CN-122023835-A - Recognition method and system for metal surface laser etching texture based on deep learning

CN122023835ACN 122023835 ACN122023835 ACN 122023835ACN-122023835-A

Abstract

The application discloses a method and a system for identifying a metal surface laser etching texture based on deep learning, which relate to the technical field of laser etching, and the method for identifying the metal surface laser etching texture based on the deep learning comprises the following steps of acquiring a metal surface texture image after laser etching treatment; the method comprises the steps of preprocessing a texture image, constructing a texture data set, inputting the texture data set into a residual network model introducing an attention mechanism for training, and identifying and classifying the texture of the metal surface to be tested by utilizing the trained model so as to judge the type of the laser etching texture of the metal surface. The recognition system of the metal surface laser etching texture based on deep learning comprises an image acquisition module, a data processing module, a model training module and a texture recognition module. The recognition method and the recognition system for the laser etching texture of the metal surface based on the deep learning can realize rapid and intelligent recognition of the laser processing metal surface texture.

Inventors

SUN FENGZHEN
WANG JIAJIA
YUAN YIDING

Assignees

同济大学

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. The identification method of the laser etching texture of the metal surface based on deep learning is characterized by comprising the following steps of: Obtaining a metal surface texture image after laser etching treatment; preprocessing the texture image and constructing a texture data set; inputting the texture data set into a residual network model of an attention introducing mechanism for training; and identifying and classifying the texture of the metal surface to be detected by using the trained model so as to judge the type of the laser etching texture of the metal surface.
2. The method for recognizing laser etched texture on a metal surface based on deep learning according to claim 1, further comprising the step of preparing different surface textures on the metal surface according to different laser processing parameters, the different surface textures having different depths, before the step of acquiring the metal surface texture image.
3. The method of claim 1, wherein the preprocessing comprises at least one of image rotation, scaling, translation, flipping, gaussian filtering, gamma correction, and contrast enhancement.
4. The method for identifying deep learning based metal surface laser etched texture of claim 1, further comprising, after the step of constructing a metal surface texture dataset, partitioning the texture dataset, the step of partitioning comprising: The method comprises the steps of preprocessing images, storing the preprocessed images in different folders according to different texture depths, randomly dividing partial images in each folder according to a preset proportion to serve as a test set, storing the test set in a test set folder independently, and storing the rest images in a training set folder.
5. The method for identifying deep learning-based metal surface laser etched textures according to claim 1, wherein in the training step, the residual network model is a ResNet series network.
6. The method for identifying deep learning-based metal surface laser etched textures according to claim 5, wherein in the training step, the attention mechanism comprises a channel attention mechanism and/or a spatial attention mechanism, and the attention mechanism is a convolution block attention module.
7. The method for identifying the laser etched texture on the metal surface based on the deep learning according to claim 1, wherein in the training step, an early stop is adopted, an optimizer Adam algorithm is adopted, and a focus loss function is used.
8. The method for recognizing laser etched texture on a metal surface based on deep learning according to claim 1, further comprising the steps of performing performance evaluation on an independent test set, calculating an overall recognition Accuracy, which is calculated by the formula accuracy=tp/(tp+fp), wherein TP represents the number of correctly classified samples in the prediction process, and FP represents the number of incorrectly classified samples in the prediction process, and obtaining a confusion matrix.
9. The method for recognizing a deep learning based metal surface laser etched texture according to claim 1, wherein in the step of recognizing and classifying, the type of the texture includes at least a bowl-shaped texture and a bulb-shaped texture.
10. The recognition system based on the deep learning of the metal surface laser etching texture is characterized in that the recognition system is used for executing the recognition method based on the deep learning of the metal surface laser etching texture according to any one of claims 1-9, and the recognition system comprises: The image acquisition module is used for acquiring a metal surface texture image after laser etching treatment; the data processing module is used for preprocessing the texture image and constructing a texture data set; The model training module is used for inputting the texture data set into a residual network model introducing an attention mechanism for training; and the texture recognition module is used for recognizing and classifying the texture of the metal surface to be detected by using the trained model so as to judge the type of the laser etching texture of the metal surface.

Description

Recognition method and system for metal surface laser etching texture based on deep learning Technical Field The present disclosure relates to the field of laser etching technologies, and in particular, to a method and system for identifying a laser etching texture on a metal surface based on deep learning. Background The statements in this section merely provide background information related to the present disclosure and may not constitute prior art. The structure formed by bonding the metal material and the same or different materials is called a metal bonding structure, has the advantages of light weight, high strength, uniform stress distribution and the like, and is widely applied to the manufacturing fields of aerospace, automobiles, ships and the like. The preparation of the material surface with good wettability, microcosmic roughness and stable mechanics through surface pretreatment is the key for obtaining the high-performance metal bonding structure. The traditional mechanical polishing, sand blasting, chemical passivation and other methods have the problems of high energy consumption, low automation degree, large noise, more dust, heavy smell and the like, and a large amount of harmful wastewater and the like are easily generated in the treatment process, so that the development trend of green low carbon is increasingly difficult to meet. Laser texturing is an advanced surface pretreatment technology with high precision, non-contact and environmental protection. As shown in fig. 1, a laser source 1 emits a laser beam 2, the laser beam 2 is focused on a metal surface through a reflector 3 and a focusing lens 4, so that the metal surface is etched, a microstructure with controllable shape and size (such as a micro-groove array, each micro-groove has the same depth h and width w) can be prepared, meanwhile, a porous oxide film with a certain thickness is formed on the surface of a material, the chemical activity and wettability of the metal surface are improved, and the contact area and the mechanical locking effect of the metal surface and an adhesive are increased. Studies have shown that the microstructure (e.g., micro-groove) geometry changes from a "bowl" with a wide top and a narrow bottom as shown in fig. 2 to a "bulb" with a narrow top and a wide bottom as shown in fig. 3, and the mechanical locking effect reaches an extreme value only when the laser etching depth reaches a certain value. However, too great an etching depth may result in the adhesive bonding not filling the bottom of the microstructure completely, the bubble defect of the bonding interface is easily formed, and too deep etching may also result in the reduction of the rigidity of the substrate. Therefore, the proper fabrication of "bulb-like" microstructures using lasers is critical to maximizing bond interface strength. At present, the geometry of the microstructure is determined mainly by observing the cross section of the laser etched metal through a scanning electron microscope (shadow effect exists in profilometer detection, concave surface profile cannot be obtained), the problems of low efficiency, high cost, incapability of online measurement and the like exist, the requirements of modern industry on material surface texture detection are difficult to be met, and intelligent identification of the laser etched texture of the metal surface becomes urgent. It should be noted that the foregoing description of the technical background is only for the purpose of facilitating a clear and complete description of the technical solutions of the present specification and for the convenience of understanding by those skilled in the art. The above-described solutions are not considered to be known to the person skilled in the art simply because they are set forth in the background section of the present description. Disclosure of Invention In view of the shortcomings of the prior art, an object of the present specification is to provide a method and a system for identifying a laser etched texture on a metal surface based on deep learning, which can realize rapid and intelligent identification of the laser processed metal surface texture. In order to achieve the above object, the embodiment of the present disclosure provides a method for identifying a laser etched texture on a metal surface based on deep learning, including the following steps: Obtaining a metal surface texture image after laser etching treatment; preprocessing the texture image and constructing a texture data set; inputting the texture data set into a residual network model of an attention introducing mechanism for training; and identifying and classifying the texture of the metal surface to be detected by using the trained model so as to judge the type of the laser etching texture of the metal surface. As a preferred embodiment, before the step of acquiring the metal surface texture image, the method further comprises the step of preparing different surface tex