Defect Detection Method and System for Wire Coating

Abstract

An embodiment of the present invention provides a method and system for defect detection of wire coatings. Obtain training samples. Each training sample includes an image of a wire irradiated by a light source, and each light source has only one specific wavelength. Mark those training samples according to the vibration frequencies of multiple sections of the wire. If the vibration frequency of the first section among those sections is greater than the frequency threshold, mark the training sample corresponding to the first section as a good product sample. If the vibration frequency of the second section among those sections is not greater than the frequency threshold, mark the training sample corresponding to the second section as a defective sample. Establish a defect detection module based on the marked training samples and a machine learning algorithm. The defect detection module is used to analyze the coating defects on the wire. Thereby, online detection can be provided.

Description

Technical Field

[0001] The present invention relates to a defect analysis technology, and in particular to a defect detection method and system for wire coatings. Background Art

[0002] The midstream products in the steel industry belong to carbon steel. These products include cold and hot rolled steel coils, steel bars, wire rod coils, and bar steel coils. In terms of stainless steel, its products include cold and hot rolled stainless steel coils, stainless steel bars and wires, and stainless steel sections, and these products require further cutting and tube manufacturing. If using general steel billets as raw materials, the raw materials are rolled into coils, and after further processing of the coils, downstream products such as screws, nuts, and steel wires can be produced. The difference between midstream wire rod coils and bar steel coils is that those with a diameter above 14 mm are called bar steel, and those with a diameter below 14 mm are called wire rods.

[0003] Facing the production of high-value fasteners, the quality requirements for wire rod coils are gradually increasing. Although intelligent manufacturing has been introduced in the production management of fasteners, defects during the wire drawing process cannot be detected in real time. Due to the old equipment in the wire drawing factory, it is necessary to manually check and confirm the status of the forming equipment and the male and female dies. However, the production of high-end fasteners has extremely high quality requirements for the provided wire rods, and manual inspection usually cannot meet customer needs and even affects the production line efficiency. Summary of the Invention

[0004] The present invention is directed to a defect detection method and system for wire coatings, which can automatically detect coating defects.

[0005] According to an embodiment of the present invention, the defect detection method for wire coatings is applicable to wires coated with coatings. The defect detection method includes (but is not limited to) the following steps: obtaining one or more training samples. Each training sample includes an image of a wire irradiated by a light source, and each light source has only one specific wavelength. Marking those training samples according to the vibration frequencies of multiple sections of the wire. If the vibration frequency of the first section among those sections is greater than the frequency threshold, mark the training sample corresponding to the first section as a good product sample. If the vibration frequency of the second section among those sections is not greater than the frequency threshold, mark the training sample corresponding to the second section as a defective sample. Establishing a defect detection module based on the marked training samples and a machine learning algorithm. The defect detection module is used to analyze the coating defects on the wire.

[0006] According to an embodiment of the present invention, a defect detection system for wire coatings is applicable to wires coated with coatings and includes an arithmetic device. The arithmetic device is configured to obtain one or more training samples, label those training samples according to the vibration frequencies of multiple sections of the wire, and establish a defect detection module based on the labeled training samples and a machine learning algorithm. Each training sample includes an image of the wire irradiated by a light source, and each light source has only one specific wavelength. If the vibration frequency of the first section among those sections is greater than the frequency threshold, the training sample corresponding to the first section is labeled as a good product sample. If the vibration frequency of the second section among those sections is not greater than the frequency threshold, the training sample corresponding to the second section is labeled as a defective sample. The defect detection module is used to analyze the coating defects on the wire.

[0007] Based on the above, for the defect detection method and system for wire coatings according to an embodiment of the present invention, given a light source with a specific wavelength, an image of the wire coating is obtained, and the light input amount threshold is adjusted to obtain a clear image. The good and defective sections of the wire are judged based on the vibration frequency during the wire drawing process. Then, after collecting a certain number of good and defective samples, a defect detection model can be established through machine learning. Using this defect detection model, the goal of on-line measurement of wire coating defects can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The accompanying drawings are included to provide a further understanding of the present invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the present invention and, together with the description, are used to explain the principles of the present invention.

[0009] Figure 1 is a schematic diagram of a defect detection system according to an embodiment of the present invention;

[0010] Figure 2 is Figure 1 a partial side view of;

[0011] Figure 3A is a perspective view of a housing according to an embodiment of the present invention;

[0012] Figure 3B is Figure 3A a side view of;

[0013] Figure 4 is a schematic diagram of a defect detection method according to an embodiment of the present invention;

[0014] Figure 5A is an example of a training sample - defective sample;

[0015] Figure 5B is an example of a training sample - good product sample;

[0016] Figure 6Ais an example of an image obtained under background blue light;

[0017] Figure 6B is an example of an image obtained under background red light;

[0018] Figure 7 is an example of a vibration frequency;

[0019] Figure 8A is an example of the vibration frequency corresponding to a defective sample;

[0020] Figure 8B is an example of the vibration frequency corresponding to a non-defective sample;

[0021] Figure 9 is a schematic flow diagram of machine learning according to an embodiment of the present invention;

[0022] Figure 10 is an example of an image to be detected.

[0023] Explanation of the reference numerals in the drawings

[0024] 1: System;

[0025] 10: Image capture device;

[0026] 20: Light source;

[0027] 21: Laser point light source;

[0028] 23: Background light source;

[0029] 25: Housing;

[0030] 30: Wire drawing module;

[0031] 40: Vibration sensor;

[0032] 50: Computing device;

[0033] W: Wire;

[0034] θ: Angle;

[0035] S410~S450, S910~S930: Steps;

[0036] S: Training sample;

[0037] 904: Neural network;

[0038] 905: Classification result. Detailed implementation manners

[0039] Reference will now be made in detail to the exemplary embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Whenever possible, the same component symbols are used in the drawings and the description to denote the same or similar parts.

[0040] Figure 1 is a schematic diagram of a defect detection system 1 according to an embodiment of the present invention. Please refer to Figure 1 , the defect detection system 1 includes (but is not limited to) an image capture device 10, a light source 20, a wire extraction module 30, a vibration sensor 40, and an arithmetic device 50. The defect detection system 1 is applicable to detecting a wire W coated with a film (for example, phosphate, oxalate, or galvanized coating) (which may be various types of metal or alloy wires).

[0041] The image capture device 10 can be a camera, a video camera, or a monitor, and is used to photograph the wire W to obtain an image. In one embodiment, the image capture device 10 includes an image sensor, an image processor, and a lens, and its specifications (for example, imaging aperture, magnification, focal length, imaging viewing angle, image sensor size, etc.) and configuration can be changed according to actual needs.

[0042] The light source 20 is used to irradiate the wire W, and the light source 20 has only one wavelength. In one embodiment, the light source 20 can be a laser point light source 21, and the wavelength of the laser point light source 21 is approximately 532 nanometers (nm). In one embodiment, the light source 20 can be a background light source 23, and the wavelength of the background light source 23 is between 430 and 455 nm (substantially blue light). In some embodiments, the background light source 23 can be implemented by a ring light source, a ring cover light source, or a plurality of strip light sources. In one embodiment, the light source 20 includes the laser point light source 21 and the background light source 23. In some embodiments, the light source 20 may also provide other wavelengths or light mask areas.

[0043] Figure 2 is Figure 1 a partial side view. Please refer to Figure 2 , the image capture device 10 photographs the wire W downward (for example, the shooting direction is substantially perpendicular to the horizontal plane). In one embodiment, the angle θ between the extension line of the laser point light source 21 to the wire W and the direction (for example, downward) in which the image capture device 10 obtains the image is between 15 and 30 degrees.

[0044] In one embodiment, the system 1 further includes a housing. For example, Figure 3A is a perspective view of a housing 25 according to an embodiment of the present invention, and Figure 3B is Figure 3A a side view of Figure 3A and Figure 3B , the housing 25 is a hollow cuboid, and the image capture device 10 and the laser point light source 21 are located in the cavity of the housing 25. It should be noted that the material of the housing 25 has the characteristics of light extinction or reducing light scattering, so that the light source 20 stably irradiates the wire W.

[0045] It should be noted thatFigure 3A and Figure 3B The shape of the housing 25 shown is only an example. In other embodiments, its shape can still be changed according to actual needs, or the housing 25 can be omitted.

[0046] The wire drawing module (or wire stretching machine) 30 is used to drive the wire W so that the image capturing device 10 can capture different sections of the wire W. In one embodiment, the wire drawing module 30 is used to stretch the wire W so as to process the wire W into a specified wire diameter.

[0047] The vibration sensor 40 is used to measure the vibration frequency of the wire W. It should be noted that during the process of the wire drawing module 30 stretching or driving the wire W, the wire W may vibrate.

[0048] The computing device 50 can be a device such as a desktop computer, a notebook computer, an AIO computer, a smart phone, a tablet computer, or a server. The computing device 50 may include (but is not limited to) a memory and a processor. In one embodiment, the memory stores program code, and the program code can be loaded and executed by the processor to implement the method flow of the embodiments of the present invention.

[0049] In one embodiment, the computing device 50 is connected to at least one of the image capturing device 10, the light source 20, the wire drawing module 30, and the vibration sensor 40, and is used to control its operation or receive data. For example, the computing device 50 obtains the image of the image capturing device 10 and the vibration frequency sensed by the vibration sensor 40. Again, for example, the computing device 50 controls the operation of the wire drawing module 30.

[0050] In one embodiment, the computing device 50 can be integrated with at least one of the image capturing device 10, the light source 20, the wire drawing module 30, and the vibration sensor 40. In some embodiments, some or all of the functions of the computing device 50 can be implemented by software or a chip.

[0051] Hereinafter, the method described in the embodiments of the present invention will be described in conjunction with the various components, modules, and devices in the defect detection system 1. Each process of this method can be adjusted according to the implementation situation and is not limited thereto.

[0052] Figure 4 is a schematic diagram of a defect detection method according to an embodiment of the present invention. Please refer to Figure 4 , the computing device 50 obtains one or more training samples (step S410). Specifically, each training sample includes an image of the wire W irradiated by the light source 20. That is, the image obtained by the image capturing device 10 capturing the wire W. In one embodiment, the image capturing device 10 can obtain an image of the wire W in response to a trigger condition (such as a timer, a control instruction, or the confirmation of a good product and a defective product) as a training sample.

[0053] For example, Figure 5A is an example of a training sample - defective sample. Please refer to Figure 5A , this image shows that the surface of the wire W has defects. Figure 5B is an example of a training sample - non - defective sample. Please refer to Figure 5B , this image shows that the surface of the wire W is smooth.

[0054] It should be noted that the training samples are taken under a light source 20 with a specific wavelength. Figure 6A is an example of an image obtained under background blue light, and Figure 6B is an example of an image obtained under background red light. Please refer to Figure 6A and Figure 6B and Figure 6B , a background light source 23 with a wavelength of approximately 430 to 455 nm (showing blue light) will help improve the image contrast. In comparison, the contrast of the image under background red light (e.g., wavelength of 633 to 635 nm) is not high.

[0055] In addition, as Figure 2 shown, the angle θ between the irradiation direction of the laser point light source 21 and the direction of obtaining the image is between 15 and 30 degrees. Such a specific angle also helps to present the defects. In some embodiments, the light input amount from the light source 20 to the wire W can be adjusted to improve the image clarity.

[0056] Please refer to Figure 4 , the computing device 50 marks those training samples according to the vibration frequencies of multiple sections in the wire W (step S430). Specifically, through experiments, it is known that during the wire drawing / pulling process, the surface quality of the wire is reflected in the vibration frequency of the wire drawing. The vibration frequency of the defective surface is less than that of the smooth surface. For example, the vibration frequency of the defective surface is 160 Hz, and the vibration frequency of the smooth surface is 1200 Hz. Therefore, the vibration frequencies of the sections of the wire W can be used to mark the corresponding training samples of this section as non - defective samples (i.e., smooth surface) or defective samples (i.e., defective surface). It should be noted that the length of the section is approximately the length of the wire W that can be photographed within the field of view of the image capture device 10, but it can still be changed according to actual needs.

[0057] If the vibration frequency of the first section among those sections of the wire W is greater than the frequency threshold, the computing device 50 marks the training sample corresponding to the first section as a non - defective sample. Assuming that the vibration frequency of the defective sample is 160 Hz and the vibration frequency of the non - defective sample is 1200 Hz, the frequency threshold can be set to, for example, 500, 600, or 1000 Hz. If the vibration frequency of the second section among those sections is not greater than the frequency threshold, the computing device 50 marks the training sample corresponding to the second section as a defective sample.

[0058] For example, during the process of stretching the wire W by the wire drawing module 30, the vibration sensor 40 reports the vibration frequencies of each section to the computing device 50. As Figure 7 are examples of vibration frequencies. Please refer to Figure 7 , on the horizontal time axis (in units of the number of samplings), the vibration frequencies of each sampling number (corresponding to a section) may be different.

[0059] The computing device 50 can distinguish each section according to time or speed, and mark the images captured by the image capturing device 10 according to the reported vibration frequencies. Figure 8A are examples of the vibration frequencies corresponding to defective samples. Please refer to Figure 8A , the highest vibration frequency of the defective samples is approximately 160 Hz. Figure 8B are examples of the vibration frequencies corresponding to good samples. Please refer to Figure 8B , the highest vibration frequency corresponding to the good samples can be as high as 1200 Hz. Thereby, a mechanism for automatically marking samples can be provided.

[0060] In one embodiment, the system 1 may further include a wire diameter sensor (not shown in the figure), and perform enhanced verification on each section of the wire W. For example, a wire diameter threshold is set, and those exceeding the threshold are defective samples.

[0061] Please refer to Figure 4 , the computing device 50 establishes a defect detection module (step S450) based on the marked training samples and based on a machine learning algorithm. Specifically, the machine learning algorithm can be a regression analysis algorithm, an extreme gradient boosting (XGboost) algorithm, a light gradient boosting machine (LightGBM), a neural network algorithm, a random forest algorithm, a support vector regression algorithm, or other algorithms. The machine learning algorithm can analyze the training samples to obtain patterns therefrom, so as to predict unknown data through the patterns. The defect detection model is a machine learning model constructed after learning, and inferences are made based on it for the data to be evaluated. In the embodiment of the present invention, the defect detection module is used to analyze film defects (such as scars, transverse cracks, scratches / folds, wormholes, scratches, or rust skins) on the wire W.

[0062] For example, Figure 9 is a schematic flowchart of machine learning according to an embodiment of the present invention. Please refer to Figure 9, the arithmetic device 50 performs feature extraction on the training sample S to obtain one or more feature maps (step S910). Feature extraction constructs derived values (or called feature values) that are informative and non-redundant from the initial measured / collected / obtained data samples. Feature extraction can assist subsequent learning and rule induction processes and can provide a better interpretation of the initial data samples. In other words, feature extraction can simplify the input data into a feature set (which can be regarded as important or useful information) and directly use the feature set to perform subsequent tasks (such as model training, component analysis, object detection, etc.). For example, for feature extraction of images, features such as edges, corners, Scale-Invariant Feature Transform (SIFT), curvature, shape, etc. can be obtained.

[0063] The arithmetic device 50 can use a Neural Network (NN) 904 to train a defect detection model (step S930) so that the defect detection model can estimate the classification result 905. For example, the defect detection model classifies the sections of the wire W in the image as good products (without defects) or defective products (with defects). Figure 10 is an example of the image to be detected. Please refer to Figure 10 , the defect detection model determines that this section has a 99.5% probability of being a defective product and only a 0.5% probability of being a good product. Another example is that the defect detection model can estimate the variation conditions of the wire drawing or post-processing process.

[0064] In an embodiment, if the quantity ratio of the good product samples and defective samples among multiple sections of the wire W conforms to a balanced ratio, the arithmetic device 50 can start the establishment of the defect detection module. Specifically, generally, most sections of the wire W are smooth and defect-free. Too many smooth sections will result in the number of good product samples being much larger than that of defective samples. Training the defect detection model in such a situation where the sample ratio is disparate may affect the recognition accuracy. Therefore, the quantity ratio of good product samples and defective samples can be set to a specific ratio before model training. This specific ratio is, for example, 6:4 (that is, good product samples account for 60% and defective samples account for 40%), 5:5 (that is, good product samples account for 50% and defective samples account for 50%), or 4:6 (that is, good product samples account for 40% and defective samples account for 60%), and can be used as the balanced ratio. If the quantity ratio does not conform to the balanced ratio, the arithmetic device 50 can disable the establishment of the defect evaluation module and wait until the quantity ratio conforms to the balanced ratio before establishing the defect evaluation module. For example, the arithmetic device 50 can first accumulate a certain number of training samples and then delete the excessive types according to the balanced ratio. Another example is that the arithmetic device 50 can accumulate good product samples and defective samples to the specified quantity respectively according to the balanced ratio.

[0065] In summary, in the method and system for defect detection of wire coatings according to the embodiments of the present invention, a wire is irradiated with a light source having a specific wavelength and angle, the types of training samples are automatically marked according to the vibration frequency during the wire drawing process, and a defect detection model is established by means of machine learning. Thereby, the quality and yield of the wire drawing process can be determined, and the goal of online defect detection of wire coatings can be achieved.

[0066] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for defect detection of wire coatings, characterized in that, Applicable to a wire covered with a film, the defect detection method includes: Obtaining a plurality of training samples, wherein each of the training samples includes an image of a wire irradiated by at least one light source, and each of the light sources has only a wavelength; Marking the training samples according to the vibration frequencies of a plurality of sections of the wire, wherein if the vibration frequency of a first section in the section is greater than a frequency threshold, marking the training sample corresponding to the first section as a good sample; and if the vibration frequency of a second section in the section is not greater than the frequency threshold, marking the training sample corresponding to the second section as a bad sample; and Establishing a defect detection module based on the marked training samples and based on a machine learning algorithm, wherein the defect detection module is used to analyze the film defects on the wire.

2. The defect detection method for wire coatings according to claim 1, wherein The light source includes a laser point light source, and the wavelength of the laser point light source is approximately 532 nanometers.

3. The defect detection method for wire coatings according to claim 1 or 2, characterized in that, The light source includes a background light source, and the wavelength of the background light source is between 430 and 455 nanometers.

4. The defect detection method for wire coatings according to claim 2, wherein, The included angle between the extension line of the laser point light source to the wire and the direction of obtaining the image is between 15 and 30 degrees.

5. The defect detection method for wire coatings according to claim 1, characterized in that, The wire includes a plurality of sections, and the step of establishing the defect detection module includes: if the quantity ratio of the good samples and the bad samples marked in the section conforms to a balance ratio, starting the establishment of the defect detection module; and if the quantity ratio does not conform to the balance ratio, disabling the establishment of the defect evaluation module.

6. A defect detection system for wire coatings, characterized in that, Applicable to a wire covered with a film, and includes: An arithmetic device configured to: Obtain a plurality of training samples, wherein each of the training samples includes an image of a wire irradiated by at least one light source, and each of the light sources has only a wavelength; Mark the training samples according to the vibration frequencies of a plurality of sections of the wire, wherein if the vibration frequency of a first section in the section is greater than a frequency threshold, marking the training sample corresponding to the first section as a good sample; and if the vibration frequency of a second section in the section is not greater than the frequency threshold, marking the training sample corresponding to the second section as a bad sample; and Establish a defect detection module based on the marked training samples and based on a machine learning algorithm, wherein the defect detection module is used to analyze the film defects on the wire.

7. The defect detection system for wire coatings according to claim 6, wherein, The defect detection system for wire film further includes: The light source, including: A laser point light source with a wavelength of approximately 532 nanometers.

8. The defect detection system for wire film according to claim 6 or claim 7, wherein the light source includes a background light source, and the wavelength of the background light source is between 430 and 455 nanometers.

9. The defect detection system for wire film according to claim 7, further includes: An image capture device, wherein the included angle between the extension line of the laser point light source to the wire and the direction of the image capture device for obtaining the image is between 15 and 30 degrees.

10. The defect detection system for wire film according to claim 6, wherein the wire includes a plurality of sections, and the arithmetic device is further configured to: If the quantity ratio of the samples marked as the good samples and the bad samples in the said section conforms to the balanced ratio, start the establishment of the defect detection module: and If the quantity ratio does not conform to the balanced ratio, disable the establishment of the defect assessment module.

Images