width measurements of each calibrated defect are ensured to be accurate. Digital measurement tools are used to automatically calculate the size of each defect, such as through a conversion formula between pixel point spacing and actual length, to accurately obtain the actual size of the defect. The generated defect size data is not only accurate, but also facilitates subsequent quality evaluation and repair work.

[0030] The position recording sub-module uses defect size data to perform coordinate positioning, records and labels the actual position of each defect, and obtains defect position information. The position recording sub-module uses the generated defect size data to perform coordinate positioning of defects. Through accurate image analysis and coordinate conversion algorithms, this module records and labels the actual position of each defect, ensuring the accuracy and traceability of the position information. The process includes converting the image position of the defect into actual physical coordinates using GIS software or similar technology, and then generating detailed defect position information. The information will be used to create a defect map, detailing the position of each defect, providing precise work guidance for the team, optimizing maintenance strategies and resource allocation.

[0031] Please refer to Figure 2 and Figure 6 , the defect type judgment module includes a feature extraction sub-module, a matching analysis sub-module, and a severity assessment sub-module. The feature extraction sub-module, based on the defect position information, refines the visual features of each defect, extracts the shape and texture parameters of the defect, and outputs the defect feature vector. The feature extraction submodule utilizes detailed defect location information for in-depth analysis, refining and extracting the morphological and texture features of each defect. Through advanced image processing techniques, such as gray-level co-occurrence matrix or Fourier transform, it analyzes the texture and shape characteristics of the defect region, accurately capturing the unique visual parameters of each defect. During this process, machine learning algorithms are applied to automatically extract and encode these features, generating defect feature vectors. These vectors represent the main visual attributes of the defects, such as edge sharpness, texture directionality, and regional contrast, providing crucial input data for subsequent matching analysis.

[0032] The matching analysis submodule compares the defect feature vector with a database of known defect types to identify similar features, determine the category of each defect, and obtain the matching analysis results. The matching analysis submodule compares the generated defect feature vectors with a comprehensive database of known defect types. By applying pattern recognition and machine learning techniques, such as support vector machines or deep learning classifiers, the system identifies features similar to known defect samples, accurately determining the category of each defect. This analysis process not only improves the accuracy of identification but also accelerates processing speed, enabling real-time matching of defects detected on-site with historical data. This ensures that each defect can be quickly and accurately classified, ultimately generating detailed matching analysis results.

[0033] The severity assessment submodule analyzes the potential impact of each defect based on the matching analysis results, calculates the severity score of the defect area on structural stability, and generates defect identification results. Severity scoring, according to the formula: ; Calculations are performed to generate a severity assessment result for the defect, whereby... Represents severity score, Representing the The area of ​​the defect, Representing the The depth of the defect This represents the total number of defects identified.

[0034] Assuming the number of defects detected The specific defect area and depth data are as follows:

[0035]

[0036]

[0037]

[0038] Calculate severity score :

[0039] The results show that the weighted average of the defect area and depth reflects the severity of the impact on the structure, and the higher the value, the greater the impact of the defect on the stability of the structure.

[0040] See Figure 2 and Figure 7 , the defect detail confirmation module includes a characteristic depth analysis submodule, a risk level assessment submodule, and a development trend prediction submodule; The characteristic depth analysis submodule analyzes the morphological characteristics of each defect based on the defect identification results, reveals the geometric structure and surface characteristics of the defect, and generates a characteristic analysis report; The characteristic depth analysis submodule comprehensively utilizes the defect identification results to conduct in-depth analysis of the morphological characteristics of each defect. Through advanced image analysis techniques such as three-dimensional modeling and surface scanning, the geometric structure and surface characteristics of each defect are revealed in detail, the volume, depth, and surface roughness of the defect are calculated, and the physical appearance and internal structure of the defect are depicted. Based on detailed analysis, a characteristic analysis report containing comprehensive defect description is generated, providing a scientific basis for subsequent risk assessment and maintenance strategy formulation.

[0041] The risk level assessment submodule uses the characteristic analysis report to analyze the impact of the defect on the mechanical strength and functional stability of the stainless steel pipe, determines the risk level of the defect, and obtains the risk assessment result; The risk level assessment submodule uses the characteristic analysis report as the basis for detailed analysis of the potential impact of each defect on the mechanical strength and functional stability of the stainless steel pipe. By comprehensively evaluating the size, location, and morphological characteristics of the defect, a risk assessment model is used to determine the risk level of each defect. The evaluation process includes quantitative analysis of stress concentration and fatigue life reduction caused by the defect, and the risk assessment result of each defect is obtained to guide the maintenance team to prioritize high-risk defects to ensure the safety and reliability of the pipe.

[0042] The development trend prediction submodule calculates the development score of the defect under future operating conditions based on the risk assessment result, analyzes the expansion speed of the defect, and generates a defect feature confirmation record; The development score is calculated according to the formula: ; wherein, represents the probability value, represents the absolute value of the defect depth predicted by the th decision tree, represents the product of the operating conditions and environmental factors corresponding to the th decision tree under the square root, Number of total decision trees in the representative model.

[0043] Absolute value of defect depth predicted by the i-th decision tree. Quantifies the severity of the defect, obtained through real-time monitoring of the defect and historical data analysis. Assume that the analysis results of 10 decision trees show depth values of 2.1, 3.2, 1.8, 2.9, 2.5, 3.3, 1.9, 2.2, 3.1, and 2.0 cm, respectively.

[0044] Represents the complexity of the operating conditions corresponding to the i-th decision tree, obtained by evaluating the operating environment through real-time sensor data.

[0045] Represents environmental factors such as humidity or chemical corrosion, obtained through relevant sensors and numerical processing. The specific numerical standard is set according to the environmental safety evaluation standard.

[0046] Number of total decision trees in the model, which is 10 in this example. and Quantified by real-time data obtained through sensors, under certain conditions, is the operating temperature (in Celsius), is the humidity percentage, with specific values of (80°C, 70%), (75°C, 65%), (82°C, 68%), (78°C, 72%), (80°C, 66%), (76°C, 70%), (85°C, 65%), (82°C, 68%), (77°C, 72%), and (79°C, 67%).

[0047] Calculate the value of each For the first decision tree, , :

[0048] Repeat this calculation method to obtain the value of each decision tree Multiply the defect depth of each decision tree and accumulate the product:

[0049] Divide the accumulated result by the total number of decision trees :

[0050] The results show that, based on the actual monitoring and calculation data, the comprehensive score of predicting the defect development probability is 183.28, reflecting the comprehensive possibility of the defect development under future operating conditions, and a higher score indicates a higher potential risk, which requires further monitoring and possible maintenance measures.

[0051] Referring to Figure 8 An intelligent stainless steel pipe defect detection method, comprising the following steps: S1: capturing the surface image of the stainless steel pipe by using a camera, adjusting the focal length and exposure time of the camera to match the different light conditions, adjusting the contrast, brightness and color saturation to obtain the optimized surface image of the stainless steel pipe; S2: based on the optimized surface image of the stainless steel pipe, searching and identifying the crack features in the image, identifying the corrosion area, automatically labeling and drawing the boundary distinguishing the different texture features, and generating a preliminary defect analysis report; S3: based on the preliminary defect analysis report, performing edge detection on each defect, identifying the geometric center point, segmenting each defect area, measuring the length and width of each calibrated defect, calculating the actual size of the defect, and obtaining the defect size data; S4: using the defect size data, recording and labeling the actual position of each defect, refining the visual features of each defect, extracting the shape and texture parameters of the defect, and outputting the defect feature vector; S5: comparing the defect feature vector with the known defect type database to determine the category of each defect, calculating the severity score of the defect area to the structural stability, and generating a defect recognition result; S6: based on the defect recognition result, analyzing the morphological features of each defect, determining the risk level of the defect, calculating the development score of the defect under future operating conditions, and generating a defect feature confirmation record.

[0052] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made in accordance with the technical essence of the present application to the above embodiments shall fall within the protection scope of the present application.

Claims

1. An intelligent stainless steel pipe fitting defect detection system, characterized in that: The system includes: The surface image acquisition module captures surface images of stainless steel pipe fittings using a camera, adjusts image contrast and brightness to highlight surface features of stainless steel pipe fittings, adjusts image detail visibility, and generates optimized surface images of stainless steel pipe fittings. The surface texture analysis module analyzes crack and corrosion areas based on the optimized surface image of the stainless steel pipe fitting, extracts texture differences within crack and corrosion areas, marks area boundaries, distinguishes textured areas, and generates a preliminary defect analysis report. Based on the preliminary defect analysis report, the defect location calibration module determines the center point of the defect, calculates the defect size, records the location of each defect, and generates defect location information. Based on the defect location information, the defect type determination module performs feature matching on the defect image, compares and analyzes it with a known defect database, classifies each defect type, determines the defect category and severity, and generates defect identification results. Based on the defect identification results, the defect details confirmation module analyzes the defect characteristics, assesses potential risks, calculates the probability of defect development, and generates a defect feature confirmation record.

2. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The optimized surface image of the stainless steel pipe fitting includes optimized image contrast, optimized image brightness, and surface feature details; the preliminary defect analysis report includes crack area features, corrosion area features, area boundary information, and differential texture features; The defect location information includes the coordinates of the defect center point, defect size parameters, defect morphology description, and defect distribution location; the defect identification results include defect category classification and defect severity level; the defect feature confirmation record includes defect characteristic description, potential risk assessment results, and defect development probability prediction results.

3. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The surface image acquisition module includes an image capture submodule, an image processing submodule, and an image optimization submodule; The image capture submodule uses a camera to capture surface images of stainless steel pipe fittings, adjusts the camera's focal length and exposure time to match different lighting conditions, and generates the original surface image; The image processing submodule adjusts the contrast and brightness based on the original surface image, strengthens the image edges, and highlights the surface features of the stainless steel pipe to obtain a feature-enhanced image. The image optimization submodule enhances the image based on the features, adjusts the detail level of the image, adjusts the color saturation, and optimizes the image texture to obtain an optimized image of the stainless steel pipe surface.

4. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The surface texture analysis module includes a crack identification submodule, a corrosion analysis submodule, and a texture annotation submodule; The crack recognition submodule searches for and identifies crack features in the image based on the optimized image of the stainless steel pipe fitting, refines the crack outline, adjusts the image parameters to distinguish between cracks and normal material, and obtains a crack feature map. The corrosion analysis submodule performs pixel intensity detection to identify the corrosion area based on the crack feature map, judges and analyzes the texture differences of the corrosion area, and generates a detailed map of the corrosion area. The texture annotation submodule automatically annotates and draws texture feature boundaries based on the detailed map of the eroded area, distinguishes the boundaries of different texture features, and generates a preliminary defect analysis report.

5. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The defect location calibration module includes a center point positioning submodule, a size calculation submodule, and a location recording submodule; Based on the preliminary defect analysis report, the center point localization submodule performs edge detection on each defect, identifies the geometric center point, segments each defect region, and generates a defect center localization map. The size calculation submodule performs image scale conversion based on the defect center location map, measures the length and width of each calibrated defect, and calculates the actual size of the defect by converting pixels to actual size, thus obtaining the defect size data. The location recording submodule uses the defect size data to perform coordinate positioning, records and marks the actual location of each defect, and obtains defect location information.

6. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The defect type determination module includes a feature extraction submodule, a matching analysis submodule, and a severity assessment submodule. Based on the defect location information, the feature extraction submodule refines the visual features of each defect, extracts the morphological and texture parameters of the defect, and outputs the defect feature vector. The matching analysis submodule compares the defect feature vector with a database of known defect types to identify similar features, determine the category of each defect, and obtain the matching analysis results. Based on the matching analysis results, the severity assessment submodule analyzes the potential impact of each defect, calculates the severity score of the defect area on structural stability, and generates defect identification results.

7. The intelligent stainless steel pipe fitting defect detection system according to claim 6, characterized in that: The severity score is calculated according to the formula: ； Calculations are performed to generate a severity assessment result for the defect, whereby... Represents severity score, Representing the The area of ​​the defect, Representing the The depth of the defect This represents the total number of defects identified.

8. The intelligent stainless steel pipe fitting defect detection system according to claim 1, characterized in that: The defect details confirmation module includes a characteristic in-depth analysis submodule, a risk level assessment submodule, and a development trend prediction submodule. Based on the defect identification results, the feature depth analysis submodule analyzes the morphological characteristics of each defect, reveals the geometric structure and surface characteristics of the defect, and generates a feature analysis report; The risk level assessment submodule uses the aforementioned characteristic analysis report to analyze the impact of defects on the mechanical strength and functional stability of stainless steel pipe fittings, determine the risk level of the defects, and obtain the risk assessment results. Based on the risk assessment results, the development trend prediction submodule calculates the development score of the defect under future operating conditions, analyzes the expansion rate of the defect, and generates a defect feature confirmation record.

9. The intelligent stainless steel pipe fitting defect detection system according to claim 8, characterized in that: The development score is calculated according to the formula: ； Calculations are performed, in which, Represents probability value, Representing the The absolute value of the defect depth predicted by each decision tree. Represents the square root of the first 100 BC. The product of the operational conditions and environmental factors corresponding to each decision tree. This represents the total number of decision trees in the model.

10. An intelligent method for detecting defects in stainless steel pipe fittings, characterized in that, The system according to any one of claims 1-9 is executed, comprising the following steps: The surface image of stainless steel pipe fittings is captured by a camera. The camera's focal length and exposure time are adjusted to match different lighting conditions. The contrast, brightness, and color saturation are adjusted to obtain an optimized image of the stainless steel pipe fitting surface. Based on the optimized surface image of the stainless steel pipe fitting, the crack features in the image are searched and identified, the corrosion area is identified, the boundaries that distinguish the different texture features are automatically marked and drawn, and a preliminary defect analysis report is generated. Based on the preliminary defect analysis report, edge detection is performed on each defect to identify the geometric center point, each defect region is segmented, the length and width of each calibrated defect are measured, the actual size of the defect is calculated, and the defect size data is obtained. Using the defect size data, the actual location of each defect is recorded and labeled, the visual features of each defect are refined, the shape and texture parameters of the defects are extracted, and the defect feature vector is output. By comparing the defect feature vectors with a database of known defect types, the category of each defect is determined, the severity score of the defect area on structural stability is calculated, and defect identification results are generated. Based on the defect identification results, the morphological characteristics of each defect are analyzed, the risk level of the defect is determined, the development score of the defect under future operating conditions is calculated, and a defect feature confirmation record is generated.

Images