Rare earth reflection heat insulation metal plate and roller coating quality detection method thereof

Abstract

The invention relates to the technical field of heat insulation metal plates, and particularly discloses a rare earth reflection heat insulation metal plate and a roller coating quality detection method of the rare earth reflection heat insulation metal plate. The surface coating is composed of a multi-layer overlapping structure of a rare earth composite reflecting material coating and a rare earth composite hollow microbead thermal insulation material coating. In the preparation process of the rare earth reflective heat insulation metal plate, machine vision is used for constructing a roller coating quality detection scheme, defects such as bubbles, cracks and non-uniformity on the surface of a coating can be automatically recognized, large-scale roller coating quality control is achieved, and therefore the heat insulation performance and durability of a final product are guaranteed.

Description

Technical Field

[0001] The present application relates to the technical field of heat-insulating metal plates, and more specifically, to a rare earth reflective heat-insulating metal plate and a roller coating quality detection method thereof. Background Art

[0002] With the development of science and technology and the enhancement of environmental awareness, the requirements for energy-saving performance of materials in the construction and industrial fields are getting higher and higher. Especially in the building envelope, the heat loss of walls and roofs is one of the important factors affecting building energy consumption. In order to effectively reduce the heat exchange between the interior and exterior environment of the building and improve the energy efficiency of the building, the development of new insulation materials has become a research hotspot.

[0003] Traditional thermal insulation materials mainly rely on their own physical properties (such as low thermal conductivity) to achieve thermal insulation effects, but in practical applications, they have problems such as heavy weight and inconvenient construction. In recent years, with the development of nanotechnology and new materials science, the technology of using the optical properties of specific substances or materials to reflect solar radiation and achieve efficient thermal insulation has gradually attracted attention. Among them, rare earth elements show great potential in the application of reflective thermal insulation coatings due to their unique spectral characteristics and chemical stability.

[0004] Rare earth reflective thermal insulation coating is a functional coating with rare earth compounds as the main ingredients. It can effectively reflect the near-infrared part of sunlight while maintaining good visible light transmittance, thereby significantly reducing the surface temperature of objects without affecting lighting, achieving the purpose of energy saving and consumption reduction.

[0005] However, in the actual coating process of this type of coating, the quality of the coating directly affects the thermal insulation performance of the final product. In order to ensure the quality of the coating, it is usually necessary to accurately test the coating quality. At present, the method for testing the coating quality mainly relies on manual sampling inspection, which is not only inefficient, but also easily affected by the subjective judgment of the inspectors, making it difficult to ensure the objectivity and accuracy of the test results.

[0006] Therefore, an optimized rare earth reflective insulation metal plate and a preparation method thereof are desired. Summary of the invention

[0007] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a new type of rare earth reflective heat-insulating metal plate and a roller coating quality detection method thereof. The rare earth reflective heat-insulating metal plate is a metal plate with high reflection ability to sunlight and infrared rays and heat insulation function, and its surface coating is composed of a multi-layer overlapping structure of a rare earth composite reflective material coating and a rare earth composite hollow microbead heat-insulating material coating. In the preparation process of the rare earth reflective heat-insulating metal plate, machine vision is used to construct a roller coating quality detection scheme, which can automatically identify defects on the coating surface, such as bubbles, cracks, unevenness, etc., to achieve large-scale roller coating quality control, thereby ensuring the heat insulation performance and durability of the final product.

[0008] Accordingly, according to one aspect of the present application, a method for preparing a rare earth reflective heat-insulating metal plate is provided, comprising: providing a rare earth reflective heat-insulating coating and a metal plate; performing surface treatment on the metal plate to obtain a surface-treated metal plate; applying the rare earth reflective heat-insulating coating to the surface-treated metal plate by a roller coating process to obtain a rare earth reflective heat-insulating metal plate; performing roller coating quality inspection on the rare earth reflective heat-insulating metal plate to obtain a roller coating quality inspection result; wherein, performing roller coating quality inspection on the rare earth reflective heat-insulating metal plate to obtain a roller coating quality inspection result comprises: acquiring a coating surface state image acquired by an industrial camera; The coating surface state image is subjected to texture feature extraction based on grid granularity to obtain a set of coating surface grid granularity texture feature vectors; the set of coating surface grid granularity texture feature vectors is subjected to feature significance modulation dynamic aggregation to obtain a coating surface texture grid granularity global significant aggregation representation vector; based on the feature difference of each coating surface grid granularity texture feature vector in the set of coating surface grid granularity texture feature vectors relative to the coating surface texture grid granularity global significant aggregation representation vector, the roller coating quality detection result is generated, wherein the roller coating quality detection result is used to indicate whether the coating surface is dense.

[0009] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, the coating surface state image is subjected to grid granularity-based texture feature extraction to obtain a set of coating surface grid granularity texture feature vectors, including: grid segmentation of the coating surface state image to obtain a set of coating surface state grid granularity image blocks; each coating surface state grid granularity image block in the set of coating surface state grid granularity image blocks is input into a texture feature extractor based on a multi-layer void convolution structure to obtain a set of coating surface grid granularity texture feature vectors.

[0010] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, the set of the coating surface grid grain texture feature vectors is subjected to feature significance modulation dynamic aggregation to obtain the coating surface texture grid grain global significant aggregation representation vector, including: determining the coating surface grid grain texture feature clustering initial center vector based on the feature distribution field of the set of the coating surface grid grain texture feature vectors; based on the spatial span of each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vectors relative to the coating surface grid grain texture feature clustering initial center vector, the set of the coating surface grid grain texture feature vectors is subjected to significance modulation dynamic aggregation to obtain the coating surface texture grid grain global significant aggregation representation vector.

[0011] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, based on the characteristic distribution field of the set of coating surface grid grain size texture feature vectors, the initial center vector of the coating surface grid grain size texture feature clustering is determined, including: calculating the static energy factor of each coating surface grid grain size texture feature vector in the set of coating surface grid grain size texture feature vectors to obtain a set of coating surface grid grain size texture feature static energy factors; selecting the coating surface grid grain size texture feature vector corresponding to the maximum value in the set of coating surface grid grain size texture feature static energy factors as the initial center vector of the coating surface grid grain size texture feature clustering.

[0012] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, the static energy factor of each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vector is calculated to obtain a set of coating surface grid grain texture feature static energy factors, including: calculating the kurtosis of the coating surface grid grain texture feature vector, and inputting the kurtosis into a sigmoid activation function to obtain the coating surface grid grain texture feature static energy factor.

[0013] In the preparation method of the rare earth reflective insulation metal plate, based on the spatial span of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector relative to the initial center vector of the coating surface grid grain size texture feature cluster, the set of the coating surface grid grain size texture feature vector is subjected to saliency modulation dynamic aggregation to obtain the coating surface texture grid grain size global saliency aggregation representation vector, including: based on the spatial span between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the initial center vector of the coating surface grid grain size texture feature cluster, and The set of static energy factors of the coating surface mesh granularity texture features is used to calculate the dynamic aggregation energy factors of each coating surface mesh granularity texture feature vector to obtain a set of dynamic aggregation energy factors of the coating surface mesh granularity texture features; the set of dynamic aggregation energy factors of the coating surface mesh granularity texture features is input into a gated mask unit to obtain a set of dynamic aggregation weight factors of the coating surface mesh granularity texture features; based on the set of dynamic aggregation weight factors of the coating surface mesh granularity texture features, the weighted sum of the set of coating surface mesh granularity texture feature vectors is calculated to obtain a global significant aggregation representation vector of the coating surface texture mesh granularity.

[0014] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, the dynamic aggregation energy factor of each coating surface grid grain texture feature vector is calculated to obtain a set of dynamic aggregation energy factors of the coating surface grid grain texture features, including: taking the square value of the number of feature vectors between the coating surface grid grain texture feature vector and the initial center vector of the coating surface grid grain texture feature cluster as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the coating surface grid grain texture feature vector and the static energy factor of the initial center vector of the coating surface grid grain texture feature cluster and the spatial span coefficient to obtain the dynamic aggregation energy factor of the coating surface grid grain texture features.

[0015] In the above-mentioned method for preparing the rare earth reflective insulation metal plate, the roller coating quality detection result is generated based on the feature difference of each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vector relative to the coating surface texture grid grain global significant aggregation representation vector, including: calculating the coating surface state semantic difference coefficient between each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vector and the coating surface texture grid grain global significant aggregation representation vector to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients; inputting the coating surface state global distribution representation vector into a classifier-based roller coating quality detection module to obtain the roller coating quality detection result.

[0016] In the preparation method of the above-mentioned rare earth reflective insulation metal plate, the coating surface state semantic difference coefficient between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the coating surface texture grid grain size global significant aggregation representation vector is calculated to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients, including: calculating the Mahalanobis distance between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the coating surface texture grid grain size global significant aggregation representation vector as the coating surface state semantic difference coefficient to obtain the coating surface state global distribution representation vector.

[0017] According to another aspect of the present application, a rare earth reflective heat insulating metal plate is provided, characterized in that the rare earth reflective heat insulating metal plate is manufactured by the preparation method of the rare earth reflective heat insulating metal plate as described above.

[0018] Compared with the prior art, the rare earth reflective heat-insulating metal plate and the roller coating quality detection method thereof described in the present application, wherein the rare earth reflective heat-insulating metal plate is a metal plate with high reflection ability to sunlight and infrared rays and heat-insulating function, and its surface coating is composed of a multi-layer overlapping structure of a rare earth composite reflective material coating and a rare earth composite hollow microbead heat-insulating material coating. In the preparation process of the rare earth reflective heat-insulating metal plate, it uses machine vision to construct a roller coating quality detection scheme, which can automatically identify defects on the coating surface, such as bubbles, cracks, unevenness, etc., to achieve large-scale roller coating quality control, thereby ensuring the heat insulation performance and durability of the final product. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 The present invention is a flow chart of performing roller coating quality inspection on the rare earth reflective heat insulating metal plate in the method for preparing the rare earth reflective heat insulating metal plate according to an embodiment of the present application to obtain the roller coating quality inspection result.

[0021] Figure 2 This is a flow chart of step S2 in the method for preparing a rare earth reflective insulation metal plate according to an embodiment of the present application.

[0022] Figure 3This is a flow chart of step S3 in the method for preparing a rare earth reflective insulation metal plate according to an embodiment of the present application.

[0023] Figure 4 This is a flow chart of step S4 in the method for preparing a rare earth reflective insulation metal plate according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0025] The present application proposes a method for preparing a rare earth reflective thermal insulation metal plate, the preparation method comprising: providing a rare earth reflective thermal insulation coating and a metal plate; performing surface treatment on the metal plate to obtain a surface-treated metal plate; applying the rare earth reflective thermal insulation coating to the surface-treated metal plate by a roller coating process to obtain a rare earth reflective thermal insulation metal plate; and performing roller coating quality inspection on the rare earth reflective thermal insulation metal plate to obtain a roller coating quality inspection result.

[0026] In the technical solution of the present application, the metal plate can be an aluminum plate, a stainless steel plate or a plate made of other metal materials. The rare earth reflective heat-insulating coating is prepared by the following preparation method: Step 1: Add solvent, dispersant and defoamer to a dispersion tank, and disperse at a low speed until uniform; Step 2: Continue to add rare earth pigments and precipitated barium sulfate to the dispersion tank, and disperse at a high speed until uniform to obtain a first mixed material; Step 3: Grind the first mixed material to obtain a first mixed material powder; Step 4: Mix the first mixed material powder, pure acrylic emulsion, wetting agent and film-forming aid and stir to obtain a second mixed material; Step 5: Add a thickener to adjust the viscosity of the second mixed material to obtain a third mixed material; Step 6: Add a fungicide to the third mixed material and stir to obtain a rare earth reflective heat-insulating coating.

[0027] At the same time, in the process of adjusting the viscosity of the rare earth reflective thermal insulation coating, the present application further introduces intelligent industrial control technology to adjust the amount of thickener applied. First, the corresponding target state image of the mixed material is generated according to the target viscosity value input by the user, and then the image processing technology based on deep learning is used to extract and enhance the image features of the target state image of the mixed material and the actual surface state image of the mixed material to extract the surface state features of the two materials, and then based on the feature difference between the actual surface state of the mixed material and the target state, the amount of thickener applied is intelligently recommended. In this way, the optimized control of the amount of thickener applied can be achieved, and the production efficiency and quality stability of the coating can be improved.

[0028] In order to ensure good adhesion between the paint and the metal plate, the surface of the metal plate needs to be pretreated, including but not limited to degreasing, pickling and rust removal, chemical conversion treatment (such as phosphating or chromizing), etc., to remove oil stains, oxides and other impurities on the metal surface and form a protective film to improve the durability and adhesion of subsequent coatings. Then, the rare earth reflective insulation paint is evenly coated on the pretreated metal plate using an efficient roller coating process. After the coating is completed, the coated metal plate is sent to a special oven for drying and curing treatment, so that the solvent in the paint evaporates quickly and causes the paint to undergo physical or chemical changes to form a strong and durable coating structure.

[0029] Considering that the traditional roller coating quality detection method mainly relies on manual sampling inspection, this method is not only inefficient, but also easily affected by the subjective judgment of the inspectors, and it is difficult to ensure the objectivity and accuracy of the detection results. In this regard, the present application further introduces computer vision technology to perform roller coating quality detection. Specifically, an industrial camera is first used to collect the coating surface state image, and then the coating surface state image is extracted based on the network granularity texture feature using the computer vision technology based on deep learning, and the global significant texture feature of the coating surface state image is used as a reference benchmark. Based on the semantic difference of the texture features of each network area relative to the global significant texture features of the image, the global texture distribution characteristics of the coating surface are captured, so as to realize the density detection of the coating surface. In this way, defects on the coating surface, such as bubbles, cracks, unevenness, etc., can be automatically identified, and large-scale roller coating quality control can be achieved, thereby ensuring the thermal insulation performance and durability of the final product.

[0030] Figure 1 The present invention is a flow chart of performing roller coating quality inspection on the rare earth reflective heat insulating metal plate in the method for preparing the rare earth reflective heat insulating metal plate according to the embodiment of the present application to obtain the roller coating quality inspection result. Figure 1 As shown, the roller coating quality inspection of the rare earth reflective insulation metal plate includes: S1, acquiring a coating surface state image collected by an industrial camera; S2, performing texture feature extraction based on grid granularity on the coating surface state image to obtain a set of coating surface grid granularity texture feature vectors; S3, performing feature significance modulation dynamic aggregation on the set of coating surface grid granularity texture feature vectors to obtain a coating surface texture grid granularity global significant aggregation representation vector; S4, generating the roller coating quality inspection result based on the feature difference of each coating surface grid granularity texture feature vector in the set of coating surface grid granularity texture feature vectors relative to the coating surface texture grid granularity global significant aggregation representation vector, wherein the roller coating quality inspection result is used to indicate whether the coating surface is dense.

[0031] In the above-mentioned method for preparing the rare earth reflective heat-insulating metal plate, the step S1 is to obtain the coating surface state image collected by the industrial camera. It should be understood that by using an industrial camera to collect the coating surface state image, high-resolution coating surface visual information can be provided, and automated visual quality inspection can be achieved, thereby improving inspection efficiency and shortening the production cycle. In addition, using an industrial camera to collect the coating surface state image is a non-contact detection method, which avoids the damage that physical contact may cause to the coating or material, while reducing interference with the production environment.

[0032] Further, in the technical solution of the present application, a high-resolution industrial camera is used, which can capture tiny details on the coating surface, such as bubbles, cracks and unevenness, which is crucial for subsequent defect detection. Secondly, the stability and reliability of industrial cameras are also important considerations. Industrial cameras can work stably for a long time in various industrial environments, without being affected by environmental factors such as temperature and humidity, ensuring the consistency and reliability of image acquisition. At the same time, a uniform light source is set, which can reduce shadows and reflections and ensure the clarity and consistency of the image. The multi-angle light source setting can illuminate the coating surface from different directions and capture surface features from more viewing angles. For example, using a ring light source can provide uniform lighting, while using a directional light source can highlight the texture and defects of the surface. In addition, the color temperature and intensity of the light source also need to be adjusted according to the specific coating color and surface characteristics to obtain the best image effect. And fix the industrial camera in a suitable position to ensure that the distance and angle between the camera and the coating surface are consistent, which can reduce the distortion and distortion of the image. Optionally, in other examples of the present application, multiple cameras can be set to shoot the same coating surface from different angles to obtain more comprehensive information. For example, installing multiple cameras at different locations on the production line can achieve all-round monitoring of the coating surface.

[0033] In the embodiment of the present application, the step S1 also includes image acquisition. First, during the image acquisition process, it is necessary to ensure that the metal plate has been coated and dried and cured. After the coating is completed, the coating surface needs to be cleaned to remove dust and impurities to avoid interfering with image acquisition. Next, the coated metal plate is placed on a fixed platform to ensure that its position is stable and flat. This step is very important because any movement or tilt will cause image distortion. Before starting the industrial camera, it is necessary to adjust the camera parameters, such as exposure time, gain, etc., to obtain the best image quality. Too long exposure time will cause the image to be overexposed, while too short exposure time will make the image too dark. Too high gain setting will introduce noise, while too low gain setting will make the image details unclear. Therefore, it is necessary to find the best parameter setting through experiments based on the actual coating surface characteristics and lighting conditions. Trigger the camera shutter to collect images of the coating surface. In actual operation, images can be collected multiple times to ensure the reliability and integrity of the data. For example, multiple images can be collected under different lighting conditions, or the same coating surface can be photographed at different positions and angles to obtain more information. The collected images need to be saved in a designated storage device in a timely manner for subsequent processing and analysis.

[0034] In an optional embodiment of the present application, the step S1 also includes image preprocessing, and the image quality and detection accuracy can be improved by image preprocessing. First, geometric correction is performed to eliminate image deformation caused by camera lens distortion and uneven platform. Geometric correction can be achieved by a calibration plate and a calibration algorithm. There are usually marking points of known positions on the calibration plate, and the intrinsic and extrinsic parameters of the camera can be calculated through these marking points, thereby correcting the geometric distortion of the image. Secondly, illumination correction is performed to eliminate the brightness inconsistency caused by uneven light sources. Illumination correction can be achieved by histogram equalization or grayscale transformation of the image, which can adjust the brightness and contrast of the image to make the details of the coating surface more obvious. In addition, a multi-scale illumination correction method can also be used to further improve the quality of the image by performing illumination correction at different scales. In an optional embodiment of the present application, the image preprocessing also includes image enhancement. By adjusting the contrast of the image, the details of the coating surface can be made more obvious. Contrast enhancement can be achieved by linear stretching or nonlinear transformation, for example, using methods such as histogram equalization or Gamma correction. In addition, a filtering algorithm can also be used to remove noise in the image to improve the quality of the image. Common filtering algorithms include median filtering, Gaussian filtering, and bilateral filtering. These filtering algorithms can effectively remove random noise in the image while retaining the detailed information of the coating surface. Furthermore, the image preprocessing also includes image quality assessment. First, evaluate the resolution of the image to ensure that the resolution of the image is high enough to clearly display the details of the coating surface. Images with insufficient resolution cannot capture tiny defects, affecting the subsequent detection accuracy. Secondly, evaluate the clarity of the image to ensure that there is no blur or out of focus. Images with insufficient clarity will cause loss of details and affect the accuracy of the detection results. Finally, evaluate the illumination uniformity of the image to ensure that there is no obvious uneven brightness. Uneven illumination will reduce the contrast of the image and affect the identification of defects. If the image quality is not ideal, troubleshooting is required. First, check whether the equipment such as the camera, light source, and platform are working properly. For example, check whether the camera lens has dust or stains, check whether the brightness and color temperature of the light source meet the requirements, and check whether the platform is flat and stable. Secondly, adjust the camera parameters such as exposure time and gain according to the actual situation to optimize the image quality. For example, if the image is too dark, the exposure time can be increased appropriately; if the image is overexposed, the exposure time can be reduced appropriately. Through these adjustments, the image quality can be gradually improved to ensure the smooth progress of subsequent processing and analysis.

[0035] In an optional embodiment of the present application, step S1 also includes data transmission and storage, and the image data security and integrity are ensured by data transmission and storage. First, the collected image is transmitted to a computer or server in real time for subsequent processing and analysis. Data transmission can be achieved by wired or wireless means. Wired transmission has higher stability and reliability, while wireless transmission is more flexible and convenient. No matter which method is adopted, it is necessary to ensure the real-time and security of data transmission to avoid data loss or damage. Data backup is another important step. Regularly backing up image data can ensure the security and integrity of data. Backup can be achieved by external storage devices or cloud storage services. External storage devices have higher storage capacity and access speed, while cloud storage services are more flexible and convenient. No matter which method is adopted, it is necessary to ensure the integrity and recoverability of backup data to avoid data loss caused by equipment failure or human error. File naming and database storage are also important links in data management. Naming image files according to certain rules can facilitate subsequent search and management. For example, information such as date, time and location can be used as part of the file name to quickly locate a specific image file. Storing image data in a database can achieve centralized management and query of image data. The database can record the acquisition time, location, parameter settings and other information of each image to facilitate subsequent quality inspection and data analysis.

[0036] In the method for preparing the rare earth reflective heat-insulating metal plate, the step S2 is to extract the texture features of the coating surface state image based on the grid granularity to obtain a set of the coating surface grid granularity texture feature vectors. Figure 2 FIG. 1 is a flow chart of step S2 in the method for preparing a rare earth reflective heat-insulating metal plate according to an embodiment of the present application. Figure 2 As shown, the step S2 includes: S21, grid-slicing the coating surface state image to obtain a set of coating surface state grid granularity image blocks; S22, inputting each coating surface state grid granularity image block in the set of coating surface state grid granularity image blocks into a texture feature extractor based on a multi-layer void convolution structure to obtain a set of coating surface grid granularity texture feature vectors.

[0037] Specifically, the step S21, grid segmentation is performed on the coating surface state image to obtain a set of grid granularity image blocks of the coating surface state. That is, in order to fully capture the detailed information of the coating surface, the present application further grid segments the coating surface state image, so as to refine the granularity of feature extraction, facilitate detailed analysis of the coating features in each grid area, so as to more accurately evaluate the uniformity and density of the coating. In addition, compared to directly processing the entire image, the amount of data processed in a single time can be reduced by dividing the image into multiple small blocks, thereby reducing the complexity of the calculation and the time required, and improving the real-time performance of the detection. In addition, grid segmentation helps to capture local details of the coating surface, such as tiny bubbles, cracks and unevenness, which may be ignored in large-size images. By analyzing the texture features in each grid area, the overall state of the coating surface can be more fully understood.

[0038] In an optional embodiment of the present application, the grid segmentation method is uniform segmentation, that is, the image is evenly divided into several small blocks of the same size. An image of 1024x1024 pixels can be divided into 16x16 small blocks of 64x64 pixels. The advantage of uniform segmentation is that it is simple and easy to implement, and the size and shape of each small block are consistent, which is convenient for subsequent processing and analysis. The specific operation steps are as follows: First, determine the grid size. According to the characteristics of the coating surface and the detection requirements, select a suitable grid size. For example, a grid size of 64x64 pixels can be selected. Then, calculate the number of grids. According to the total size and grid size of the image, calculate the number of grids that need to be segmented. For example, for an image of 1024x1024 pixels, it is divided into a grid of 64x64 pixels, and 16x16 grids can be obtained. Finally, segment the image. Use image processing software or an image processing library in a programming language to segment the image, and evenly divide the large-size image into multiple small blocks.

[0039] In an optional embodiment of the present application, the grid segmentation method is adaptive segmentation, which performs dynamic segmentation according to the local features of the image, and can divide smaller grids in areas with complex textures, and divide larger grids in areas with simple textures. The advantage of adaptive segmentation is that it can handle areas of different complexity more flexibly, and improve the pertinence and accuracy of feature extraction. The specific operation steps are as follows: First, feature extraction is performed. A preliminary feature extraction is performed on the image. For example, edge detection, texture analysis and other methods can be used to extract local features of the image. Then, feature analysis is performed. Based on the extracted features, the complexity of each area is analyzed. For example, the edge density or texture complexity of each area can be calculated. Finally, the grid is dynamically divided. Based on the results of the feature analysis, the grid is dynamically divided. For areas with high complexity, smaller grids are divided; for areas with low complexity, larger grids are divided.

[0040] In an optional embodiment of the present application, the grid segmentation method is overlapping segmentation, and overlapping segmentation refers to allowing adjacent small blocks to have a certain overlapping area during segmentation. For example, an image of 1024x1024 pixels can be segmented into 16x16 small blocks of 64x64 pixels, but there is an overlap of 16 pixels between each small block. The advantage of overlapping segmentation is that it can avoid the features at the boundary from being truncated, ensuring that each feature can be fully analyzed in multiple small blocks. The specific operation steps are as follows: First, determine the grid size and overlap ratio. According to the characteristics of the coating surface and the detection requirements, select the appropriate grid size and overlap ratio. For example, a grid size of 64x64 pixels can be selected, and the overlap ratio is 25%. Then, calculate the number of grids. According to the total size, grid size and overlap ratio of the image, calculate the number of grids that need to be segmented. For example, for an image of 1024x1024 pixels, it is segmented into a grid of 64x64 pixels, with an overlap ratio of 25%, and 21x21 grids can be obtained. Finally, segment the image. Use image processing software or image processing libraries in programming languages ​​to segment images and divide large images into multiple small blocks with overlapping areas.

[0041] Specifically, in step S22, each coating surface state grid granularity image block in the set of the coating surface state grid granularity image blocks is input into a texture feature extractor based on a multi-layer hole convolution structure to obtain a set of the coating surface grid granularity texture feature vectors. It should be understood that, considering that the coating surface state feature has a multi-scale characteristic, the present application further uses a texture feature extractor based on a multi-layer hole convolution structure to extract features from each coating surface state grid granularity image block in the set of the coating surface state grid granularity image blocks. Those of ordinary skill in the art should know that the hole convolution can expand the receptive field without reducing the spatial resolution by inserting holes in the convolution kernel, which helps to retain more detail information. The multi-layer hole convolution structure forms a hierarchical feature representation by learning and abstracting texture features layer by layer, which can not only capture the local detail information in the coating surface state grid granularity image block, but also effectively integrate the contextual information in a larger range, which helps to understand the texture state of the coating surface more comprehensively.

[0042] Specifically, texture feature extraction is the process of extracting information describing surface texture from an image. Texture features can reflect the local structure and pattern of an image, which is particularly important for defect detection on coating surfaces. Common texture feature extraction methods include gray-level co-occurrence matrix (GLCM), local binary pattern (LBP), and deep learning-based methods. In the technical solution of this application, a texture feature extractor based on a multi-layer hole convolution structure is used. This method combines the powerful representation ability of a convolutional neural network (CNN) and the extended receptive field characteristics of a hole convolution, and can more effectively capture the multi-scale texture features of the coating surface. The core of texture feature extraction is to extract feature vectors that can describe surface texture from an image. These feature vectors can be used for subsequent quality inspection and classification tasks. For example, the gray-level co-occurrence matrix (GLCM) extracts statistical features describing texture by calculating the gray value relationship between pixel pairs in an image; the local binary pattern (LBP) extracts local pattern features describing texture by comparing each pixel with its neighboring pixels. However, these traditional methods have limitations in processing complex textures and multi-scale features, and deep learning-based methods, especially multi-layer hole convolution structures, can overcome these limitations.

[0043] The multi-layer dilated convolution structure is a special convolutional neural network architecture that can expand the receptive field without reducing the spatial resolution by inserting holes in the convolution kernel (i.e. skipping some pixels). This feature enables the multi-layer dilated convolution structure to capture a wider range of contextual information while retaining more detail information. Specifically, the multi-layer dilated convolution structure has the following characteristics: The dilated convolution expands the receptive field of each convolution kernel by inserting holes in the convolution kernel, so that it can capture a wider range of contextual information. This is very beneficial for capturing multi-scale texture features of the coating surface. For example, for a 3x3 convolution kernel, if the dilation rate is 1, the receptive field is 3x3; if the dilation rate is 2, the receptive field is 7x7. Through multi-layer dilated convolution, the receptive field can be gradually expanded to capture texture information of different scales. Unlike ordinary convolution operations, dilated convolution does not reduce the spatial resolution of the image, so it can retain more detail information. This is very important for detecting tiny defects (such as bubbles, cracks, etc.) on the coating surface. Ordinary convolution reduces the spatial resolution of the image after each convolution operation, while the atrous convolution maintains the spatial resolution of the image by inserting holes, so that the detail information can be retained. The multi-layer atrous convolution structure forms a hierarchical feature representation by learning and abstracting texture features layer by layer. This hierarchical feature representation can not only capture local detail information, but also effectively integrate contextual information in a larger range, which helps to more comprehensively understand the texture state of the coating surface. For example, the shallow convolution layer can capture local detail features, while the deep convolution layer can capture contextual information in a larger range, forming a multi-level feature representation.

[0044] In an optional embodiment of the present application, a multi-layer dilated convolution structure is composed of multiple convolution layers, and each convolution layer uses a dilated convolution kernel of different sizes. In the present application, a 3x3 convolution kernel can be used, and the dilation rate is 1, 2, 4, etc., to gradually expand the receptive field. Each convolution layer is followed by an activation function (such as ReLU) and a pooling layer (such as maximum pooling) to enhance the nonlinear representation of features and reduce the size of feature maps. In a multi-layer dilated convolution structure, the convolution operation of each layer will extract texture features of different scales. The first convolution layer uses a 3x3 convolution kernel with a dilation rate of 1, which can capture local detail features; the second convolution layer uses a 3x3 convolution kernel with a dilation rate of 2, which can capture a slightly larger range of context information; the third convolution layer uses a 3x3 convolution kernel with a dilation rate of 4, which can capture a larger range of context information. Through multi-layer convolution operations, high-level texture features are gradually abstracted. In each convolution layer, the dilated convolution operation will capture texture features of different scales. For example, a smaller void ratio (such as 1) can capture local details, and a larger void ratio (such as 4) can capture a wider range of contextual information. Through multiple layers of convolution operations, high-level texture features are gradually abstracted. These features can more comprehensively describe the texture state of the coating surface.

[0045] In the multi-layer dilated convolution structure, each convolution operation will generate a new feature map. These feature maps contain rich texture information and reflect the different scale characteristics of the coating surface. For example, the feature map generated by the first convolution layer may reflect local detail features, the feature map generated by the second convolution layer may reflect a slightly larger range of context information, and the feature map generated by the third convolution layer may reflect an even larger range of context information. Through multi-layer convolution operations, high-level texture features are gradually abstracted to form a multi-level feature representation. After the last layer of convolution operation, the obtained feature map is flattened into a one-dimensional vector. Each image block will correspond to a feature vector, which contains rich texture information.

[0046] In the method for preparing the rare earth reflective insulation metal plate, the step S3 performs feature significance modulation dynamic aggregation on the set of the coating surface grid granularity texture feature vectors to obtain the coating surface texture grid granularity global significant aggregation representation vector. It should be understood that since the quality assessment of the coating surface needs to take into account its overall state, in order to reveal the global texture distribution characteristics of the coating surface, the present application further performs global aggregation analysis on the set of the coating surface grid granularity texture feature vectors to mine the global significant texture features of the coating surface, which is used as a reference benchmark to achieve the evaluation of the texture feature difference distribution of each local grid area, thereby effectively revealing the uniformity and density of the coating. Accordingly, the present application proposes a dynamic aggregation method based on a feature distribution field to process the set of the coating surface grid granularity texture feature vectors, which dynamically adjusts the feature aggregation weight based on the correlation action mode between the grid granularity texture feature vectors of each coating surface, thereby effectively mining the global significant texture feature pattern of the coating surface.

[0047] Figure 3 FIG. 1 is a flow chart of step S3 in the method for preparing the rare earth reflective heat-insulating metal plate according to an embodiment of the present application. Figure 3 As shown, the step S3 comprises: S31, determining the initial center vector of the coating surface grid grain size texture feature clustering based on the characteristic distribution field of the set of the coating surface grid grain size texture feature vectors; S32, based on the spatial span of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vectors relative to the initial center vector of the coating surface grid grain size texture feature clustering, performing significance modulation dynamic aggregation on the set of the coating surface grid grain size texture feature vectors to obtain the coating surface texture grid grain global significance aggregation representation vector.

[0048] Specifically, the step S31 includes: calculating the static energy factor of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector to obtain a set of static energy factors of the coating surface grid grain size texture feature; selecting the coating surface grid grain size texture feature vector corresponding to the maximum value in the set of static energy factors of the coating surface grid grain size texture feature as the initial center vector of the coating surface grid grain size texture feature clustering. Among them, calculating the static energy factor of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector to obtain a set of static energy factors of the coating surface grid grain size texture feature includes: calculating the kurtosis of the coating surface grid grain size texture feature vector, and inputting the kurtosis into the sigmoid activation function to obtain the static energy factor of the coating surface grid grain size texture feature.

[0049] That is, firstly, the static energy factor of each coating surface grid grain size texture feature vector is calculated based on its kurtosis to evaluate its stability and importance in the feature space, and the coating surface grid grain size texture feature vector corresponding to the largest static energy factor is selected as the initial center vector of the coating surface grid grain size texture feature clustering to determine the cluster core of the global texture feature distribution of the coating surface, thereby providing a stable reference point for subsequent feature aggregation.

[0050] The above step S31 can be expressed as follows:

[0051]

[0052]

[0053]

[0054] in, represents the set of grid granularity texture feature vectors of the coating surface, , , , and Respectively represent the first, second, and third in the set of the coating surface grid grain texture feature vectors. , and The coating surface grid granularity texture feature vector, The value of is the number of the grid granularity texture feature vectors on the coating surface. Indicates the The coating surface grid granularity texture feature vector The eigenvalues ​​at the positions, and Respectively represent the The characteristic mean and square of the characteristic variance of the grid grain texture feature vector of the coating surface are It means to calculate the expected value of a set. represents the sigmoid activation function, Indicates the The static energy factor of the coating surface grid granularity texture feature vector, Indicates the index corresponding to the maximum value. represents the index of the largest static energy factor in the set of static energy factors of the coating surface grid granularity texture features, The coating surface mesh granularity texture feature vector corresponding to the index of the largest static energy factor in the set of static energy factors of the coating surface mesh granularity texture feature is represented. Represents the initial center vector of the clustering of the grid grain size texture features on the coating surface.

[0055] Specifically, the step S32 includes: based on the spatial span between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the initial center vector of the coating surface grid grain size texture feature cluster, and the set of static energy factors of the coating surface grid grain size texture feature, calculating the dynamic aggregation energy factor of each coating surface grid grain size texture feature vector to obtain a set of dynamic aggregation energy factors of the coating surface grid grain size texture feature; inputting the set of dynamic aggregation energy factors of the coating surface grid grain size texture feature into a gated mask unit to obtain a set of dynamic aggregation weight factors of the coating surface grid grain size texture feature; calculating the weighted sum of the set of the coating surface grid grain size texture feature vectors based on the set of dynamic aggregation weight factors of the coating surface grid grain size texture feature to obtain the global significant aggregation representation vector of the coating surface texture grid grain size.

[0056] In a specific example of the present application, the dynamic aggregation energy factors of the individual coating surface grid grain size texture feature vectors are calculated to obtain a set of dynamic aggregation energy factors of the coating surface grid grain size texture features, including: taking the square value of the number of feature vectors between the coating surface grid grain size texture feature vector and the initial center vector of the coating surface grid grain size texture feature cluster as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the coating surface grid grain size texture feature vector and the static energy factor of the initial center vector of the coating surface grid grain size texture feature cluster and the spatial span coefficient to obtain the dynamic aggregation energy factor of the coating surface grid grain size texture features.

[0057] In a specific example of the present application, the set of dynamic aggregation energy factors of the grid granularity texture features of the coating surface is input into a gated mask unit, including: inputting the set of dynamic aggregation energy factors of the grid granularity texture features of the coating surface into a sigmoid function for normalization processing to obtain a set of normalized dynamic aggregation energy factors of the grid granularity texture features of the coating surface; based on a preset mask threshold, deactivating the set of normalized dynamic aggregation energy factors of the grid granularity texture features of the coating surface to obtain a set of dynamic aggregation weight factors of the grid granularity texture features of the coating surface.

[0058] That is, considering that adjacent areas in an image often have more similar texture characteristics, the present application further derives the dynamic aggregation energy factor of each coating surface grid granularity texture feature vector based on the spatial span of each coating surface grid granularity texture feature vector relative to the initial center vector of the coating surface grid granularity texture feature cluster, and combines the static energy factors of both. That is, by considering the feature space proximity of each coating surface grid granularity texture feature vector relative to the core of the coating surface global texture feature distribution cluster and the importance difference of the intrinsic attributes, its contribution to the coating surface global texture feature pattern is more accurately measured. Then, based on the gated mask mechanism, the obtained dynamic aggregation energy factor is nonlinearly transformed and screened to generate corresponding weights, so as to perform weighted aggregation on the set of the coating surface grid granularity texture feature vectors, thereby enhancing the expression of important features and suppressing the influence of irrelevant or noise features, so as to obtain a representative coating surface texture grid granularity global significant aggregation representation vector.

[0059] The above step S32 can be expressed as follows:

[0060]

[0061]

[0062]

[0063]

[0064] in, Indicates the The static energy factor of the coating surface grid granularity texture feature vector, represents the static energy factor of the initial center vector of the clustering of the grid granularity texture features on the coating surface, and For different weight parameters, Indicates the The number of feature vectors between the coating surface grid granularity texture feature vector and the initial center vector of the coating surface grid granularity texture feature cluster, Indicates the The dynamic aggregation energy factor of the coating surface grid granularity texture feature vector, Indicates Normalized coating surface grid granularity texture feature dynamic aggregation energy factor, Indicates mask processing, is the preset mask threshold, The value of is the number of the grid granularity texture feature vectors on the coating surface. Indicates Dynamic aggregation weight factor of the coating surface grid granularity texture features, The global significant aggregation representation vector of the grid granularity representing the coating surface texture.

[0065] In the method for preparing the rare earth reflective heat-insulating metal plate, the step S4 generates the roller coating quality detection result based on the characteristic difference of each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vector relative to the coating surface texture grid grain global significant aggregation representation vector, wherein the roller coating quality detection result is used to indicate whether the coating surface is dense. Figure 4 FIG. 4 is a flow chart of step S4 in the method for preparing the rare earth reflective heat-insulating metal plate according to an embodiment of the present application. Figure 4 As shown, the step S4 includes: S41, calculating the coating surface state semantic difference coefficient between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the coating surface texture grid grain size global significant aggregation representation vector to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients; S42, inputting the coating surface state global distribution representation vector into a classifier-based roller coating quality detection module to obtain the roller coating quality detection result.

[0066] Specifically, the S41 calculates the coating surface state semantic difference coefficient between each coating surface grid granularity texture feature vector in the set of the coating surface grid granularity texture feature vector and the coating surface texture grid granularity global significant aggregation representation vector to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients. That is, by further calculating the semantic difference coefficient between each coating surface grid granularity texture feature vector and the coating surface texture grid granularity global significant aggregation representation vector, the difference between the texture features of each local area in the coating surface state image relative to the main texture features of the image as a whole is quantified, and then based on each coating surface state semantic difference coefficient, the global texture distribution pattern of the coating surface is revealed, and the overall texture distribution uniformity information of the coating surface state image is captured, thereby providing an important basis for the density evaluation of the coating. In a specific example of the present application, the Mahalanobis distance between each coating surface grid granularity texture feature vector in the set of the coating surface grid granularity texture feature vector and the coating surface texture grid granularity global significant aggregation representation vector is calculated as the coating surface state semantic difference coefficient to obtain the coating surface state global distribution representation vector.

[0067] Specifically, in S42, the coating surface state global distribution characterization vector is input into a roller coating quality detection module based on a classifier to obtain the roller coating quality detection result. In the technical solution of the present application, the classifier can effectively learn the coating surface texture distribution uniformity information contained in the coating surface state global distribution characterization vector through training and learning, and distinguish the dense and non-dense states of the coating surface accordingly, thereby realizing automatic classification and evaluation of roller coating quality, providing real-time feedback for the production process, guiding production adjustments, and ensuring the consistency and reliability of coating quality.

[0068] In the technical solution of the present application, since the coating surface texture grid granularity global significant aggregation representation vector is obtained by clustering analysis of the set of the coating surface grid granularity texture feature vectors, it is not the true surface state semantic measurement center. Therefore, when using the coating surface texture grid granularity global significant aggregation representation vector to calculate the coating surface state semantic difference coefficient of each coating surface grid granularity texture feature vector in the set of the coating surface grid granularity texture feature vectors, additional interference information will be introduced. The interference component of the coating surface state global distribution representation vector will make the overall feature manifold of the coating surface state global distribution representation vector in the high-dimensional feature space ensure the existence of fine-grained feature structure holes. The existence of fine-grained feature structure holes will not only make the features of the coating surface state global distribution representation vector corresponding to the semantic coverage of the class probability label insufficient, but also cause the outlier class regression inference mapping to miss the target, affecting the accuracy of the roller coating quality detection result obtained by inputting the coating surface state global distribution representation vector into the classifier-based roller coating quality detection module.

[0069] In the technical solution of the present application, the coating surface state global distribution characterization vector is input into a roller coating quality detection module based on a classifier to obtain a roller coating quality detection result, including: inputting the coating surface state global distribution characterization vector into a pre-classifier based on a Softmax function to obtain a global fine-grained class probability label vector of the coating surface state; Calculating the product between the coating surface state global fine-grained class probability label vector and its transposed vector to obtain the coating surface state global fine-grained full label domain modulation matrix; Dividing the eigenvalues ​​of each position in the global fine-grained full-label domain modulation matrix of the coating surface state by the square root of the scale value of the global distribution characterization vector of the coating surface state to obtain the global fine-grained full-label domain scale modulation matrix of the coating surface state; Inputting the coating surface state global fine-grained full-label domain scale modulation matrix into a quasi-probability domain attention sparse module based on a multi-level gating function to obtain the coating surface state global fine-grained full-label domain scale modulation matrix; Taking the coating surface state global distribution characterization vector as a query feature vector, matrix-multiplying the coating surface state global fine-grained full-label domain scale modulation matrix with the coating surface state global distribution characterization vector to obtain an optimized coating surface state global distribution characterization vector; and The optimized coating surface state global distribution characterization vector is input into the classifier-based roller coating quality detection module to obtain the roller coating quality detection result.

[0070] The optimization process of the global distribution characterization vector of the coating surface state is expressed by the formula:

[0071]

[0072]

[0073]

[0074]

[0075] in, The first vector representing the global distribution characterization vector of the coating surface state The eigenvalues ​​at the positions, represents the scale value of the global distribution characterization vector of the coating surface state, The first vector representing the global fine-grained class probability label vector of the coating surface state The eigenvalues ​​at the positions, Represents the global fine-grained class probability label vector of the coating surface state, represents matrix multiplication, represents the transpose of a vector, Represents the global fine-grained full-label domain scale modulation matrix of the coating surface state, represents a predetermined threshold, The first part of the global fine-grained full-label domain scale modulation matrix representing the coating surface state The eigenvalues ​​at the positions, represents the predetermined hyperparameters, represents the mask function, The first part of the sparse modulation matrix representing the global fine-grained full-label domain scale of the coating surface state The eigenvalues ​​at the positions, Represents the global fine-grained full-label domain-scale sparse modulation matrix of the coating surface state, Represents the optimized global distribution characterization vector of the coating surface state.

[0076] Therefore, by adding a class probability domain-level optimizable disturbance to the original feature vector to strengthen the dependency entanglement between different variables in the original feature vector based on label domain modulation, the source domain feature vector can be optimized based on class probability query-type scale modulation. In this way, the significant inherent modal information in the source domain features can be more effectively retained, and the interference components in the feature distribution can be effectively masked, thereby improving the adversarial robustness of the manifold expression of the feature vector, thereby improving the accuracy of the roller coating quality detection result obtained by inputting the classifier-based roller coating quality detection module.

[0077] According to another aspect of the present application, a rare earth reflective heat insulating metal plate is also provided, and the rare earth reflective heat insulating metal plate is prepared by the preparation method of the rare earth reflective heat insulating metal plate as described above. After testing, the performance indicators of the rare earth reflective heat insulating coating are shown in Table 1: Table 1 Performance indicators of rare earth reflective insulation metal panels

[0078] It can be seen from the data in Table 1 that the rare earth reflective insulation metal plate described in the embodiment of the present application has excellent insulation performance and reflection ability, can effectively reduce the temperature of the metal plate surface, and maintain good physical and chemical stability. In addition, the rare earth reflective insulation metal plate also has good waterproof and aging resistance, ensuring the long-term stability and reliability of the coating in various environments. Its salt spray resistance also shows that the metal plate can effectively resist corrosion and extend its service life when used in a marine environment or areas with more salt spray. At the same time, the drying time is short and the coating thickness is moderate, which meets the requirements for fast drying and coating protection performance in industrial applications.

[0079] Specifically, the reflective heat-insulating coating has the following technical features: First of all, it uses double reflection technology, with a reflectivity of >92%, and full-band reflection from 150-2000nm. Ordinary traditional reflective coatings only have single reflection and do not have the ability of double reflection. Rare earth composite materials are used to reflect the sunlight spectrum band, and the molecular microstructure forms an overlapping double-layer film system, which changes the single reflection of the coating and achieves the purpose of double reflection. When sunlight shines on the coating, the first layer of paint film will reflect and remove more than 85% of infrared, ultraviolet and visible light, and the second layer of paint film will reflect and remove the light that passes through the first layer of paint film, increasing the reflectivity by another 5-8%. Under the action of double reflection, it can effectively prevent the accumulation of solar heat energy and inhibit the accumulation of heat energy; under double reflection, the average reflectivity is >90% Secondly, the hemispherical emissivity of the reflective heat-insulating coating is greater than 85%, the infrared heat radiation is enhanced, the heat radiation dissipates heat and cools down, and the radiation band is: Excellent reflective sunscreen and cooling paint must have excellent spatial thermal radiation performance. Using rare earth composite coating ultrafine hollow microsphere technology and materials, curing into a film, forming a special micro-convex and concave high-radiation surface, the surface area is increased several times, and the solar heat that is not reflected and removed from the surface is radiated in the 2-15.5μm band, further reducing the surface temperature and achieving effective heat dissipation and cooling control.

[0080] In addition, the reflective heat-insulating coating has a vacuum heat shielding layer, which blocks heat, cuts off heat, and prevents heat exchange and conduction. It adopts micro-bead vacuum shielding layer composite structure technology to reduce the thermal conductivity of the coating and reduce heat conduction. The solidification forms a tangential bonding of the spatial three-dimensional structure. Particles of different particle sizes will fill the gaps between molecules and eventually form a composite heat shielding layer, which effectively prevents heat exchange and conduction between internal and external heat energy and avoids internal temperature rise. A 1mm thick rare earth reflective heat-insulating coating layer reflects about 90%-95% of radiant heat, which is equivalent to a 10mm thick polystyrene foam with an R value of 20.

[0081] Thirdly, the reflective heat insulation coating has both water-based coating and oil-based coating technology. The oil-based coating technology has the excellent characteristics of high weather resistance, aging resistance, long life, salt spray resistance, waterproof, corrosion resistance and strong adhesion.

[0082] In addition, the reflective heat-insulating coating is suitable for spraying and roller coating, and is suitable for making pre-roller-coated metal plates. The film thickness is controlled at 80um-200um, and factory production can be achieved. It has strong machinability and is easy to punch, shear and fold, overcoming the weaknesses of on-site construction.

[0083] In particular, when the reflective thermal insulation coating is suitable for making pre-rolled metal plates and obtaining the rare earth reflective thermal insulation metal plates, since the thickness of the reflective thermal insulation coating is only 80um-200um, the prepared rare earth reflective thermal insulation metal plates are easy to punch, cut, fold and roll, thereby being convenient for carrying, packaging and processing.

[0084] Finally, it should be noted that the embodiments described above are only some embodiments of the present invention, rather than all embodiments. The detailed description of the embodiments of the present invention is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Claims

1. A method for preparing a rare earth reflective heat-insulating metal plate, characterized in that: include: Provide rare earth reflective thermal insulation coatings and metal panels; performing surface treatment on the metal plate to obtain a surface-treated metal plate; Applying the rare earth reflective heat-insulating coating to the surface-treated metal plate by a roller coating process to obtain a rare earth reflective heat-insulating metal plate; Performing a roller coating quality test on the rare earth reflective heat-insulating metal plate to obtain a roller coating quality test result; Among them, the roller coating quality inspection is carried out on the rare earth reflective insulation metal plate to obtain the roller coating quality inspection result, including: acquiring a coating surface state image collected by an industrial camera; performing grid granularity-based texture feature extraction on the coating surface state image to obtain a set of coating surface grid granularity texture feature vectors; performing feature significance modulation dynamic aggregation on the set of coating surface grid granularity texture feature vectors to obtain a coating surface texture grid granularity global significant aggregation representation vector; based on the feature difference of each coating surface grid granularity texture feature vector in the set of coating surface grid granularity texture feature vectors relative to the coating surface texture grid granularity global significant aggregation representation vector, the roller coating quality inspection result is generated, wherein the roller coating quality inspection result is used to indicate whether the coating surface is dense.

2. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 1, characterized in that: Extracting texture features based on grid granularity from the coating surface state image to obtain a set of coating surface grid granularity texture feature vectors, including: Grid-slicing the coating surface state image to obtain a set of coating surface state grid granularity image blocks; Each coating surface state grid granularity image block in the set of coating surface state grid granularity image blocks is input into a texture feature extractor based on a multi-layer hole convolution structure to obtain a set of coating surface grid granularity texture feature vectors.

3. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 2, characterized in that: The set of the coating surface grid granularity texture feature vectors is subjected to feature saliency modulation dynamic aggregation to obtain a coating surface texture grid granularity global saliency aggregation representation vector, including: Determining an initial center vector of a cluster of grid grain size texture feature of the coating surface based on a characteristic distribution field of a set of grid grain size texture feature vectors of the coating surface; Based on the spatial span of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vectors relative to the initial center vector of the coating surface grid grain size texture feature cluster, the set of the coating surface grid grain size texture feature vectors is dynamically aggregated by significance modulation to obtain the coating surface texture grid grain size global significant aggregation representation vector.

4. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 3, characterized in that: Based on the characteristic distribution field of the set of the coating surface grid grain size texture feature vectors, determining the initial center vector of the coating surface grid grain size texture feature clustering includes: Calculating the static energy factor of each coating surface grid grain size texture feature vector in the set of coating surface grid grain size texture feature vectors to obtain a set of coating surface grid grain size texture feature static energy factors; The coating surface grid grain size texture feature vector corresponding to the maximum value in the set of static energy factors of the coating surface grid grain size texture feature is selected as the initial center vector of the coating surface grid grain size texture feature clustering.

5. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 4, characterized in that: Calculating the static energy factor of each coating surface grid grain size texture feature vector in the set of coating surface grid grain size texture feature vectors to obtain a set of coating surface grid grain size texture feature static energy factors, including: The kurtosis of the coating surface grid grain size texture feature vector is calculated, and the kurtosis is input into a sigmoid activation function to obtain the coating surface grid grain size texture feature static energy factor.

6. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 5, characterized in that: Based on the spatial span of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vectors relative to the initial center vector of the coating surface grid grain size texture feature clustering, the set of the coating surface grid grain size texture feature vectors is dynamically aggregated by significance modulation to obtain the coating surface texture grid grain size global significant aggregation representation vector, including: Based on the spatial span between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vectors and the initial center vector of the coating surface grid grain size texture feature cluster, and the set of the coating surface grid grain size texture feature static energy factors, the dynamic aggregation energy factors of each coating surface grid grain size texture feature vector are calculated to obtain the set of the coating surface grid grain size texture feature dynamic aggregation energy factors; Inputting the set of dynamic aggregation energy factors of the grid granularity texture features of the coating surface into a gated mask unit to obtain a set of dynamic aggregation weight factors of the grid granularity texture features of the coating surface; The weighted sum of the set of the coating surface grid grain texture feature vectors is calculated based on the set of the coating surface grid grain texture feature dynamic aggregation weight factors to obtain the coating surface texture grid grain global significant aggregation representation vector.

7. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 6, characterized in that: Calculating the dynamic aggregation energy factor of each coating surface grid grain size texture feature vector to obtain a set of coating surface grid grain size texture feature dynamic aggregation energy factors, including: The square value of the number of feature vectors between the coating surface grid grain size texture feature vector and the initial center vector of the coating surface grid grain size texture feature cluster is used as the spatial span coefficient, and the weighted ratio between the product of the static energy factor of the coating surface grid grain size texture feature vector and the static energy factor of the initial center vector of the coating surface grid grain size texture feature cluster and the spatial span coefficient is calculated to obtain the dynamic aggregation energy factor of the coating surface grid grain size texture feature.

8. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 7, characterized in that: The roller coating quality detection result is generated based on the characteristic difference of each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vectors relative to the coating surface texture grid grain size global significant aggregation representation vector, including: Calculating the coating surface state semantic difference coefficient between each coating surface grid grain size texture feature vector in the set of the coating surface grid grain size texture feature vector and the coating surface texture grid grain size global significant aggregation representation vector to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients; The coating surface state global distribution characterization vector is input into a roller coating quality detection module based on a classifier to obtain the roller coating quality detection result.

9. The method for preparing the rare earth reflective heat-insulating metal plate according to claim 8, characterized in that: Calculating the coating surface state semantic difference coefficient between each coating surface grid granularity texture feature vector in the set of the coating surface grid granularity texture feature vectors and the coating surface texture grid granularity global significant aggregation representation vector to obtain a coating surface state global distribution representation vector composed of multiple coating surface state semantic difference coefficients, including: The Mahalanobis distance between each coating surface grid grain texture feature vector in the set of the coating surface grid grain texture feature vectors and the coating surface texture grid grain global significant aggregation representation vector is calculated as the coating surface state semantic difference coefficient to obtain the coating surface state global distribution representation vector.

10. A rare earth reflective heat-insulating metal plate, characterized in that: The rare earth reflective heat-insulating metal plate is prepared by the preparation method of the rare earth reflective heat-insulating metal plate as described in any one of claims 1-9.

Images