A method for coating multi-layer profile analysis based on data analysis

Abstract

The present invention discloses a method for analyzing the multi-layer profile of a coating based on data analysis, belonging to the technical field of data processing. An image acquisition device is used to acquire an image of the profile to determine the multi-layer profile image of the sample coating. Then, the multi-layer feature information corresponding to the multi-layer profile image of the sample coating is extracted. A machine learning algorithm is used to establish a data relationship between the multi-layer feature information of the coating and the performance indexes of the coating sample, and a coating performance analysis model is obtained. Finally, the target multi-layer profile image corresponding to the coating to be analyzed can be acquired, and the target multi-layer feature information corresponding to the target multi-layer profile image is extracted. The coating performance analysis model is used to analyze the target multi-layer feature information of the coating to determine the analysis result of the multi-layer profile of the coating to be analyzed, which can effectively improve the accuracy and reliability of coating analysis, is applicable to the profile analysis of various coating materials, and is of great significance for improving the quality of coating products and optimizing the production process.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to a method for analyzing the multi-layer profile of a coating based on data analysis. Background Art

[0002] A coating is a thin covering material applied to the surface of an object for protection, decoration, or to endow specific functions such as corrosion resistance, wear resistance, insulation, conductivity, etc. It can be paint, plastic, metal, or other materials, and is widely used in industry, construction, and daily life by improving the appearance and extending the service life of the substrate. With the development of technology, coating materials have been widely used in fields such as aerospace, automotive, and electronics. The analysis of the multi-layer profile of a coating is of great significance for understanding the coating structure, performance, and optimizing the production process. However, traditional profile analysis methods have problems such as low resolution and poor reliability, and it is difficult to meet the high-precision analysis requirements of modern coating materials. Summary of the Invention

[0003] The present invention provides a method for analyzing the multi-layer profile of a coating based on data analysis to solve the technical problems existing in the prior art.

[0004] A method for analyzing the multi-layer profile of a coating based on data analysis includes:

[0005] Cut the coating sample, ensuring that the cutting direction is perpendicular to the coating surface, and use an image acquisition device to acquire an image of the profile to determine the multi-layer profile image of the sample coating;

[0006] Preprocess the multi-layer profile image of the sample coating to obtain the preprocessed multi-layer profile image of the sample coating, and extract the multi-layer feature information of the coating corresponding to the preprocessed multi-layer profile image of the sample coating;

[0007] Provide both the preprocessed multi-layer profile image of the sample coating and the multi-layer feature information to the staff to obtain the performance indicators of the coating sample input by the staff through human-computer interaction;

[0008] Use a machine learning algorithm to establish a data relationship between the multi-layer feature information of the coating and the performance indicators of the coating sample to obtain a coating performance analysis model;

[0009] Collect the target multi-layer profile image of the coating to be analyzed, extract the target multi-layer feature information corresponding to the target multi-layer profile image of the coating, and use the coating performance analysis model to analyze the target multi-layer feature information to determine the multi-layer profile analysis result of the coating to be analyzed.

[0010] Further, after determining the coating multi-layer profile analysis result corresponding to the coating to be analyzed, it further includes: generating a coating multi-layer profile analysis report according to the target coating multi-layer profile image, the target coating multi-layer feature information, and the coating multi-layer profile analysis result according to a preset template; wherein, the format of the coating multi-layer profile analysis report is PDF format or Word format.

[0011] Further, an image acquisition device is used to acquire an image of the profile to determine the sample coating multi-layer profile image, including: scanning the coating profile with a scanning electron microscope or an atomic force microscope to obtain the sample coating multi-layer profile image.

[0012] Further, preprocessing the sample coating multi-layer profile image to obtain the preprocessed sample coating multi-layer profile image, including: denoising, enhancing, and segmenting the sample coating multi-layer profile image to obtain the preprocessed sample coating multi-layer profile image.

[0013] Further, extracting the coating multi-layer feature information corresponding to the preprocessed sample coating multi-layer profile image, including:

[0014] Based on the preprocessed sample coating multi-layer profile image, obtaining the thickness and boundary morphology data of each layer in the sample coating multi-layer profile image to obtain the physical morphology characteristics corresponding to the sample coating multi-layer profile image;

[0015] Obtaining the composition type code of each layer in the sample coating multi-layer profile image input by the staff to obtain the chemical morphology characteristics corresponding to the sample coating multi-layer profile image;

[0016] Taking the physical morphology characteristics and the chemical morphology characteristics together as the coating multi-layer feature information corresponding to the sample coating multi-layer profile image.

[0017] Further, a machine learning algorithm is used to establish a data relationship between the coating multi-layer feature information and the coating sample performance index to obtain a coating performance analysis model, including:

[0018] Using a machine learning algorithm to construct an initial learning model, and using a chaotic mapping initialization method to initialize the model parameters of the initial learning model to obtain multiple different model parameter vectors;

[0019] For any model parameter vector, taking the coating multi-layer feature information as the input of the initial learning model and taking the coating sample performance index corresponding to the coating multi-layer feature information as the expected output to obtain the loss function value corresponding to the model parameter vector;

[0020] According to the loss function values corresponding to all model parameter vectors, obtaining the optimal model parameter vector;

[0021] For any model parameter vector, according to the optimal model parameter vector, adopt an adaptive following strategy to perform local fast search on the model parameter vector to obtain the model parameter vector after local fast search;

[0022] For the model parameter vector after local fast search, adopt an oscillation interval search strategy to perform local oscillation search on the model parameter vector to obtain the model parameter vector after local oscillation search;

[0023] For the model parameter vector after local oscillation search, adopt a double global mutation strategy to perform global search on the model parameter vector to obtain the model parameter vector after global search;

[0024] Judge whether the number of training times has reached the maximum number of training times. If so, according to the model parameter vector after global search, re-determine the optimal model parameter vector, and use the re-determined optimal model parameter vector as the final parameter of the initial learning model to obtain the coating performance analysis model. Otherwise, return to the step of obtaining the loss function value corresponding to the model parameter vector.

[0025] Furthermore, use the chaos mapping initialization method to initialize the model parameters of the initial learning model to obtain multiple different model parameter vectors, including:

[0026] For the model parameters of the initial learning model, perform random initialization between the upper and lower limits of the model parameters, and encode the initialized model parameters into a vector to determine the model parameter vector after random initialization;

[0027] Based on the model parameter vector after the random initialization, perform chaos mapping to obtain multiple different model parameter vectors.

[0028] Furthermore, for any model parameter vector, according to the optimal model parameter vector, adopt an adaptive following strategy to perform local fast search on the model parameter vector to obtain the model parameter vector after local fast search, including:

[0029] Based on the current number of training times, use an exponential function to obtain an adaptive balance search factor;

[0030] According to the adaptive balance search factor and the optimal model parameter vector, perform local fast search on the model parameter vector to obtain the model parameter vector after local fast search.

[0031] Furthermore, for the model parameter vector after local fast search, adopt an oscillation interval search strategy to perform local oscillation search on the model parameter vector to obtain the model parameter vector after local oscillation search, including:

[0032] For the model parameter vector after local fast search, an oscillation search factor is obtained, and an adaptive oscillation search weight is obtained according to the oscillation search factor and the current number of training times;

[0033] For the model parameter vector after local fast search, another model parameter vector is randomly matched for the model parameter vector to obtain a local cooperation vector corresponding to each model parameter vector;

[0034] According to the adaptive oscillation search weight and the local cooperation vector, local oscillation search is performed on the model parameter vector to obtain the model parameter vector after local oscillation search.

[0035] Further, for the model parameter vector after local oscillation search, a global search is performed on the model parameter vector using a double global mutation strategy to obtain the model parameter vector after global search, including:

[0036] Generate a first global transformation factor and a second global transformation factor;

[0037] For the model parameter vector after local oscillation search, according to the first global transformation factor and the second global transformation factor, a global search is performed on the model parameter vector using a spiral search path to obtain the model parameter vector after global search.

[0038] A method for analyzing the multi-layer profile of a coating based on data analysis provided by the present invention uses an image acquisition device to acquire an image of the profile, determines a sample multi-layer profile image of the coating, then extracts the coating multi-layer feature information corresponding to the preprocessed sample multi-layer profile image of the coating, uses a machine learning algorithm to establish a data relationship between the coating multi-layer feature information and the performance index of the coating sample, obtains a coating performance analysis model, and finally can acquire a target multi-layer profile image of the coating to be analyzed and extract the target coating multi-layer feature information corresponding to the target multi-layer profile image of the coating, and uses the coating performance analysis model to analyze the target coating multi-layer feature information to determine the multi-layer profile analysis result of the coating to be analyzed, which can effectively improve the accuracy and reliability of coating analysis, is applicable to the profile analysis of various coating materials, and is of great significance for improving the quality of coating products and optimizing the production process. Brief Description of the Drawings

[0039] The drawings here are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present invention, and are used together with the specification to explain the principles of the present invention.

[0040] Figure 1 It is a flowchart of a method for analyzing the multi-layer profile of a coating based on data analysis provided by an embodiment of the present invention.

[0041] Through the above-mentioned accompanying drawings, specific embodiments of the present invention have been shown, and will be described in more detail hereinafter. These drawings and the textual description are not intended to limit the scope of the inventive concept in any way, but to illustrate the concept of the present invention to those skilled in the art by reference to specific embodiments. Detailed Description of the Embodiments

[0042] Here, the exemplary embodiments will be described in detail, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numerals in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0043] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0044] As Figure 1 shown, an embodiment of the present invention provides a method for analyzing the multi-layer profile of a coating based on data analysis, including:

[0045] S1. Cut the coating sample, and ensure that the cutting direction is perpendicular to the coating surface, and use an image acquisition device to acquire an image of the profile to determine the multi-layer profile image of the sample coating;

[0046] For the coating sample, the cutting direction perpendicular to the coating surface can more effectively collect coating information, and a high-precision image acquisition device can be used to acquire the multi-layer profile image of the sample coating, so as to better identify the multi-layer feature information of the coating.

[0047] S2. Preprocess the multi-layer profile image of the sample coating, obtain the multi-layer profile image of the sample coating after preprocessing, and extract the multi-layer feature information of the coating corresponding to the multi-layer profile image of the sample coating after preprocessing;

[0048] Preprocessing the multi-layer profile image of the sample coating can be some processing operations that make the multi-layer profile image of the sample coating easier to identify, and then some characteristic data related to the coating can be extracted from the multi-layer profile image of the sample coating after preprocessing to obtain the multi-layer feature information of the coating corresponding to the multi-layer profile image of the sample coating.

[0049] S3. Provide both the multi-layer profile image of the sample coating after preprocessing and the multi-layer feature information to the staff to obtain the performance indexes of the coating sample input by the staff through human-computer interaction;

[0050] Both the multi-layer cross-sectional image of the sample coating after preprocessing and the multi-layer feature information of the coating are provided to the staff, which enables the staff to customize the performance indicators of the coating samples corresponding to each coating, thereby assisting the staff in quickly evaluating the coating performance. Only when collecting samples, professional personnel need to make marks to achieve the identification of customized coating performance, which can effectively reduce the requirements for the staff and the labor intensity of the staff in the subsequent process.

[0051] The performance indicators of the coating samples can be scratch resistance, corrosion resistance, bending resistance, and / or other properties, which can be set by the staff according to actual needs to achieve rapid analysis.

[0052] S4. Establish a data relationship between the multi-layer feature information of the coating and the performance indicators of the coating samples using a machine learning algorithm to obtain a coating performance analysis model.

[0053] For example, a convolutional neural network can be used to learn and establish a data relationship between the multi-layer feature information of the coating and the performance indicators of the coating samples, thereby obtaining a coating performance analysis model.

[0054] S5. Collect the target multi-layer cross-sectional image of the coating to be analyzed, extract the target multi-layer feature information corresponding to the target multi-layer cross-sectional image of the coating, and use the coating performance analysis model to analyze the target multi-layer feature information of the coating to determine the multi-layer cross-sectional analysis result of the coating to be analyzed (i.e., the performance indicators of the coating samples corresponding to the coating to be analyzed).

[0055] In the embodiments of the present invention, the data marked by the staff is identified to form a coating performance analysis model that can automatically identify data, ultimately effectively improving the efficiency and accuracy of the staff's multi-level analysis of the coating and effectively avoiding errors caused by manual analysis.

[0056] In the embodiments of the present invention, after determining the multi-layer cross-sectional analysis result of the coating to be analyzed, it further includes: generating a multi-layer cross-sectional analysis report of the coating according to the target multi-layer cross-sectional image, the target multi-layer feature information, and the multi-layer cross-sectional analysis result according to a preset template; wherein, the format of the multi-layer cross-sectional analysis report of the coating is in PDF format or Word format.

[0057] It should be noted that the format of the multi-layer cross-sectional analysis report of the coating can also be set to other formats to be applicable to different information systems.

[0058] In the embodiments of the present invention, an image acquisition device is used to acquire an image of the cross-section to determine the multi-layer cross-sectional image of the sample coating, including: scanning the coating cross-section using a scanning electron microscope or an atomic force microscope to obtain the multi-layer cross-sectional image of the sample coating.

[0059] Since the coating information is difficult to be observed by an ordinary camera, in the embodiments of the present invention, a scanning electron microscope or an atomic force microscope is used to scan the coating profile to obtain a relatively clear multi-layer profile image of the sample coating. However, it should be noted that other devices can also be used to collect the multi-layer profile image of this coating.

[0060] In the embodiments of the present invention, preprocessing is performed on the multi-layer profile image of the sample coating to obtain the multi-layer profile image of the sample coating after preprocessing, including: denoising, enhancing, and segmenting the multi-layer profile image of the sample coating to obtain the multi-layer profile image of the sample coating after preprocessing.

[0061] The preprocessing method provided in the embodiments of the present invention is only used as a preferred implementation manner. It is also possible to denoise or enhance the multi-layer profile image of the sample coating, and coating analysis can also be performed.

[0062] In the embodiments of the present invention, extracting the multi-layer coating feature information corresponding to the multi-layer profile image of the sample coating after preprocessing includes:

[0063] Based on the multi-layer profile image of the sample coating after preprocessing, obtaining the thickness and boundary morphology data of each layer in the multi-layer profile image of the sample coating to obtain the physical morphology features corresponding to the multi-layer profile image of the sample coating; for example: assuming that the pixels of the multi-layer profile image of the sample coating are A×B, then there are a total of B columns and A rows of pixel points. Also assuming that the boundary of each layer should be a straight line in an ideal state, that is, the values generated by the same boundary in B columns should be at the same position. Therefore, in order to avoid data changes caused by boundary offset, for any boundary, the pixel point closest to the lower edge of the image among the pixel points in each column of the boundary (assuming that the coating boundary should be parallel to the lower edge in an ideal state) is used as the boundary morphology data, that is, B pixel coordinate values are obtained as the boundary morphology data.

[0064] Obtaining the composition type coding of each layer in the multi-layer profile image of the sample coating input by the staff to obtain the chemical morphology features corresponding to the multi-layer profile image of the sample coating;

[0065] Taking the physical morphology features and the chemical morphology features together as the multi-layer coating feature information corresponding to the multi-layer profile image of the sample coating.

[0066] In the embodiments of the present invention, a machine learning algorithm is used to establish a data relationship between the multi-layer coating feature information and the coating sample performance index to obtain a coating performance analysis model, including:

[0067] Using a machine learning algorithm to construct an initial learning model, and using a chaotic mapping initialization method to initialize the model parameters of the initial learning model to obtain a plurality of different model parameter vectors;

[0068] For any model parameter vector, using the coating multi-layer feature information as the input of the initial learning model and the coating sample performance index corresponding to the coating multi-layer feature information as the expected output, obtain the loss function value corresponding to the model parameter vector;

[0069] According to the loss function values corresponding to all model parameter vectors, obtain the optimal model parameter vector; that is, the model parameter vector with the smallest loss function value is the optimal model parameter vector.

[0070] For any model parameter vector, according to the optimal model parameter vector, adopt an adaptive following strategy to perform local rapid search on the model parameter vector to obtain the model parameter vector after local rapid search;

[0071] For the model parameter vector after local rapid search, adopt an oscillation interval search strategy to perform local oscillation search on the model parameter vector to obtain the model parameter vector after local oscillation search;

[0072] For the model parameter vector after local oscillation search, adopt a double global mutation strategy to perform global search on the model parameter vector to obtain the model parameter vector after global search;

[0073] Judge whether the number of training times has reached the maximum number of training times. If so, according to the model parameter vector after global search, re-determine the optimal model parameter vector, and use the re-determined optimal model parameter vector as the final parameter of the initial learning model to obtain the coating performance analysis model; otherwise, return to the step of obtaining the loss function value corresponding to the model parameter vector.

[0074] In the prior art, the gradient descent algorithm or other algorithms are often used to optimize the parameters of the machine learning algorithm, resulting in the training process falling into a local optimum, and ultimately leading to the inability to achieve data relationship learning. Therefore, the embodiments of the present invention provide a new training algorithm to improve the problems existing in the prior art, enhance the global search ability and search effect of the algorithm, and ultimately improve the coating multi-layer profile analysis effect and accuracy to assist the staff in achieving better coating performance analysis.

[0075] In the embodiments of the present invention, a chaotic mapping initialization method is used to initialize the model parameters of the initial learning model to obtain multiple different model parameter vectors, including:

[0076] For the model parameters of the initial learning model, randomly initialize between the upper limit and the lower limit of the model parameters, and encode the initialized model parameters into a vector to determine the model parameter vector after random initialization;

[0077] Based on the model parameter vector after the random initialization, perform a chaotic mapping to obtain multiple different model parameter vectors as follows:

[0078]

[0079] Among them, represents the k-th model parameter vector, and when k = 1, represents the model parameter vector after random initialization, represents the (k + 1)-th model parameter vector, represents the chaotic mapping parameter.

[0080] The chaotic mapping initialization method provided by the embodiments of the present invention can make the initial solutions more evenly distributed in the solution space, and can effectively improve the training effect and training speed of the algorithm.

[0081] In the embodiments of the present invention, for any model parameter vector, according to the optimal model parameter vector, an adaptive following strategy is adopted to perform local fast search on the model parameter vector to obtain the model parameter vector after local fast search, including:

[0082] Based on the current training times, use an exponential function to obtain the adaptive balance search factor as:

[0083]

[0084] Among them, represents the adaptive balance search factor, e represents the natural constant, t represents the current training times, and T represents the maximum training times;

[0085] According to the adaptive balance search factor and the optimal model parameter vector, perform local fast search on the model parameter vector to obtain the model parameter vector after local fast search as:

[0086]

[0087] Among them, represents the i-th model parameter vector in the t-th training process, i = 1, 2,..., K, K represents the total number of model parameter vectors, i represents the model parameter vector after local fast search , represents the optimal model parameter vector, represents the model parameter vector corresponding historical optimal value, represents the search factor as a constant, represents the first random number between (0, 1), represents the worst model parameter vector.​

[0088] The adaptive following strategy provided by the embodiments of the present invention enables each model parameter vector to learn the information of its historical optimal value, and at the same time can avoid the worst positions, and finally search around the optimal position, so that the algorithm can perform fast search.

[0089] In the embodiments of the present invention, for the model parameter vector after local fast search, an oscillation interval search strategy is adopted to perform local oscillation search on the model parameter vector, and the model parameter vector after local oscillation search is obtained, including:

[0090] For the model parameter vector after local fast search, obtain the oscillation search factor as:

[0091]

[0092] Wherein, represents the oscillation search factor in the t-th training process, represents the oscillation search factor in the (t - 1)-th training process, and in the first training process, the oscillation search factor is set to a random number between (0, 1), represents the adjustment parameter and is set to a constant factor between (0, 2);

[0093] According to the oscillation search factor and the current training times, obtain the adaptive oscillation search weight as:

[0094]

[0095] Wherein, represents the adaptive oscillation search weight, represents the maximum value of the adaptive oscillation search weight, represents the minimum value of the adaptive oscillation search weight.

[0096] For the model parameter vector after local fast search, randomly match an other model parameter vector for the model parameter vector to obtain the local cooperation vector corresponding to each model parameter vector;

[0097] According to the adaptive oscillation search weight and the local cooperation vector, perform local oscillation search on the model parameter vector, and the model parameter vector after local oscillation search is:

[0098]

[0099]

[0100] Wherein, denotes the model parameter vector after the m-th local fast search in the t-th training process, where m = 1, 2, …, K, and K represents the total number of model parameter vectors. denotes the model parameter vector corresponding local collaboration vector denotes the collaboration factor denotes the second random number between (0, 1) denotes the model parameter vector in the t-th training process corresponding search speed denotes the model parameter vector in the (t + 1)-th training process corresponding search speed denotes the model parameter vector after local oscillatory search .

[0101] The oscillatory interval search strategy provided by the embodiments of the present invention can perform oscillatory search between two model parameter vectors, which can not only effectively search unfamiliar regions, but also provide more possibilities to search for the global optimal solution, and will not affect the search accuracy of the algorithm in the later stage of the algorithm, and can significantly improve the training effect of the algorithm.

[0102] In the embodiments of the present invention, for the model parameter vector after local oscillatory search, a double global mutation strategy is adopted to perform global search on the model parameter vector to obtain the model parameter vector after global search, including:

[0103] Generate the first global transformation factor and the second global transformation factor as:

[0104]

[0105]

[0106] Among them, denotes the first global transformation factor denotes the natural constant denotes the cosine function denotes the global transformation scale parameter, and , h represents the upper limit value of the global transformation scale parameter, and b represents the transformation shape control factor; denotes the second global transformation factor denotes generating a random step size through Lévy flight; denotes (-2 , 2 ) random number;

[0107] For the model parameter vector after local oscillation search, according to the first global transformation factor and the second global transformation factor, a spiral search path is used to globally search the model parameter vector, and the model parameter vector after global search is obtained as follows:

[0108]

[0109] Wherein, represents the model parameter vector after the nth local oscillation search in the tth training process, n = 1, 2,..., K, and K represents the total number of model parameter vectors, represents a random model parameter vector, represents the model parameter vector after global search , represents , the search control factor between, and represents the minimum value of the search control factor, represents the maximum value of the search control factor, represents a random search direction factor with a uniform distribution between (0, 1), represents pi, and cos represents the cosine function.

[0110] Optionally, after global search, a greedy algorithm can also be used to control the global search to ensure the search speed of the algorithm. After each search, the model parameter vector can also be processed to prevent it from going out of bounds, so as to ensure that the training is always effective.

[0111] A method for coating multi-layer profile analysis based on data analysis provided by the present invention uses an image acquisition device to acquire an image of the profile to determine a sample coating multi-layer profile image, then extracts the coating multi-layer feature information corresponding to the preprocessed sample coating multi-layer profile image, uses a machine learning algorithm to establish a data relationship between the coating multi-layer feature information and the coating sample performance index to obtain a coating performance analysis model, and finally can acquire a target coating multi-layer profile image corresponding to the coating to be analyzed and extract the target coating multi-layer feature information corresponding to the target coating multi-layer profile image, and use the coating performance analysis model to analyze the target coating multi-layer feature information to determine the coating multi-layer profile analysis result corresponding to the coating to be analyzed, which can effectively improve the accuracy and reliability of coating analysis, is applicable to the profile analysis of various coating materials, and is of great significance for improving the quality of coating products and optimizing the production process.

[0112] Other embodiments of the present invention will be readily apparent to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the invention following the general principles of the invention and including known common general knowledge or conventional technical means in the technical field not disclosed by the present invention. It should be understood that the present invention is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A coating multilayer profile analysis method based on data analysis, characterized in that: include: The coating sample is cut, and the cutting direction is ensured to be perpendicular to the coating surface, and an image acquisition device is used to acquire an image of the cross section to determine a multi-layer cross-sectional image of the sample coating; Preprocessing the sample coating multi-layer cross-sectional image to obtain the sample coating multi-layer cross-sectional image after the preprocessing, and extracting the coating multi-layer feature information corresponding to the sample coating multi-layer cross-sectional image after the preprocessing; The multi-layer cross-sectional image of the sample coating after preprocessing and the multi-layer characteristic information of the coating are provided to the staff to obtain the coating sample performance index input by the staff through human-computer interaction; A machine learning algorithm is used to establish the data relationship between the multi-layer characteristic information of the coating and the performance indicators of the coating samples, and a coating performance analysis model is obtained; Collecting a multi-layer cross-sectional image of a target coating corresponding to the coating to be analyzed, extracting multi-layer characteristic information of the target coating corresponding to the multi-layer cross-sectional image of the target coating, and analyzing the multi-layer characteristic information of the target coating using a coating performance analysis model to determine a multi-layer cross-sectional analysis result of the coating corresponding to the coating to be analyzed; The machine learning algorithm is used to establish the data relationship between the coating multi-layer feature information and the coating sample performance indicators, and the coating performance analysis model is obtained, including: An initial learning model is constructed by using a machine learning algorithm, and model parameters of the initial learning model are initialized by using a chaotic mapping initialization method to obtain multiple different model parameter vectors; For any model parameter vector, the coating multilayer feature information is used as the input of the initial learning model, the coating sample performance index corresponding to the coating multilayer feature information is used as the expected output, and the loss function value corresponding to the model parameter vector is obtained; Obtain the optimal model parameter vector according to the loss function values ​​corresponding to all model parameter vectors; For any model parameter vector, according to the optimal model parameter vector, an adaptive following strategy is used to perform a local fast search on the model parameter vector to obtain the model parameter vector after the local fast search; For the model parameter vector after the local fast search, an oscillation interval search strategy is used to perform a local oscillation search on the model parameter vector to obtain the model parameter vector after the local oscillation search; For the model parameter vector after the local oscillation search, a dual global mutation strategy is used to perform a global search on the model parameter vector to obtain the model parameter vector after the global search; Determine whether the number of training times reaches the maximum number of training times. If so, re-determine the optimal model parameter vector according to the model parameter vector after the global search, and use the re-determined optimal model parameter vector as the final parameter of the initial learning model to obtain the coating performance analysis model. Otherwise, return to the step of obtaining the loss function value corresponding to the model parameter vector; The model parameters of the initial learning model are initialized using the chaotic mapping initialization method to obtain multiple different model parameter vectors, including: For the model parameters of the initial learning model, randomly initialize them between the upper limit and the lower limit of the model parameters, encode the model parameters after the initialization into a vector, and determine the model parameter vector after the random initialization; Based on the model parameter vector after random initialization, chaotic mapping is performed to obtain a plurality of different model parameter vectors; For any model parameter vector, according to the optimal model parameter vector, an adaptive following strategy is used to perform a local fast search on the model parameter vector to obtain the model parameter vector after the local fast search, including: Based on the current number of training times, an exponential function is used to obtain the adaptive balance search factor; According to the adaptive balance search factor and the optimal model parameter vector, a local fast search is performed on the model parameter vector to obtain the model parameter vector after the local fast search; For the model parameter vector after the local fast search, an oscillation interval search strategy is used to perform a local oscillation search on the model parameter vector, and the model parameter vector after the local oscillation search is obtained, including: For the model parameter vector after the local fast search, an oscillation search factor is obtained, and according to the oscillation search factor and the current number of training times, an adaptive oscillation search weight is obtained; For the model parameter vector after the local fast search, randomly match another model parameter vector to the model parameter vector to obtain the local cooperation vector corresponding to each model parameter vector; Performing a local oscillation search on the model parameter vector according to the adaptive oscillation search weight and the local cooperation vector to obtain the model parameter vector after the local oscillation search; For the model parameter vector after the local oscillation search, a dual global mutation strategy is used to perform a global search on the model parameter vector to obtain the model parameter vector after the global search, including: generating a first global transformation factor and a second global transformation factor; For the model parameter vector after the local oscillation search, a spiral search path is used to perform a global search on the model parameter vector according to the first global transformation factor and the second global transformation factor to obtain the model parameter vector after the global search.

2. The coating multilayer profile analysis method based on data analysis according to claim 1, characterized in that: After determining the coating multi-layer profile analysis result corresponding to the coating to be analyzed, it also includes: generating a coating multi-layer profile analysis report according to a preset template based on the target coating multi-layer profile image, the target coating multi-layer feature information and the coating multi-layer profile analysis result; wherein the coating multi-layer profile analysis report is in PDF format or Word format.

3. The coating multilayer profile analysis method based on data analysis according to claim 1, characterized in that: The image acquisition device is used to acquire images of the cross section to determine the multi-layer cross-sectional image of the sample coating, including: scanning the coating cross section with a scanning electron microscope or an atomic force microscope to obtain the multi-layer cross-sectional image of the sample coating.

4. The coating multilayer profile analysis method based on data analysis according to claim 1, characterized in that: Preprocessing the sample coating multi-layer cross-sectional image to obtain the preprocessed sample coating multi-layer cross-sectional image includes: denoising, enhancing and segmenting the sample coating multi-layer cross-sectional image to obtain the preprocessed sample coating multi-layer cross-sectional image.

5. The coating multilayer profile analysis method based on data analysis according to claim 1, characterized in that: Extract the coating multi-layer feature information corresponding to the sample coating multi-layer cross-section image after preprocessing, including: Based on the preprocessed multi-layer cross-sectional image of the sample coating, the thickness and boundary morphology data of each layer in the multi-layer cross-sectional image of the sample coating are obtained to obtain the physical morphological features corresponding to the multi-layer cross-sectional image of the sample coating; Obtain the component type code of each layer in the multi-layer cross-sectional image of the sample coating input by the staff, and obtain the chemical morphological characteristics corresponding to the multi-layer cross-sectional image of the sample coating; The physical morphological features and the chemical morphological features are used together as the coating multi-layer feature information corresponding to the sample coating multi-layer cross-sectional image.

Images