Rare earth reflective thermal insulation coating and preparation method thereof

Abstract

The invention relates to the technical field of rare earth coatings, and particularly discloses a rare earth reflective thermal insulation coating and a preparation method thereof.In the viscosity adjusting process of the rare earth reflective thermal insulation coating, firstly, a corresponding mixed material target state image is generated according to a target viscosity value input by a user; carrying out image feature extraction and feature enhancement processing on the mixed material target state image and the actual mixed material surface state image by adopting an image processing technology based on deep learning so as to extract material surface state features of the mixed material target state image and the actual mixed material surface state image; and the application amount of the thickening agent is intelligently recommended based on the characteristic difference between the actual surface state of the mixed material and the target state. In this way, optimal control over the application amount of the thickening agent can be achieved, and the production efficiency and the quality stability of the coating are improved.

Description

Technical Field

[0001] The present application relates to the technical field of rare earth coatings, and more specifically, to a rare earth reflective heat-insulating coating and a preparation method thereof. Background Art

[0002] With the increasing requirements for building energy conservation, thermal insulation coatings have attracted widespread attention due to their excellent heat reflection performance. Especially in the high temperature environment in summer, thermal insulation coatings can effectively reduce the internal temperature of buildings, reduce air conditioning energy consumption, and improve living comfort. However, traditional thermal insulation coatings have problems such as low reflectivity and single color, which makes it difficult to meet the diverse needs of the market.

[0003] In this regard, the invention patent with publication number CN115558356A proposes a rare earth reflective thermal insulation coating, which solves the problem of low reflectivity of existing colored coatings by adding specific rare earth pigments and fillers as well as other auxiliary materials such as dispersants, wetting agents, defoaming agents, water, film-forming aids, thickeners, etc., and meets the demand for color diversity. At the same time, patent CN115558356A also proposes a preparation method of rare earth reflective thermal insulation coating, which includes: step 1: adding solvent, dispersant and defoaming agent to a dispersion tank, and dispersing at a low speed until uniform; step 2: continuing to add rare earth pigments and precipitated barium sulfate to the dispersion tank, and dispersing at a high speed until uniform to obtain a first mixed material; step 3: grinding the first mixed material to obtain a first mixed material powder; step 4: mixing the first mixed material powder, pure acrylic emulsion, wetting agent and film-forming aid and stirring evenly to obtain a second mixed material; step 5: adding a thickener to adjust the viscosity of the second mixed material to obtain a third mixed material; step 6: adding a bactericide to the third mixed material and stirring until uniform to obtain a rare earth reflective thermal insulation coating.

[0004] However, in the actual production process of rare earth reflective thermal insulation coatings, the viscosity adjustment of the coating is a crucial link. Specifically, the appropriateness of the coating viscosity is directly related to the coating's construction performance, drying speed, and final coating quality. For example, if the viscosity is too high, the coating is prone to sagging during spraying or brushing; if the viscosity is too low, the coating may not form a uniform coating, affecting the thermal insulation effect.

[0005] At present, the adjustment of coating viscosity mainly depends on the operator's manual experience and repeated trials. This method is not only inefficient, but also prone to errors, making it difficult to ensure the consistency and stability of coating quality. Especially in rare earth reflective thermal insulation coatings with complex formula ingredients and a wide variety of pigments and fillers, precise control of viscosity has become a technical difficulty.

[0006] Therefore, an optimized rare earth reflective thermal insulation coating and a preparation method thereof are expected. 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 thermal insulation coating and a preparation method thereof. In the viscosity adjustment process of the rare earth reflective thermal insulation coating, the corresponding target state image of the mixed material is first 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 image features and enhance features of the target state image of the mixed material and the actual surface state image of the mixed material, so as 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 to be 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.

[0008] Correspondingly, according to one aspect of the present application, a method for preparing a rare earth reflective thermal insulation coating is provided, comprising: step 1: adding a solvent, a dispersant and a defoaming agent to a dispersion tank, and dispersing at a low speed until uniform, wherein the solvent is toluene; step 2: continuing to add rare earth pigments and precipitated barium sulfate to the dispersion tank, and dispersing at a high speed until uniform to obtain a first mixed material; step 3: grinding the first mixed material to obtain a first mixed material powder; step 4: mixing the first mixed material powder, pure acrylic resin, a wetting agent and a film-forming aid and stirring them uniformly to obtain a second mixed material; step 5: adding a thickener to adjust the viscosity of the second mixed material to obtain a third mixed material; step 6: adding a bactericide to the third mixed material and stirring until uniform to obtain a rare earth reflective thermal insulation coating.

[0009] In the above-mentioned method for preparing the rare earth reflective thermal insulation coating, the step 5 also includes: receiving a target viscosity value input by a user, and simultaneously acquiring a surface state image of the second mixture material collected by an industrial camera; generating a second mixture material target state generation image based on the target viscosity value; respectively extracting the material surface state features of the second mixture material target state generation image and the second mixture material surface state image to obtain a second mixture material target state image semantic coding feature map and a second mixture material surface state image semantic coding feature map; performing pixel-level semantic association spatial enhancement on the second mixture material target state image semantic coding feature map and the second mixture material surface state image semantic coding feature map to obtain a second mixture material target state image semantic space enhancement coding feature map and a second mixture material surface state image semantic space enhancement coding feature map; based on the feature difference between the second mixture material target state image semantic space enhancement coding feature map and the second mixture material surface state image semantic space enhancement coding feature map, generating a recommended value for the amount of thickener to be applied.

[0010] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, a second mixed material target state generation image is generated based on the target viscosity value, including: performing one-hot encoding on the target viscosity value to obtain a target viscosity one-hot encoding vector; inputting the target viscosity one-hot encoding vector into a second mixed material target state image generator based on a diffusion model to obtain the second mixed material target state generation image.

[0011] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, the material surface state characteristics of the second mixture material target state generation image and the second mixture material surface state image are extracted respectively to obtain the second mixture material target state image semantic coding feature map and the second mixture material surface state image semantic coding feature map, including: inputting the second mixture material target state generation image and the second mixture material surface state image into a surface state feature extractor based on a deep separable convolutional neural network model to obtain the second mixture material target state image semantic coding feature map and the second mixture material surface state image semantic coding feature map.

[0012] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, the second mixture material target state image semantic coding feature map and the second mixture material surface state image semantic coding feature map are subjected to pixel-level semantic association spatial enhancement to obtain the second mixture material target state image semantic space enhancement coding feature map and the second mixture material surface state image semantic space enhancement coding feature map, including: calculating the pixel-level semantic association of the second mixture material target state image semantic coding feature map to obtain the second mixture material target state image pixel granularity semantic association topological matrix; based on the pixel spatial position information of the second mixture material target state image semantic coding feature map, performing spatial attention attenuation modulation on the second mixture material target state image pixel granularity semantic association topological matrix to obtain the second mixture material target state semantic association space soft constraint topological feature matrix; based on the second mixture material target state semantic association space soft constraint topological feature matrix, performing pixel granularity spatial enhancement processing on the second mixture material target state image semantic coding feature map to obtain the second mixture material target state image semantic space enhancement coding feature map.

[0013] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, the pixel-level semantic association of the semantic coding feature map of the target state image of the second mixture material is calculated to obtain the pixel granularity semantic association topological matrix of the target state image of the second mixture material, including: feature dispersion of the semantic coding feature map of the target state image of the second mixture material along the channel dimension to obtain a set of pixel granularity feature vectors of the target state image of the second mixture material; calculating the semantic association score between any two pixel granularity feature vectors of the target state image of the second mixture material in the set of pixel granularity feature vectors of the target state image of the second mixture material to obtain the pixel granularity semantic association topological matrix of the target state image of the second mixture material.

[0014] In the preparation method of the rare earth reflective heat-insulating coating, based on the pixel spatial position information of the second mixed material target state image semantic coding feature map, the second mixed material target state image pixel granularity semantic association topological matrix is ​​subjected to spatial attention attenuation modulation to obtain the second mixed material target state semantic association space soft constraint topological feature matrix, including: taking the natural exponential function value of the Euclidean distance between any two second mixed material target state image pixel granularity feature vectors in the set of the second mixed material target state image pixel granularity feature vectors as the spatial distance factor to obtain the second mixed material target state image pixel granularity space distance topological matrix; using a preset attenuation modulation constant to multiply the second mixed material target state image pixel granularity space distance topological matrix to obtain the spatial attention attenuation modulation matrix, and calculating the position point division of the second mixed material target state image pixel granularity semantic association topological matrix and the spatial attention attenuation modulation matrix to obtain the second mixed material target state semantic association space soft constraint topological matrix. The second mixed material target state semantic association space soft constraint topological matrix is ​​subjected to dilated convolution coding to obtain the second mixed material target state semantic association space soft constraint topological feature matrix.

[0015] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, based on the second mixture material target state semantic association space soft constraint topological feature matrix, the second mixture material target state image semantic coding feature map is subjected to pixel granularity space enhancement processing to obtain the second mixture material target state image semantic space enhanced coding feature map, including: inputting the set of pixel granularity feature vectors of the second mixture material target state image and the second mixture material target state semantic association space soft constraint topological feature matrix into the graph convolutional coding module to obtain the set of context semantic association enhanced second mixture material target state pixel granularity feature vectors; and performing feature shape reshaping on the set of context semantic association enhanced second mixture material target state pixel granularity feature vectors to obtain the second mixture material target state image semantic space enhanced coding feature map.

[0016] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, based on the feature difference between the semantic space enhancement coding feature map of the target state image of the second mixture and the semantic space enhancement coding feature map of the surface state image of the second mixture, a recommended value of the thickener application amount is generated, including: calculating the positional difference between the semantic space enhancement coding feature map of the target state image of the second mixture and the semantic space enhancement coding feature map of the surface state image of the second mixture to obtain a state difference image semantic coding feature map; inputting the state difference image semantic coding feature map into a decoder-based thickener application recommendation module to obtain the recommended value of the thickener application amount.

[0017] According to another aspect of the present application, a rare earth reflective thermal insulation coating is provided, wherein the rare earth reflective thermal insulation coating is prepared by the preparation method of the rare earth reflective thermal insulation coating as described above.

[0018] Compared with the prior art, the rare earth reflective heat-insulating coating and its preparation method described in the present application, in the process of adjusting the viscosity of the rare earth reflective heat-insulating coating, firstly generates a corresponding target state image of the mixed material according to the target viscosity value input by the user, and then uses the image processing technology based on deep learning 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, so as to extract the surface state features of the two materials, and then intelligently recommend the application amount of the thickener based on the feature difference between the actual surface state of the mixed material and the target state. In this way, the application amount of the thickener can be optimized and controlled, and the production efficiency and quality stability of the coating can be improved. 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 adding a thickener to adjust the viscosity of the second mixed material to obtain a third mixed material in the preparation method of the rare earth reflective thermal insulation coating according to an embodiment of the present application.

[0021] Figure 2 This is a flow chart of step S52 in the method for preparing the rare earth reflective thermal insulation coating according to an embodiment of the present application.

[0022] Figure 3 This is a flow chart of step S54 in the method for preparing the rare earth reflective thermal insulation coating according to an embodiment of the present application.

[0023] Figure 4 This is a flow chart of step S55 in the method for preparing the rare earth reflective thermal insulation coating 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] In the technical solution of the present application, in order to adjust the rare earth reflective heat-insulating coating from water-based to oil-based, water is replaced with toluene (that is, the solvent water is replaced with toluene). At the same time, the pure acrylic emulsion is replaced with pure acrylic resin, and other materials and processes are fine-tuned on the basis of the existing technology.

[0026] In a specific example, the formula composition of the rare earth reflective thermal insulation coating described in this application is shown in Table 1: Table 1 Formula composition of rare earth reflective thermal insulation coating

[0027] After testing, the performance of the rare earth reflective thermal insulation coating is shown in Table 2: Table 2 Performance indicators of rare earth reflective thermal insulation coatings

[0028] It can be seen from the data in Table 2 that the rare earth reflective thermal insulation coating described in the embodiment of the present application has excellent temperature resistance, the color remains stable, and it also performs well in performance indicators such as waterproofness, adhesion, aging resistance, and salt spray resistance. It has a short drying time and uniform thickness, meeting all requirements of high-performance thermal insulation coatings.

[0029] In particular, in the technical solution of the present application, the rare earth reflective thermal insulation coating is an oily reflective thermal insulation coating, which has the advantages of durability, acid and alkali resistance, and aging resistance. In specific applications, it can be sprayed or used for pre-roll coating. In particular, the rare earth reflective thermal insulation coating is a rare earth composite material with high reflectivity (reflectivity greater than 92%), a reflection range of 200nm--2500nm, and high radiation capacity (hemispherical emissivity greater than 86%), and radiation in the 2-15μm band. It is worth mentioning that it has excellent thermal insulation performance. Due to the oily paint and ultra-thin level (film thickness level is controlled within 200um), pre-roll coated metal plates can be produced with friendly processing characteristics such as stamping and bending.

[0030] It is also worth mentioning that the rare earth reflective heat-insulating coating achieves the purpose of lowering the temperature by efficiently reflecting sunlight and radiant heat energy and blocking heat insulation. Based on the reflection effect of rare earth compounds on infrared sunlight and the refraction and reflection effect of spherical microbeads, it can highly reflect the entire band of visible sunlight and infrared rays in the range of 200nm--2500nm. The rare earth reflective heat-insulating coating adopts a special microscopic convex and concave high-radiation surface, which increases the surface area several times, and removes the solar heat that is not reflected and removed from the surface with 2-15μm band radiation, further reducing the surface temperature and achieving effective heat dissipation and cooling control.

[0031] In practical applications, the appropriateness of the viscosity of rare earth reflective thermal insulation coatings is directly related to the coating's construction performance, drying speed, and final coating quality. For example, if the viscosity is too high, the coating is prone to sagging during spraying or brushing; if the viscosity is too low, the coating may not form a uniform coating, affecting the thermal insulation effect. Therefore, in the actual production process of rare earth reflective thermal insulation coatings, the viscosity of the coating needs to be precisely controlled.

[0032] However, the current regulation of coating viscosity mainly relies on the operator's manual experience and repeated tests. This method is not only inefficient, but also prone to errors, making it difficult to ensure the consistency and stability of coating quality. Especially in rare earth reflective thermal insulation coatings with complex formula ingredients and a wide variety of pigments and fillers, the precise control of viscosity has become a technical difficulty. In response to the above technical problems, the present application further introduces intelligent industrial control technology to adjust the amount of thickener applied. Specifically, in the viscosity adjustment process of rare earth reflective thermal insulation coatings, the corresponding target state image of the mixed material is first 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 image features and enhance features of the target state image of the mixed material and the actual surface state image of the mixed material, so as to extract the surface state characteristics of the two materials, and then based on the characteristic 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.

[0033] Figure 1 This is a flow chart of adding a thickener to adjust the viscosity of the second mixed material to obtain a third mixed material in the method for preparing the rare earth reflective heat-insulating coating according to an embodiment of the present application. Figure 1As shown, the step 5 also includes: S51, receiving the target viscosity value input by the user, and simultaneously acquiring the second mixed material surface state image collected by the industrial camera; S52, generating the second mixed material target state generation image based on the target viscosity value; S53, respectively extracting the material surface state features of the second mixed material target state generation image and the second mixed material surface state image to obtain the second mixed material target state image semantic coding feature map and the second mixed material surface state image semantic coding feature map; S54, performing pixel-level semantic association space enhancement on the second mixed material target state image semantic coding feature map and the second mixed material surface state image semantic coding feature map to obtain the second mixed material target state image semantic space enhanced coding feature map and the second mixed material surface state image semantic space enhanced coding feature map; S55, generating a recommended value for the amount of thickener to be applied based on the feature difference between the second mixed material target state image semantic space enhanced coding feature map and the second mixed material surface state image semantic space enhanced coding feature map.

[0034] In the above-mentioned method for preparing rare earth reflective thermal insulation coating, the step S51 receives the target viscosity value input by the user, and simultaneously obtains the surface state image of the second mixed material captured by the industrial camera. It should be understood that by collecting the surface state image of the mixed material through the industrial camera, the current material viscosity information is provided, so as to facilitate the recommendation of the amount of thickener to be applied based on the current material state. At the same time, different application scenarios may require coatings of different viscosities. For example, in some coating processes, a higher viscosity helps prevent sagging; in other cases, a lower viscosity can improve spraying efficiency. Therefore, by receiving the target viscosity value input by the user, various specific application requirements can be better adapted.

[0035] In the method for preparing the rare earth reflective heat-insulating coating, the step S52 generates a second mixed material target state generation image based on the target viscosity value. Figure 2 FIG. 5 is a flow chart of step S52 in the method for preparing the rare earth reflective heat-insulating coating according to an embodiment of the present application. Figure 2 As shown, the step S52 includes: S521, performing one-hot encoding on the target viscosity value to obtain a target viscosity one-hot encoding vector; S522, inputting the target viscosity one-hot encoding vector into a second mixed material target state image generator based on a diffusion model to obtain the second mixed material target state generation image.

[0036] Specifically, the step S521 performs one-hot encoding on the target viscosity value to obtain a one-hot encoding vector of the target viscosity. It should be understood that, considering that the target viscosity value is a discrete value, in the machine learning model, directly processing its raw data may be affected by the size of the value itself, resulting in difficulty in model training. Therefore, the present application further adopts a one-hot encoding method to convert the target viscosity value into a binary vector of fixed length, in which only one element is 1 and the rest of the elements are 0, and each binary vector corresponds to a specific viscosity value. In this way, the numerical value is converted into category information, so that the machine learning model can better understand and utilize the target viscosity information, thereby improving the generalization ability of the model.

[0037] Specifically, in step S522, the target viscosity one-hot encoding vector is input into the second mixed material target state image generator based on the diffusion model to obtain the second mixed material target state generated image. That is, in order to provide a clear reference standard for the recommended control of the thickener application amount, the present application uses the image generation technology based on the diffusion model to process the target viscosity one-hot encoding vector, so as to utilize the powerful image generation ability of the diffusion model to gradually iterate and generate the mixed material target state image corresponding to the target viscosity value, so that by comparing the difference between the target state image and the actual mixed material surface state image, the required thickener application amount can be more intuitively evaluated, thereby improving the accuracy of the thickener application amount recommendation.

[0038] In the above-mentioned method for preparing the rare earth reflective thermal insulation coating, the step S53 extracts the material surface state features of the second mixture target state generated image and the second mixture surface state image respectively to obtain the second mixture target state image semantic coding feature map and the second mixture surface state image semantic coding feature map. It should be understood that in order to realize the comparative analysis of the feature differences between the second mixture target state generated image and the second mixture surface state image, it is necessary to further use image processing technology to extract the material state feature representations in the two. In a specific example of the present application, the step S53 includes: inputting the second mixture target state generated image and the second mixture surface state image into a surface state feature extractor based on a deep separable convolutional neural network model to obtain the second mixture target state image semantic coding feature map and the second mixture surface state image semantic coding feature map. Among them, the deep separable convolutional neural network significantly reduces the amount of calculation and the number of model parameters by decomposing the standard convolution layer into depthwise convolution and pointwise convolution, while maintaining a high feature extraction capability. It can effectively extract key surface state features from the image, including but not limited to texture, color, glossiness, and surface roughness, providing an accurate basis for subsequent recommendations on the amount of thickener to be applied.

[0039] In the preparation method of the above-mentioned rare earth reflective thermal insulation coating, the step S54 performs pixel-level semantic association spatial enhancement on the semantic coding feature map of the target state image of the second mixture and the semantic coding feature map of the surface state image of the second mixture to obtain the semantic spatial enhancement coding feature map of the target state image of the second mixture and the semantic spatial enhancement coding feature map of the surface state image of the second mixture. It should be understood that, considering that the above-mentioned convolution operation can usually only capture local image features, the understanding of the global context information of the image is limited. Therefore, in order to further enhance the global expression ability of image features, the present application proposes a feature enhancement method based on pixel granularity, which realizes an in-depth understanding of the global context information of the image by mining the pixel-level feature semantic associations in the image, so that the model can better identify and understand the surface state of the mixture.

[0040] Figure 3 FIG. 5 is a flow chart of step S54 in the method for preparing the rare earth reflective heat-insulating coating according to an embodiment of the present application. Figure 3As shown, the step S54 includes: S541, calculating the pixel-level semantic association of the second mixed material target state image semantic coding feature map to obtain the second mixed material target state image pixel granularity semantic association topological matrix; S542, based on the pixel spatial position information of the second mixed material target state image semantic coding feature map, performing spatial attention attenuation modulation on the second mixed material target state image pixel granularity semantic association topological matrix to obtain the second mixed material target state semantic association space soft constraint topological feature matrix; S543, based on the second mixed material target state semantic association space soft constraint topological feature matrix, performing pixel granularity spatial enhancement processing on the second mixed material target state image semantic coding feature map to obtain the second mixed material target state image semantic space enhanced coding feature map.

[0041] Specifically, the step S541 includes: performing feature discretization on the semantic encoding feature map of the second mixture material target state image along the channel dimension to obtain a set of pixel granularity feature vectors of the second mixture material target state image; calculating the semantic association score between any two second mixture material target state image pixel granularity feature vectors in the set of the second mixture material target state image pixel granularity feature vectors to obtain the second mixture material target state image pixel granularity semantic association topological matrix.

[0042] In a specific example of the present application, the semantic association score between any two second mixture material target state image pixel granularity feature vectors is calculated, including: performing nonlinear mapping on each second mixture material target state image pixel granularity feature vector in the set of the second mixture material target state image pixel granularity feature vectors to obtain a set of mapped second mixture material target state image pixel granularity feature vectors; calculating the Poincare distance between any two mapped second mixture material target state image pixel granularity feature vectors in the set of mapped second mixture material target state image pixel granularity feature vectors as the semantic association score to obtain the second mixture material target state image pixel granularity semantic association topological matrix.

[0043] Here, taking the semantic coding feature map of the target state image of the second mixed material as an example, the features are first dispersed along the channel dimension to split it into a set of feature vectors of pixel granularity. Then, by calculating the semantic association between any two pixel granularity feature vectors in the set, the pixel granularity semantic association topological matrix of the target state image of the second mixed material is constructed, thereby revealing the intrinsic semantic association of the image. In particular, in this process, considering that the hyperbolic space can more effectively capture the nonlinear relationship between data compared with the Euclidean space, the present application first maps each pixel granularity feature vector from the Euclidean space to the Poincare space, so that it adapts to the geometric characteristics of the hyperbolic space while maintaining the original semantic information, and calculates the Poincare distance between any two mapped pixel granularity feature vectors as the semantic association score between the two, so as to utilize the geometric characteristics of the Poincare space to enhance the nonlinear association between features and improve the model's ability to distinguish features of different categories, thereby more accurately simulating the intrinsic semantic association structure of the image.

[0044] That is, in the above specific example, the implementation process of step S541 can be expressed as follows:

[0045]

[0046]

[0047] in, is the semantic encoding feature map of the target state image of the second mixed material, , and represent the height, width and number of channels of the semantic encoding feature map of the target state image of the second mixed material, respectively, represents the feature shape reshaping, represents the set of pixel granularity feature vectors of the target state image of the second mixed material, and respectively represent the first and The pixel size feature vector of the second mixed material target state image, The value of is the number of feature vectors in the set of pixel particle size feature vectors of the target state image of the second mixed material, and , The value of is the length of the pixel granularity feature vector of the target state image of the second mixed material, is the weight transformation matrix, is the weight transformation vector, and respectively represent the first and The pixel granularity feature vector of the second mixed material target state image after mapping, represents the inverse hyperbolic cosine function in hyperbolic space, represents the norm of the eigenvector, Indicates the the pixel granularity feature vector of the target state image of the second mixed material after mapping and the The Poincare distance between the pixel granularity feature vectors of the second mixed material target state image after mapping, The pixel-granular semantic association topological matrix representing the target state image of the second mixture is The semantic relevance score value of the location.

[0048] Specifically, the step S542 includes: using the natural exponential function value of the Euclidean distance between any two second mixed material target state image pixel granularity feature vectors in the set of the second mixed material target state image pixel granularity feature vectors as the spatial distance factor to obtain the second mixed material target state image pixel granularity spatial distance topological matrix; using a preset attenuation modulation constant to multiply the second mixed material target state image pixel granularity spatial distance topological matrix to obtain a spatial attention attenuation modulation matrix, and calculating the second mixed material target state image pixel granularity semantic association topological matrix and the spatial attention attenuation modulation matrix divided by the position point to obtain the second mixed material target state semantic association space soft constraint topological matrix; and, performing hole convolution coding on the second mixed material target state semantic association space soft constraint topological matrix to obtain the second mixed material target state semantic association space soft constraint topological feature matrix.

[0049] In this specific example, the above process can be expressed as:

[0050]

[0051] in, The pixel-granular semantic association topological matrix representing the target state image of the second mixture is The semantic relevance score of the location, is the preset attenuation modulation constant, Indicates the the pixel granularity feature vector of the second mixed material target state image and the The Euclidean distance between the pixel size feature vectors of the second mixed material target state image, is the soft constraint topological matrix of the second mixture target state semantic association space, The second mixture target state semantic association space soft constraint topology matrix is The eigenvalues ​​of the positions, represents the dilated convolutional coding, It is the soft constraint topological feature matrix of the second mixture material target state semantic association space.

[0052] That is, considering that in practical applications, adjacent image regions often have stronger correlations than distant regions, the present application further performs spatial attention attenuation modulation on the pixel granularity semantic association topological matrix of the second mixed material target state image based on the spatial distance between any two pixel granularity feature vectors to simulate the association attenuation characteristics between local image regions, so that the model pays more attention to the feature association of adjacent regions. Furthermore, the topological matrix adjusted by spatial soft constraints is subjected to atrous convolution encoding using the atrous convolution technique to capture the semantic association topological features between each pixel granularity feature, and obtain the second mixed material target state semantic association spatial soft constraint topological feature matrix.

[0053] Specifically, the step S543 includes: inputting the set of pixel granularity feature vectors of the second mixed material target state image and the second mixed material target state semantic association space soft constraint topological feature matrix into a graph convolutional coding module to obtain a set of contextual semantic association enhanced second mixed material target state pixel granularity feature vectors; and performing feature reshaping on the set of contextual semantic association enhanced second mixed material target state pixel granularity feature vectors to obtain the second mixed material target state image semantic space enhanced coding feature map.

[0054] That is, the set of pixel granularity feature vectors of the second mixed material target state image and the soft constraint topological feature matrix of the second mixed material target state semantic association space are further processed by graph convolution coding technology, and the image's individual pixel granularity feature representations are updated based on the semantic association information between pixels by utilizing the graph convolution characteristics, and the global semantic context association feature expression of the image pixel granularity features is enhanced by learning the intrinsic semantic association structure of the image. Finally, the feature shape of each pixel granularity feature after the context semantic association is strengthened is reshaped to restore its original feature structure, so as to obtain the second mixed material target state image semantic space enhanced coding feature map, thereby realizing the refined enhancement of the local detail features of the image and the optimized expression of the global semantic information.

[0055] In this specific example, the execution process of this step can be expressed as:

[0056]

[0057] in, is the soft constraint topological feature matrix of the second mixed material target state semantic association space, represents graph convolution processing, represents the set of pixel granularity feature vectors of the second mixed material target state enhanced by contextual semantic association, Represents a semantic space enhanced encoding feature map of the target state image of the second mixed material.

[0058] In the method for preparing the rare earth reflective heat-insulating coating, the step S55 generates a recommended value for the amount of thickener to be applied based on the feature difference between the semantic space enhanced coding feature map of the target state image of the second mixture and the semantic space enhanced coding feature map of the surface state image of the second mixture. Figure 4 FIG. 5 is a flow chart of step S55 in the method for preparing the rare earth reflective heat-insulating coating according to an embodiment of the present application. Figure 4 As shown, the step S55 includes: S551, calculating the positional difference between the semantic space enhancement coding feature map of the target state image of the second mixture and the semantic space enhancement coding feature map of the surface state image of the second mixture to obtain a state difference image semantic coding feature map; S552, inputting the state difference image semantic coding feature map into a decoder-based thickener application recommendation module to obtain a recommended value of the thickener application amount.

[0059] Specifically, the step S551 calculates the positional difference between the semantic space enhanced coding feature map of the target state image of the second mixture and the semantic space enhanced coding feature map of the surface state image of the second mixture to obtain the state difference image semantic coding feature map. That is, in order to reveal the characteristic difference between the target state image of the mixture and the actual surface state image, the present application further calculates the positional difference between the semantic space enhanced coding feature map of the target state image of the second mixture and the semantic space enhanced coding feature map of the surface state image of the second mixture to quantify the specific difference between the target state image and the actual surface state image, thereby providing a direct characteristic basis for the precise application of the thickener.

[0060] Specifically, in step S552, the state difference image semantic coding feature map is input into the thickener application recommendation module based on the decoder to obtain the recommended value of the thickener application amount. In the technical solution of the present application, the decoder is based on a multi-layer neural network architecture, and by performing a series of nonlinear transformations on the state difference image semantic coding feature map, it can fully understand the degree of difference between the mixed material and the target state, thereby accurately predicting the required thickener application amount and providing intuitive guidance for the operator.

[0061] In particular, the state difference image semantic coding feature map is used to represent the image semantic difference coding features between the image semantic coding features of the surface state image of the second mixture and the image semantic coding features of the second mixture target state generated image generated by the target viscosity one-hot coding vector through the diffusion model. However, considering that the image feature is not a minimum feature unit per position, in the process of calculating the positional difference between the second mixture target state image semantic space enhancement coding feature map and the second mixture surface state image semantic space enhancement coding feature map, the state difference image semantic coding feature map will have fine-grained image semantic difference coding diversity, which affects its decoding accuracy when it is input into the decoder-based thickener application recommendation module for decoding.

[0062] Based on this, in the technical solution of the present application, before the state difference image semantic coding feature map is input into the thickener application recommendation module based on the decoder, the state difference image semantic coding feature map is optimized for feature distribution, and the feature distribution optimization process includes: Determine the feature mean and feature variance of the state difference image semantic coding feature map, and divide the feature mean by the feature variance to obtain the state difference image semantic coding probability statistics value:

[0063] in, and represent the feature mean and feature variance of the semantic encoding feature map of the state difference image respectively, Represents the probability statistics of the semantic encoding of the state difference image; The state difference image semantic coding feature map is point-multiplied with the inverse of the maximum eigenvalue in the state difference image semantic coding feature map to obtain the state difference image semantic coding probability constraint map:

[0064] in, represents the semantic encoding feature map of the state difference image, represents the maximum eigenvalue in the semantic encoding feature map of the state difference image, It means point multiplication by position. Represents the probability constraint graph of semantic encoding of state difference image; After performing point addition on the state difference image semantic coding probability constraint graph and the state difference image semantic coding probability statistical value, a logarithmic value with base 2 is calculated to obtain the state difference image semantic coding information interaction graph:

[0065] in, represents the probability constraint graph of semantic encoding of state difference image, represents the probability statistics of the semantic encoding of the state difference image, Indicates adding by position point. Represents the interactive graph of semantically encoded information of state difference images; After performing point subtraction on the state difference image semantic coding probability constraint map with the state difference image semantic coding probability statistics value, the inverse of each eigenvalue is calculated to obtain the state difference image semantic coding sequence constraint map:

[0066] in, represents the probability constraint graph of semantic encoding of state difference image, represents the probability statistics of the semantic encoding of the state difference image, It means to subtract by position point. Represents the state difference image semantic encoding sequence constraint graph; The state difference image semantic coding information interaction graph and the state difference image semantic coding sequence constraint graph are interpolated to obtain an optimized state difference image semantic coding feature graph.

[0067] Furthermore, the optimized state difference image semantic coding feature map is passed through a decoder-based thickener application recommendation module to obtain a recommended value of the thickener application amount.

[0068] Accordingly, in this embodiment, the probabilistic statistical characteristics of the state difference image semantic coding feature map are used to simulate the mesoscale interaction structure under probability constraints between the eigenvalue scale and the feature map scale of the state difference image semantic coding feature map, so as to construct a bidirectional latent variable motif based on the mesoscale short sequence relative to the probability statistical value to perform mesoscale bidirectional migration, and to perform posterior recovery based on the sequence constraints on the interactive information, thereby improving the convergence effect in the probability density domain and improving the accuracy of the recommended value of the thickener application amount obtained by the decoder-based thickener application recommendation module of the state difference image semantic coding feature map.

[0069] In summary, the rare earth reflective heat-insulating coating and its preparation method according to the embodiment of the present application are explained. In the viscosity adjustment process of the rare earth reflective heat-insulating coating, the corresponding target state image of the mixed material is first 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 image features and enhance features of the target state image of the mixed material and the actual surface state image of the mixed material, so as 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 to be applied is intelligently recommended. In this way, the amount of thickener applied can be optimized and controlled, and the production efficiency and quality stability of the coating can be improved.

[0070] 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 coating, comprising: Step 1: Add solvent, dispersant and defoamer to a dispersion tank, and disperse at a low speed until uniform, wherein the solvent is toluene; 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 resin, 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 bactericide to the third mixed material and stir to obtain a rare earth reflective thermal insulation coating, characterized in that the step 5 comprises: receiving a target viscosity value input by a user, and acquiring a surface state image of the second mixed material captured by an industrial camera; generating a second mixed material target state generation image based on the target viscosity value; Respectively extracting material surface state features of the second mixed material target state generation image and the second mixed material surface state image to obtain a second mixed material target state image semantic coding feature map and a second mixed material surface state image semantic coding feature map; Performing pixel-level semantic association spatial enhancement on the second mixed material target state image semantic coding feature map and the second mixed material surface state image semantic coding feature map to obtain the second mixed material target state image semantic space enhancement coding feature map and the second mixed material surface state image semantic space enhancement coding feature map; Based on the feature difference between the semantic space enhanced coding feature map of the target state image of the second mixture and the semantic space enhanced coding feature map of the surface state image of the second mixture, a recommended value for the amount of thickener to be applied is generated.

2. The method for preparing the rare earth reflective heat-insulating coating according to claim 1, characterized in that: Generating a second mixed material target state generation image based on the target viscosity value includes: One-hot encoding the target viscosity value to obtain a target viscosity one-hot encoding vector; The target viscosity one-hot encoding vector is input into a second mixed material target state image generator based on a diffusion model to obtain the second mixed material target state generation image.

3. The method for preparing the rare earth reflective heat-insulating coating according to claim 2, characterized in that: Extracting the material surface state features of the second mixed material target state generation image and the second mixed material surface state image respectively to obtain the second mixed material target state image semantic coding feature map and the second mixed material surface state image semantic coding feature map, including: The second mixture material target state generation image and the second mixture material surface state image are input into a surface state feature extractor based on a deep separable convolutional neural network model to obtain a semantic coding feature map of the second mixture material target state image and a semantic coding feature map of the second mixture material surface state image.

4. The method for preparing the rare earth reflective heat-insulating coating according to claim 3, characterized in that: Performing pixel-level semantic association spatial enhancement on the second mixed material target state image semantic coding feature map and the second mixed material surface state image semantic coding feature map to obtain the second mixed material target state image semantic spatial enhancement coding feature map and the second mixed material surface state image semantic spatial enhancement coding feature map, including: Calculating the pixel-level semantic association of the semantic encoding feature map of the second mixed material target state image to obtain a pixel-level semantic association topological matrix of the second mixed material target state image; Based on the pixel spatial position information of the semantic encoding feature map of the target state image of the second mixed material, spatial attention attenuation modulation is performed on the pixel granularity semantic association topological matrix of the target state image of the second mixed material to obtain a second mixed material target state semantic association spatial soft constraint topological feature matrix; Based on the soft constraint topological feature matrix of the second mixture material target state semantic association space, the second mixture material target state image semantic coding feature map is subjected to pixel granularity space enhancement processing to obtain the second mixture material target state image semantic space enhanced coding feature map.

5. The method for preparing the rare earth reflective heat-insulating coating according to claim 4, characterized in that: Calculating the pixel-level semantic association of the semantic encoding feature map of the second mixed material target state image to obtain a pixel-level semantic association topological matrix of the second mixed material target state image includes: Performing feature dispersion on the semantic encoding feature map of the second mixed material target state image along the channel dimension to obtain a set of pixel granularity feature vectors of the second mixed material target state image; The semantic association score between any two second mixed material target state image pixel granularity feature vectors in the set of the second mixed material target state image pixel granularity feature vectors is calculated to obtain the second mixed material target state image pixel granularity semantic association topological matrix.

6. The method for preparing the rare earth reflective heat-insulating coating according to claim 5, characterized in that: Based on the pixel spatial position information of the semantic coding feature map of the target state image of the second mixed material, a spatial attention attenuation modulation is performed on the pixel granularity semantic association topological matrix of the target state image of the second mixed material to obtain a second mixed material target state semantic association space soft constraint topological feature matrix, including: Using the natural exponential function value of the Euclidean distance between any two second mixed material target state image pixel particle size feature vectors in the set of the second mixed material target state image pixel particle size feature vectors as a spatial distance factor to obtain a second mixed material target state image pixel particle size spatial distance topological matrix; Multiplying the second mixed material target state image pixel granularity space distance topological matrix by a preset attenuation modulation constant to obtain a spatial attention attenuation modulation matrix, and calculating the position point-by-position division of the second mixed material target state image pixel granularity semantic association topological matrix and the spatial attention attenuation modulation matrix to obtain a second mixed material target state semantic association space soft constraint topological matrix; The second mixed material target state semantic association space soft constraint topology matrix is ​​subjected to dilated convolution coding to obtain the second mixed material target state semantic association space soft constraint topology feature matrix.

7. The method for preparing the rare earth reflective heat-insulating coating according to claim 6, characterized in that: Based on the second mixed material target state semantic association space soft constraint topological feature matrix, the second mixed material target state image semantic coding feature map is subjected to pixel granularity space enhancement processing to obtain the second mixed material target state image semantic space enhanced coding feature map, including: Inputting the set of pixel granularity feature vectors of the second mixed material target state image and the second mixed material target state semantic association space soft constraint topological feature matrix into a graph convolutional coding module to obtain a set of contextual semantic association enhanced second mixed material target state pixel granularity feature vectors; The set of pixel granularity feature vectors of the second mixed material target state enhanced by the contextual semantic association is reshaped to obtain the second mixed material target state image semantic space enhanced coding feature map.

8. The method for preparing the rare earth reflective heat-insulating coating according to claim 7, characterized in that: Generating a recommended value of the amount of thickener to be applied based on a feature difference between the semantic space enhanced coding feature map of the target state image of the second mixed material and the semantic space enhanced coding feature map of the surface state image of the second mixed material, including: Calculating the position difference between the semantic space enhanced coding feature map of the target state image of the second mixed material and the semantic space enhanced coding feature map of the surface state image of the second mixed material to obtain a state difference image semantic coding feature map; The state difference image semantic coding feature map is input into a thickener application recommendation module based on a decoder to obtain a recommended value of the thickener application amount.

9. A rare earth reflective heat-insulating coating, characterized in that: The rare earth reflective thermal insulation coating is prepared by the preparation method of the rare earth reflective thermal insulation coating as described in any one of claims 1-8.

Images