High-speed enameled wire surface defect detection system and method

Abstract

The invention relates to the technical field of industrial visual inspection, in particular to a high-speed enameled wire surface defect detection system and method. The real-time diameter measuring module adopts a laser diameter measuring instrument to measure the diameter of the enameled wire in real time, and the precision reaches + / -0.001 mm; the high-voltage pinhole detection module is used for verifying electrical defects; and a control processing module. Optimized combination of a high-speed line-scan digital camera and an optimized Yolo11 model is realized by innovatively fusing visual detection, triple technologies of optical measurement and physical detection are realized, full-dimensional quality verification and multispectral imaging are combined with the optimized Yolo11 model to improve the detection rate of 0.1 mm-level pinholes and scratches, and meanwhile, a laser diameter measuring instrument is integrated to feed back the line diameter in real time, so that the detection rate of the 0.1 mm-level pinholes and scratches is improved. In addition, high-voltage pinhole electrical defect signals are aligned with visual results in a space-time mode, and the problem of high-speed misinformation is solved.

Description

Technical Field

[0001] This invention relates to the field of industrial visual inspection technology, specifically to a high-speed enameled wire surface defect detection system and method. Background Technology

[0002] In the field of enameled wire manufacturing, online surface defect detection has long faced the following technical bottlenecks: Traditional visual inspection uses area scan cameras, where frame rate and resolution are directly related to price, with high-resolution and high-frame-rate area scan cameras being prohibitively expensive. However, when inspecting enameled wire with low frame rate and low resolution area scan cameras, at line speeds >30m / min, single-frame motion blur >0.1mm, resulting in a missed detection rate of over 15% for pinholes with a diameter of 0.1mm. Meanwhile, existing high-speed solutions barely support speeds of 60m / min by increasing illumination intensity, but the edge imaging distortion rate for wires with diameters greater than 5mm reaches as high as 30%. Furthermore, the industry commonly uses small-aperture optical lenses, with a depth of field range of only ±1mm. When inspecting special enameled wires with diameters of 10-20mm: the wire curvature causes edge defocusing, and enamel layer reflection interferes with defect identification. Existing technologies attempt multi-camera stitching; however, at speeds of 200m / min, the synchronization error is >10ms, failing to meet the requirements for continuous inspection. While high-voltage pinhole detection can capture electrical defect signals, the electrical signals and visual defects are spatially and temporally misaligned in high-speed scenarios. Furthermore, it cannot distinguish between real pinholes and dust interference, leading to a high false alarm rate. Therefore, we propose a high-speed enameled wire surface defect detection system and method. Summary of the Invention

[0003] The purpose of this invention is to provide a high-speed enameled wire surface defect detection system and method, which solves the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A high-speed enameled wire surface defect detection system includes an optical imaging module, which comprises a line array camera with a line frequency ≥80kHz, an electric zoom lens, and a multispectral combined light source. The multispectral combined light source supports switching between 450nm blue light, 630nm red light, and 850nm near-infrared light modes and is equipped with a polarization module. The real-time diameter measurement module uses a laser diameter measuring instrument to measure the diameter of the enameled wire in real time with an accuracy of ±0.001mm. High-voltage pinhole detection module, used for verifying electrical defects; The control processing module is configured to perform the following operations: Speed ​​V is detected by dynamic matching of wire diameter D. max ; Synchronous visual imaging, high-pressure pinhole detection module signal and diameter measurement data, with a time deviation of ≤±5ms; Run the optimized YOLO11 model for defect identification.

[0005] Preferably, the line frequency of the line scan camera is dynamically adjusted according to the line diameter: When the wire diameter D≤1.0mm, the line frequency ≥400kHz; When the wire diameter is 1.0mm < D ≤ 5.0mm, 400kHz > line frequency ≥ 160kHz; When the wire diameter is 5.0mm < D ≤ 10.0mm, 160kHz > line frequency ≥ 110kHz; When the wire diameter is 10.0mm < D ≤ 20.0mm, the line frequency is 110kHz > 80kHz.

[0006] Preferably, the operating modes of the multispectral combined light source include: for enameled wires with a diameter ≤ 5 mm, using a combination of 450 nm blue light and 630 nm red light for illumination; and for enameled wires with a diameter > 5 mm, using a combination of 850 nm near-infrared light and a polarization module for illumination.

[0007] Preferably, the optimized YOLO11 model includes: The C3k2 feature interaction module enables cross-layer feature fusion. C2PSA dual attention mechanism; Depthwise separable convolutional layers have a computational cost that is 1 / 8 that of standard convolutions; Multi-scale SPPF pooling layers execute 5×5, 9×9, and 13×13 pooling kernel operations in parallel.

[0008] Preferably, the optimized YOLO11 model has a detection rate of ≥99.1% for pinholes with a diameter of 0.1mm and an inference delay of ≤2ms.

[0009] Preferably, it also includes a defect verification module, which is configured to: when the vision system detects a surface defect, trigger the high-voltage pinhole detection module to perform electrical signal verification within a ±5ms time window; if the high-voltage pinhole detection module verifies successfully, it is determined to be a valid defect; if the high-voltage pinhole detection module fails to verify, it is otherwise marked as a false alarm.

[0010] A method for detecting surface defects in high-speed enameled wire, used in a high-speed enameled wire surface defect detection system as described in any one of claims 1-6, characterized by comprising the following steps: S1. Receive user-input enameled wire specifications, including wire diameter range of 1.0-20.0mm; S2. Automatically calculate the maximum detection speed based on the wire diameter and configure the line scan camera's line frequency; S3. Select light source mode based on wire diameter: Enable visible light combination when diameter ≤ 5mm, and enable near-infrared polarized light when diameter > 5mm; S4. Surface defects are identified in real time by optimizing the Yolo11 model, and the high-voltage pinhole detection module is triggered simultaneously for electrical verification. S5. When the number of consecutive defects or the density of defects per unit length exceeds the preset threshold, an audible and visual alarm is triggered.

[0011] Preferably, the defect determination in step S4 must simultaneously meet the following conditions: visual features conform to the defect classification criteria; the defect size exceeds the threshold settings for length, width, and area; and the similarity with the sample library is ≥70%.

[0012] Preferably, it also includes: S6, real-time monitoring of the enameled wire position, and automatic adjustment of the guide wheel position when it deviates from the center of the imaging area; S7: Dynamically maintains clear imaging within a working distance of 200-1200mm using an electric zoom lens.

[0013] Preferably, the alarm threshold setting in step S5 includes: ≥2 pinhole defects appearing continuously within every 10 meters; ≥20 small particle defects appearing within every 100 meters; and ≥2 large particle defects appearing within every 100 meters.

[0014] By employing the above technical solution, the present invention provides a high-speed enameled wire surface defect detection system and method, which has at least the following beneficial effects: (1) This invention achieves full-dimensional quality verification by innovatively integrating visual inspection to optimize the combination of high-speed linear array camera and optimized YOLO11 model, as well as optical measurement and physical inspection technologies. Multispectral imaging combined with optimized YOLO11 model improves the detection rate of pinholes and scratches at the 0.1mm level. At the same time, it integrates a laser diameter measuring instrument to provide real-time feedback of wire diameter and achieves dynamic speed matching. In addition, the high-voltage pinhole electrical defect signal is spatiotemporally aligned with the visual results, solving the problem of high-speed false alarms.

[0015] (2) The present invention achieves adaptive speed adjustment of wire diameter through dynamic speed matching technology, and effectively reduces motion blur with 0.014mm pixel accuracy. At the same time, it effectively reduces the interference of paint layer reflection by using near-infrared polarization penetration technology with 850nm light source + 45° polarization mirror. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention, form part of this application: Figure 1 This is a schematic diagram of the system architecture of the present invention. Detailed Implementation

[0017] 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 some embodiments of the present invention, and not all 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.

[0018] Example 1 A high-speed enameled wire surface defect detection system, such as Figure 1 As shown, it includes an optical imaging module, which contains a line scan camera with a line frequency ≥80kHz, a motorized zoom lens, and a multispectral combined light source. The line scan camera is the ECO PN2KCL-400KM model, the motorized zoom lens has a working distance of 200-1200mm and supports full-surface focusing on wires with diameters of 1-20mm, and the multispectral combined light source supports switching between 450nm blue light, 630nm red light and 850nm near-infrared light modes, and is equipped with a polarization module.

[0019] The line scan camera's line frequency is dynamically adjusted according to the line diameter: when the line diameter D≤1.0mm, the line frequency ≥400kHz; when the line diameter 1.0mm<D≤5.0mm, 400kHz>line frequency ≥160kHz; when the line diameter 5.0mm<D≤10.0mm, 160kHz>line frequency ≥110kHz; when the line diameter 10.0mm<D≤20.0mm, 110kHz>line frequency ≥80kHz.

[0020] The operating modes of the multispectral combined light source include: for enameled wires with a diameter ≤ 5mm, a combination of 450nm blue light and 630nm red light is used for illumination; for enameled wires with a diameter > 5mm, a combination of 850nm near-infrared light and a polarization module is used for illumination.

[0021] The real-time diameter measurement module uses a laser diameter gauge to measure the diameter of enameled wire in real time with an accuracy of ±0.001mm and a measurement frequency of ≥1kHz.

[0022] The high-voltage pinhole detection module is used for electrical defect verification and outputs electrical defect signals.

[0023] The control processing module is used to perform the following operations: Speed ​​V is detected by dynamic matching of wire diameter D. max The following relational expressions are satisfied: Synchronize visual imaging, high-pressure pinhole detection module signals and diameter measurement data, and align visual imaging, high-pressure pinhole detection module signals and diameter measurement data through FPGA timestamps, with a synchronization time deviation of ≤±5ms; Run the optimized YOLO11 model for defect identification. The optimized YOLO11 model includes: The C3k2 feature interaction module is the foundational feature extraction module of the model. It's a bottleneck structure containing three convolutional layers, enabling efficient feature extraction. It uses a 2x2 convolutional kernel, a relatively small kernel suitable for capturing subtle features. The core function of this module is to fuse information from different channels in an image, including edges, color, and texture, allowing the network to integrate multiple information sources to understand the image and achieve cross-channel feature fusion. The C3k2 feature interaction module improves the recall rate for small objects by 3.2%. In defect detection, scratches, small bubbles, and other small objects are often difficult to detect. This module significantly improves the model's ability to find these small defects.

[0024] The C2PSA dual attention mechanism is also known as position-channel dual attention. Channel attention determines the more important detection and recognition features and assigns higher weights to the corresponding detection and recognition features. Position attention, on the other hand, determines the more important detection and recognition locations, allowing the model to focus on the regions in the image most likely to contain defects. After processing by the position-channel dual attention module, the model's attention to defective regions in the image is increased by 3 times, greatly enhancing the model's ability to accurately locate minute defects in complex backgrounds.

[0025] Depthwise separable convolutional layers are a lightweight convolutional operation. They decompose standard convolution into two steps, significantly reducing the computational cost and number of parameters, thus improving model speed and efficiency. Although the depthwise separable convolutional layer follows the position-channel dual attention module, it greatly offsets the computational overhead of the attention mechanism, resulting in an 87% reduction in overall model computation compared to standard convolution. This is crucial for industrial scenarios requiring real-time detection.

[0026] Multi-scale SPPF pooling layers: Perform 5×5, 9×9, and 13×13 pooling kernel operations in parallel. The Spatial Pyramid Pooling - Fast module is used for multi-scale feature fusion. This module uses pooling kernels of various sizes, such as 5x5, 9x13, and 13x13, to process feature maps from the backbone network. Defects vary in size and type, including scratches of different lengths and bubbles of different sizes. The SPPF module can simultaneously capture feature information at different scales and fuse these features together, enabling the model to achieve good detection results for defects of different sizes.

[0027] The above modules enable the optimized YOLO11 model to achieve a detection rate of ≥99.1% for pinholes with a diameter of 0.1mm and an inference latency of ≤2ms.

[0028] It also includes a defect verification module, which is configured to: when the vision system detects a surface defect, trigger the high-pressure pinhole detection module to perform electrical signal verification within a ±5ms time window; if the high-pressure pinhole detection module verifies successfully, it is determined to be a valid defect; if the high-pressure pinhole detection module fails to verify, it is otherwise marked as a false alarm.

[0029] Example 2 A method for detecting surface defects in high-speed enameled wire, such as Figure 1 As shown, it includes the following steps: S1. Receive user-input enameled wire specifications, including wire diameter range of 1.0-20.0mm; S2. Automatically calculate the maximum detection speed based on the wire diameter and configure the line scan camera's line frequency; S3. Select light source mode based on wire diameter: Enable visible light combination when diameter ≤ 5mm, and enable near-infrared polarized light when diameter > 5mm; S4. Surface defects are identified in real time by optimizing the Yolo11 model, and the high-voltage pinhole detection module is triggered simultaneously for electrical verification. In step S4, the defect determination must simultaneously meet the following conditions: the visual features conform to the defect classification criteria; the defect size exceeds the threshold settings for length, width, and area; and the similarity with the sample library is ≥70%.

[0030] S5. When the number of consecutive defects or the defect density per unit length exceeds a preset threshold, an audible and visual alarm is triggered. The alarm threshold settings in step S5 include: ≥2 pinhole defects appearing consecutively within every 10 meters; ≥20 small particle defects appearing within every 100 meters; ≥2 large particle defects appearing within every 100 meters.

[0031] S6. Real-time monitoring of the enameled wire position; when it deviates from the center of the imaging area, the guide wheel position is automatically adjusted. S7: Dynamically maintains clear imaging within a working distance of 200-1200mm using an electric zoom lens.

[0032] This invention is performed through the following steps when in use: First, the system hardware is configured. In the optical imaging module, the linear array camera is selected from the ECO PN2KCL-400KM; the lens is a motorized zoom lens with a focal length of 35-150mm, and the working distance of the lens is 200-1200mm; the light source adopts a three-channel LED light source box, including 450nm blue light, 630nm red light and 850nm near infrared, and the light source output end of the three-channel LED light source box is equipped with a 0°-90° adjustable polarizer.

[0033] In the real-time diameter measurement module, the laser diameter gauge is the Proton InteliSENS DGK model, which has a sampling rate of 1kHz and an accuracy of ±0.001mm; the high-voltage pinhole detection module has an adjustable output voltage of 0.5-6kV and a leakage current detection range of 0.1-10mA.

[0034] The industrial computer in the control processing module uses a combination of Intel i7-12700K and NVIDIA RTX 4090; the synchronous controller is a Xilinx Kintex-7 FPGA, which can achieve spatial alignment of ±5ms.

[0035] The software is then initialized. First, the vision inspection system is started, and then the system performs a self-test. After the self-test is completed, the system parameters are configured, including product model, wire diameter D, width tolerance, pinhole voltage, and other parameters.

[0036] Next, dynamic detection is performed, including wire diameter-speed matching. The system automatically calculates the line frequency of the line scan camera based on the real-time wire diameter D, and then switches the light source mode according to the wire diameter. In the small wire diameter mode with a wire diameter of less than or equal to 5mm, the 450nm+630nm combined light is activated and polarization is turned off; in the large wire diameter mode with a wire diameter greater than 5mm, the 850nm near-infrared light + 45° polarization is activated to suppress paint layer reflection.

[0037] Subsequently, defect detection was performed on the surface of the enameled wire. A linear array camera captured images of the surface of the enameled wire with a pixel accuracy of 0.014mm. The YOLO11 was optimized to sequentially perform cross-layer fusion feature extraction using the C3k2 module, the C2PSA attention mechanism, and SPPF multi-scale pooling on the acquired images. The cross-layer fusion feature extraction using the C3k2 module was used to extract pinhole defects with a diameter of 0.1mm, the C2PSA attention mechanism was used to locate scratch edges, and SPPF multi-scale pooling was used to identify large-diameter marks. Simultaneously, multimodal verification was performed: after visual detection of the defect, the coordinates of the defect were located, and the FPGA verified whether the current of the high-voltage pinhole detection module exceeded the threshold within the time window of [t-5ms, t+5ms]. The laser diameter gauge simultaneously detected the height of the defect protrusion.

[0038] Before the official power-on, perform a real-time calibration: if the enameled wire deviates from the center by more than 5%, adjust the guide wheel to the green baseline; adjust the lens sharpness until the MTF curve peak value is ≥0.8.

[0039] Technical effect verification data Tests were conducted on enameled wires with diameters of 1.0-20.0 mm at speeds of 60-300 m / min: After testing with this system, the detection rate of pinholes with a wire diameter of 0.1mm can reach 99.1%, the false alarm rate of large wire diameters is less than 0.5%, the algorithm delay is 1.8ms, and the false alarm rate of the high-voltage pinhole detection module is as low as 0.3% after verification. However, after testing with traditional systems, the detection rate of pinholes with a wire diameter of 0.1mm is less than 85%, the false alarm rate of large wire diameters is more than 12%, the algorithm delay is 15ms, and the false alarm rate of the high-voltage pinhole detection module after verification can reach 8.7%. Compared with existing technologies, the system in this invention improves the detection rate of pinholes with a wire diameter of 0.1mm by more than 14.1%, reduces the false alarm rate of large wire diameters by 11.5%, reduces algorithm latency by 88%, and reduces the false alarm rate of the high-voltage pinhole detection module by 8.4% after verification.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0041] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-speed enameled wire surface defect detection system, characterized in that: It includes an optical imaging module, which includes a line array camera with a line frequency of ≥80kHz, a motorized zoom lens, and a multispectral combined light source. The multispectral combined light source supports switching between 450nm blue light, 630nm red light, and 850nm near-infrared light modes and is equipped with a polarization module. The real-time diameter measurement module uses a laser diameter measuring instrument to measure the diameter of the enameled wire in real time with an accuracy of ±0.001mm. High-voltage pinhole detection module, used for verifying electrical defects; The control processing module is configured to perform the following operations: Speed ​​V is detected by dynamic matching of wire diameter D. max ; Synchronous visual imaging, high-pressure pinhole detection module signal and diameter measurement data, with a time deviation of ≤±5ms; Run the optimized YOLO11 model for defect identification.

2. The high-speed enameled wire surface defect detection system according to claim 1, characterized in that, The line scan camera's line frequency is dynamically adjusted according to the line diameter: When the wire diameter D≤1.0mm, the line frequency ≥400kHz; When the wire diameter is 1.0mm < D ≤ 5.0mm, 400kHz > line frequency ≥ 160kHz; When the wire diameter is 5.0mm < D ≤ 10.0mm, 160kHz > line frequency ≥ 110kHz; When the wire diameter is 10.0mm < D ≤ 20.0mm, 110kHz > line frequency ≥ 80kHz.

3. The high-speed enameled wire surface defect detection system according to claim 1, characterized in that: The operating modes of the multispectral combined light source include: for enameled wires with a diameter ≤ 5mm, a combination of 450nm blue light and 630nm red light is used for illumination; for enameled wires with a diameter > 5mm, a combination of 850nm near-infrared light and a polarization module is used for illumination.

4. The high-speed enameled wire surface defect detection system according to claim 1, characterized in that, The optimized YOLO11 model includes: The C3k2 feature interaction module enables cross-layer feature fusion. C2PSA dual attention mechanism; Depthwise separable convolutional layers have a computational cost that is 1 / 8 that of standard convolutions; Multi-scale SPPF pooling layers execute 5×5, 9×9, and 13×13 pooling kernel operations in parallel.

5. The high-speed enameled wire surface defect detection system according to claim 1, characterized in that: The optimized YOLO11 model has a detection rate of ≥99.1% for pinholes with a diameter of 0.1mm and an inference latency of ≤2ms.

6. The high-speed enameled wire surface defect detection system according to claim 1, characterized in that, It also includes a defect verification module, which is configured to: when the vision system detects a surface defect, trigger the high-pressure pinhole detection module to perform electrical signal verification within a ±5ms time window; if the high-pressure pinhole detection module verifies successfully, it is determined to be a valid defect; if the high-pressure pinhole detection module fails to verify, it is otherwise marked as a false alarm.

7. A method for detecting surface defects in high-speed enameled wire, used in a high-speed enameled wire surface defect detection system as described in any one of claims 1-6, characterized in that, Includes the following steps: S1. Receive user-input enameled wire specifications, including wire diameter range of 1.0-20.0mm; S2. Automatically calculate the maximum detection speed based on the wire diameter and configure the line scan camera's line frequency; S3. Select light source mode based on wire diameter: Enable visible light combination when diameter ≤ 5mm, and enable near-infrared polarized light when diameter > 5mm; S4. Surface defects are identified in real time by optimizing the Yolo11 model, and the high-voltage pinhole detection module is triggered simultaneously for electrical verification. S5. When the number of consecutive defects or the density of defects per unit length exceeds the preset threshold, an audible and visual alarm is triggered.

8. The method for detecting surface defects in high-speed enameled wire according to claim 7, characterized in that, In step S4, the defect determination must simultaneously meet the following conditions: visual features conform to the defect classification criteria; the defect size exceeds the threshold settings for length, width, and area; and the similarity with the sample library is ≥70%.

9. The high-speed enameled wire surface defect detection system and method according to claim 7, characterized in that, Also includes: S6. Real-time monitoring of the enameled wire position; when it deviates from the center of the imaging area, the guide wheel position is automatically adjusted. S7: Dynamically maintains clear imaging within a working distance of 200-1200mm using an electric zoom lens.

10. The method for detecting surface defects in high-speed enameled wire according to claim 7, characterized in that, The alarm threshold settings in step S5 include: ≥2 pinhole defects appearing consecutively within every 10 meters; ≥20 small particle defects appearing within every 100 meters; and ≥2 large particle defects appearing within every 100 meters.

Images