A laser-excited infrared weld internal defect detection system and method

Abstract

This invention discloses a laser-excited infrared weld internal defect detection system and method, belonging to the field of weld non-destructive testing and infrared thermal imaging detection technology. The system consists of a laser thermal excitation module, an infrared thermal imaging acquisition module, a host computer (PC), an image preprocessing module, a defect detection algorithm module, and a detection result display and storage module. The laser thermal excitation module includes a fiber-coupled laser, a transmission fiber, and an end collimator. The laser generated by the fiber-coupled laser is transmitted through the transmission fiber to the end collimator and irradiates the weld area to be inspected to form local thermal excitation. The infrared thermal imaging acquisition module acquires infrared thermal images of the weld area during the laser heating and cooling stages and transmits them to the host computer (PC). The image preprocessing module performs region extraction and format conversion on the infrared thermal images. The defect detection algorithm module identifies and locates internal weld defects based on a multi-scale edge perception network (MEDB). The MEDB module extracts multi-scale edge features, the MSSA module enhances the defect area response and suppresses complex thermal background interference, and the EUCB module enhances the feature representation capability during the upsampling stage. The detection result and storage module outputs and saves the defect type, location, confidence level, and corresponding infrared image. This invention can improve the detection accuracy of internal defects in welds under complex thermal conditions and is applicable to non-contact detection of defects such as flat-bottomed holes, cracks, internal grooves, and lack of fusion.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing and infrared thermal imaging detection technology for welds. Specifically, it relates to a laser-excited infrared weld internal defect detection system and method, and more particularly to a detection system and method that uses laser thermal excitation to acquire infrared thermal response images of welds and combines them with a multi-scale edge sensing network to realize the identification and location of internal defects in welds.

[0002] Welded structures are widely used in pressure vessels, pipelines, rail transportation, ships, engineering machinery, and special equipment. The quality of the weld directly affects the load-bearing capacity, sealing performance, and service safety of structural components. Due to the influence of process parameters, material condition, heat input, and cooling conditions during welding, defects such as cracks, lack of fusion, porosity, and internal grooves are prone to occur inside the weld. If these defects are not detected in time, they may further expand during service, leading to structural failure or even safety accidents.

[0003] Currently, commonly used methods for detecting internal defects in welds include radiographic testing, ultrasonic testing, magnetic particle testing, penetrant testing, and eddy current testing. Radiographic testing provides intuitive imaging, but suffers from high radiation protection requirements, high testing costs, and inconvenient on-site deployment. Ultrasonic testing is highly dependent on operator experience and coupling conditions, and its stability is insufficient for detecting welds with complex morphologies. Magnetic particle and penetrant testing are mainly suitable for surface or near-surface defects and cannot meet the needs of internal defect detection. Therefore, there is an urgent need for a non-contact, intuitive imaging method suitable for rapid on-site detection of internal weld defects.

[0004] Infrared thermal imaging nondestructive testing reflects internal structural differences by acquiring temperature field changes on the surface of the object under test. It has advantages such as being non-contact, having a fast detection speed, and providing visualized results. Laser-excited infrared thermal imaging uses a laser as an active thermal excitation source, which can apply controllable heat input to local areas of the weld. When defects exist inside the weld, there are differences in the heat diffusion path and thermal response characteristics between the defect area and the normal area. Infrared thermal imagers can capture this temperature field anomaly, thus providing a basis for the detection of internal defects in the weld.

[0005] However, in practical laser-excited infrared weld inspection, the weld surface morphology is complex, thermal diffusion weakens the thermal response of defects, and infrared images may also contain background noise, blurred boundaries, indistinct small target defects, and similar thermal responses among different defects. Traditional image processing methods struggle to reliably extract these weak defect features, and general target detection algorithms are prone to missed detections and false detections in the detection of small-scale cracks, lack of fusion, and other weak boundary defects. Therefore, a weld internal defect detection system that combines a laser-excited infrared detection platform with a dedicated defect recognition algorithm is needed to improve detection accuracy and engineering applicability under complex thermal backgrounds. Summary of the Invention

[0006] To address the problems of low detection efficiency, insufficient non-contact detection capability, blurred defect boundaries in infrared thermal images, difficulty in identifying small target defects, and severe interference from complex thermal backgrounds in existing weld internal defect detection methods, this invention proposes a laser-excited infrared weld internal defect detection system and method. This system uses a laser thermal excitation module to locally thermally excite the weld area, acquires infrared thermal images of the weld area during the heating and cooling stages using an infrared thermal imaging acquisition module, and utilizes a defect detection algorithm module based on a multi-scale edge perception network to identify and locate internal weld defects.

[0007] To achieve the above objectives, this invention discloses a laser-excited infrared weld internal defect detection system, which consists of a laser thermal excitation module (1), an infrared thermal imaging acquisition module (2), a host computer PC (3), an image preprocessing module (4), a defect detection algorithm module (5), and a detection result display and storage module (6).

[0008] The laser thermal excitation module (1) includes a fiber-coupled laser (11), a transmission fiber (12), and an end collimator (13). The fiber-coupled laser (11) generates a laser thermal excitation signal, which is transmitted to the end collimator (13) via the transmission fiber (12). The end collimator (13) collimates and shapes the laser beam before irradiating the weld area to be inspected, thereby forming a local thermal excitation area on the weld surface.

[0009] The infrared thermal imaging acquisition module (2) includes an infrared thermal imager (21) for acquiring infrared thermal images of the weld area during the laser heating and cooling stages. The host computer (PC) (3) is used to receive infrared thermal images (32), call the defect detection algorithm (33), and display the detection results on the system interface (34).

[0010] The image preprocessing module (4) is used to extract the region of interest, enhance the image, and convert the format of the infrared thermal image. The defect detection algorithm module (5) adopts a weld defect detection model based on a multi-scale edge perception network, including a MEDB module (51), an MSSA module (52), and an EUCB module (53), for identifying and locating internal defects in the weld. The detection result display and storage module (6) is used to display and store the defect category, defect location, detection confidence level, and corresponding infrared image.

[0011] This invention also discloses a laser-excited infrared weld internal defect detection method, based on the above system, including the following steps: After the system is powered on, the host computer PC (3) initializes the infrared thermal imaging acquisition module (2), image preprocessing module (4), defect detection algorithm module (5), and detection result display and storage module (6); after initialization, the laser power, excitation time, excitation distance, infrared acquisition frame rate, and acquisition time are set; the fiber-coupled laser (11) generates a laser thermal excitation signal, which is irradiated onto the weld area to be detected through the transmission fiber (12) and the end collimator (13); the infrared thermal imager (21) and the laser thermal imager (21) are used to detect the internal defects of the weld. The infrared thermal image of the weld area is acquired and transmitted to the host computer PC (3); the image preprocessing module (4) processes the infrared thermal image and inputs it into the defect detection algorithm module (5); the defect detection algorithm module (5) extracts multi-scale edge features through the MEDB module (51), enhances the response of the defect area and suppresses complex thermal background interference through the MSSA module (52), and enhances the high-resolution feature expression after upsampling through the EUCB module (53), and finally outputs the category, location and confidence of the internal defect of the weld; the detection result display and storage module (6) displays the detection result on the system interface (34) and stores it.

[0012] The fiber-coupled laser (11) is an 808nm high-power fiber-coupled diode laser system that supports power adjustment, temperature control, and TTL modulation. Its model number is FC-W-808H. The end collimator (13) shapes the laser into a 40mm × 40mm planar spot.

[0013] The infrared thermal imager (21) is a network-type temperature-measuring infrared thermal imager, model IPT 640M, with a resolution of 640×512, a response band of 8~14μm, and a frame rate of 25Hz. It communicates with the host computer PC (3) via an Ethernet interface.

[0014] The internal defects of the weld include one or more of the following: flat-bottomed hole defects, crack defects, internal groove defects, and lack of fusion defects. The defect-marked areas in the infrared thermal image are the main thermal anomaly response areas visible in the image.

[0015] The MEDB module (51) includes a multi-scale convolution branch, an asymmetric convolution branch, a Sobel edge-aware branch, and a QKV channel attention fusion branch. Among them, the multi-scale convolution branch is used to extract defect features at different scales; the asymmetric convolution branch is used to extract directional features of slender defects; the Sobel edge-aware branch is used to enhance the contour response of defect edges; and the QKV channel attention fusion branch is used to adaptively fuse multi-source features.

[0016] The MSSA module (52) includes a directional pooling branch and a selective state space modeling branch. The directional pooling branch is used to extract the axial and transverse structural features of the weld; the selective state space modeling branch is used to establish long-range context relationships and suppress complex thermal background interference.

[0017] The EUCB module (53) is set on the feature fusion network of the weld defect detection model in the sampling stage. It is used to replace conventional interpolation upsampling and improves the positioning accuracy of small-scale defects and weak boundary defects through scale recovery, depth convolution feature enhancement, batch normalization, nonlinear activation and channel mapping operations.

[0018] Compared with existing technologies, the advantages of this invention are as follows: First, this invention combines active laser thermal excitation with infrared thermal imaging to achieve non-contact detection of internal defects in welds; Second, this invention achieves locally controllable thermal excitation through fiber-coupled lasers and end collimators, which is beneficial to enhancing the thermal response of internal weld defects; Third, this invention employs a multi-scale edge sensing network based on MEDB, MSSA, and EUCB, which can enhance the feature representation of small-scale defects, slender cracks, weak boundary defects, and defect regions under complex thermal backgrounds; Fourth, this invention can output defect category, location, and confidence level, and save the corresponding infrared images and detection results, facilitating subsequent review and quality assessment. Attached Figure Description

[0019] Figure 1 This is a physical image of a laser-excited infrared weld internal defect detection system according to the present invention.

[0020] Figure 2 This is a schematic diagram of the principle of laser-excited infrared weld internal defect detection according to the present invention.

[0021] Figure 3 This is a schematic diagram of the overall structure of the defect detection algorithm module of the present invention.

[0022] Figure 4 This is a schematic diagram of the MEDB module structure of the present invention.

[0023] Figure 5 This is a schematic diagram of the MSSA module structure of the present invention.

[0024] Figure 6 This is a schematic diagram of the defect detection results of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] like Figure 1 As shown, the laser-excited infrared weld internal defect detection system of the present invention consists of a laser thermal excitation module (1), an infrared thermal imaging acquisition module (2), a host computer PC (3), an image preprocessing module (4), a defect detection algorithm module (5), and a detection result display and storage module (6). The laser thermal excitation module (1) is arranged on one side of the weld to be inspected area, the infrared thermal imaging acquisition module (2) is aligned with the weld to be inspected area, and the host computer PC (3) is communicatively connected to the infrared thermal imager (21) and is used to run the image preprocessing module (4), the defect detection algorithm module (5), and the detection result display and storage module (6).

[0027] The laser thermal excitation module (1) includes a fiber-coupled laser (11), a transmission fiber (12), and an end collimator (13). Before inspection, the end collimator (13) is adjusted to a suitable position so that its output spot covers the weld area to be inspected. After the fiber-coupled laser (11) generates a laser thermal excitation signal, the laser is transmitted to the end collimator (13) via the transmission fiber (12), and then collimated and shaped by the end collimator (13) before irradiating the weld surface. Because the heat conduction characteristics of the defect area inside the weld are different from those of the normal area, the defect area will show an abnormal thermal response in the infrared thermal image during the laser heating and cooling process.

[0028] The infrared thermal imaging acquisition module (2) includes an infrared thermal imager (21). The infrared thermal imager (21) continuously acquires infrared thermal images of the weld area during the laser heating and cooling stages, and transmits the acquired infrared thermal images to the host computer PC (3). In a preferred embodiment, the infrared thermal imager (21) is a network-type temperature-measuring infrared thermal imager with an infrared resolution of 640×512 and an acquisition frame rate of 25Hz, and communicates with the host computer PC (3) through an Ethernet interface.

[0029] like Figure 2As shown, the detection principle of this invention is as follows: a local heat input is applied to the weld area to be detected by a laser thermal excitation module (1), and the surface temperature field response of the weld is collected by an infrared thermal imager (21). Then, the upper computer PC (3) performs image preprocessing and intelligent recognition on the infrared thermal image. When there are flat-bottomed holes, cracks, inner grooves or non-fusion defects inside the weld, the defect area will affect the heat flow diffusion path, causing a thermal anomaly response in the infrared thermal image that is different from that of the normal weld area. The defect detection algorithm module (5) identifies the thermal anomaly area and outputs the defect type, location and confidence level.

[0030] When the system is running, the host computer PC (3) is started first, and the laser thermal excitation module (1), infrared thermal imaging acquisition module (2), image preprocessing module (4), defect detection algorithm module (5), and detection result display and storage module (6) are initialized. After initialization, the laser power, excitation time, excitation distance, infrared acquisition frame rate, and acquisition time are set according to the weld material, defect depth, detection distance, and thermal diffusion characteristics. Then, the fiber-coupled laser (11) is started to generate a laser thermal excitation signal, which is then irradiated onto the weld area to be inspected through the transmission fiber (12) and the end collimator (13).

[0031] During the laser heating stage and the cooling stage after the laser stops, the infrared thermal imager (21) continuously acquires infrared thermal images of the weld area. After the acquired infrared thermal images are transmitted to the host computer PC (3), the image preprocessing module (4) performs region of interest extraction, image enhancement, and format conversion. The region of interest is preferably the weld and its surrounding thermal anomaly response area. In the case of multiple adjacent defects, each defect and its surrounding area can be cropped separately to reduce thermal diffusion interference between adjacent defects.

[0032] The overall structure of the algorithm is as follows Figure 3 As shown. The preprocessed infrared thermal image is input into the defect detection algorithm module (5). The defect detection algorithm module (5) adopts a weld defect detection model based on a multi-scale edge perception network. This model is based on a target detection network, and introduces the MEDB module (51), MSSA module (52), and EUCB module (53) in the feature extraction and feature fusion stages, respectively.

[0033] The MEDB module (51) is used to enhance the multi-scale morphological features and edge contour features of weld defects. The module structure is as follows: Figure 4As shown in the diagram. Specifically, the multi-scale convolution branch is used to extract local details and contextual information under a large receptive field simultaneously; the asymmetric convolution branch is used to extract the directional features of slender or strip-shaped defects such as cracks and lack of fusion; the Sobel edge-aware branch is used to enhance the edge response of the defect region; and the QKV channel attention fusion branch is used to adaptively weight and fuse the features of multiple branches, thereby improving the model's ability to represent weak boundary defects.

[0034] The MSSA module (52) is used to enhance the response of defect areas and suppress interference from complex thermal backgrounds. The module structure is as follows: Figure 5 As shown, the directional pooling branch aggregates contextual information along the weld seam direction, across the weld seam, and in the global spatial range, enabling the model to perceive the directional structural features of the weld seam; the selective state space modeling branch unfolds the two-dimensional feature map into sequential features, establishing long-distance contextual relationships, enabling the model to dynamically focus on defect-related regions in complex thermal backgrounds.

[0035] The EUCB module (53) is set on the feature fusion network of the weld defect detection model during the sampling stage. Compared with conventional interpolation upsampling, the EUCB module further performs deep convolution feature enhancement, batch normalization, nonlinear activation and channel mapping operations during the scale recovery process, thereby improving the spatial representation ability of the feature map after upsampling and improving the localization effect of small-scale defects and weak boundary defects.

[0036] After the detection is completed, the detection result display and storage module (6) overlays the defect category, defect location and detection confidence level on the infrared image in the system interface (34) and saves the detection results simultaneously. Figure 6 This diagram illustrates the results of the detection of internal defects in welds according to the present invention.

[0037] This invention discloses a laser-excited infrared weld internal defect detection system and method, which can be used to detect flat-bottomed hole defects, crack defects, internal groove defects, and lack of fusion defects inside welds. This method utilizes active laser thermal excitation to enhance the thermal response of defects and combines it with a multi-scale edge sensing network to improve the accuracy of weld defect detection under complex thermal backgrounds. It can reduce reliance on manual interpretation and improve the efficiency of non-destructive testing of welds.

[0038] The above description is merely a preferred embodiment of the present invention, demonstrating its main functional features and innovative advantages, and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A laser-excited infrared weld internal defect detection system, characterized in that: It consists of a laser thermal excitation module (1), an infrared thermal imaging acquisition module (2), a host computer (PC) (3), an image preprocessing module (4), a defect detection algorithm module (5), and a detection result display and storage module (6). The laser thermal excitation module (1) includes a fiber-coupled laser (11), a transmission fiber (12), and an end collimator (13). The fiber-coupled laser (11) generates a laser thermal excitation signal, which is transmitted to the end collimator (13) via the transmission fiber (12). The end collimator (13) collimates and shapes the laser beam before irradiating the weld area to be inspected. The infrared thermal imaging acquisition module (2) includes an infrared thermal imager (21) used to acquire the weld area under laser thermal excitation signal. Infrared thermal images of the heating and cooling stages; the host computer PC (3) is used to receive infrared thermal images (32), call the defect detection algorithm (33), and display the detection results on the system interface (34); the image preprocessing module (4) is used to extract the region of interest, enhance the image, and convert the format of the infrared thermal images; the defect detection algorithm module (5) adopts a weld defect detection model based on a multi-scale edge perception network, including the MEDB module (51), MSSA module (52), and EUCB module (53), which is used to identify and locate internal defects in the weld; the detection result display and storage module (6) is used to display and store the defect category, defect location, detection confidence level, and corresponding infrared image.

2. A method for detecting internal defects in laser-excited infrared welds, characterized in that: The laser-excited infrared weld internal defect detection system according to claim 1 includes the following steps: After the system is powered on, the host computer PC (3) initializes the infrared thermal imaging acquisition module (2), image preprocessing module (4), defect detection algorithm module (5), and detection result display and storage module (6); after initialization, the laser power, excitation time, excitation distance, infrared acquisition frame rate, and acquisition time are set; the fiber-coupled laser (11) generates a laser thermal excitation signal, which is irradiated onto the weld area to be detected through the transmission fiber (12) and the end collimator (13); the infrared thermal imager (21) acquires the signal synchronously. The infrared thermal image of the weld area is transmitted to the host computer PC (3); the image preprocessing module (4) processes the infrared thermal image and inputs it into the defect detection algorithm module (5); the defect detection algorithm module (5) extracts multi-scale edge features through the MEDB module (51), enhances the response of the defect area and suppresses complex thermal background interference through the MSSA module (52), and enhances the high-resolution feature expression after upsampling through the EUCB module (53), and finally outputs the category, location and confidence of the internal defects of the weld; the detection result display and storage module (6) displays the detection results on the system interface (34) and stores them.

3. The laser-excited infrared weld internal defect detection system according to claim 1, characterized in that: The fiber-coupled laser (11) adopts an 808nm high-power fiber-coupled diode laser system with an output wavelength of 808±10nm. It supports power adjustment, temperature control, and TTL modulation. The model number is FC-W-808H. The end collimator (13) shapes the laser into a 40mm×40mm planar spot with a spot uniformity of better than 95%.

4. The laser-excited infrared weld internal defect detection system according to claim 1, characterized in that: The infrared thermal imager (21) is a network-type temperature measurement infrared thermal imager, model Guide IPT640M, with an infrared resolution of 640×512, a response band of 8~14μm, and a frame rate of 25Hz. It communicates with the host computer PC (3) via an Ethernet interface.

5. The laser-excited infrared weld internal defect detection method according to claim 2, characterized in that: The internal defects of the weld include one or more of the following: flat-bottom hole defects, crack defects, internal groove defects, and non-fusion defects. The defect-marked areas in the infrared thermal image are the main thermal anomaly response areas visible in the image.

6. The laser-excited infrared weld internal defect detection method according to claim 2, characterized in that: The MEDB module (51) includes a multi-scale convolution branch, an asymmetric convolution branch, a Sobel edge-aware branch, and a QKV channel attention fusion branch. The multi-scale convolution branch is used to extract defect features at different scales, the asymmetric convolution branch is used to extract directional features of slender defects, and the Sobel edge-aware branch is used to enhance the response of defect edge contours.

7. The laser-excited infrared weld internal defect detection method according to claim 2, characterized in that: The MSSA module (52) includes a directional pooling branch and a selective state space modeling branch. The directional pooling branch is used to extract the directional and transverse structural features of the weld, while the selective state space modeling branch is used to establish long-range context relationships and suppress complex thermal background interference.

Images