A coating wear life prediction method and related devices

Abstract

This invention discloses a method and related equipment for predicting coating wear life, belonging to the field of coating life prediction technology. The method includes: acquiring a surface wear morphology image and a coating thickness sequence of the coating to be predicted within a specified time window; inputting the surface wear morphology image and the coating thickness sequence into a pre-trained coating life prediction model to obtain a predicted remaining life value; wherein, the coating life prediction model is constructed based on a two-stream neural network and is used to fuse the spatial features of the surface wear morphology image and the temporal features of the coating thickness sequence for life prediction. This invention overcomes the shortcomings of existing single-modal prediction techniques, such as insufficient information utilization and incomplete degradation characterization, by simultaneously utilizing two modal data—surface morphology image and thickness sequence—to comprehensively capture coating degradation information from both spatial and temporal dimensions, thus improving the accuracy of coating wear life prediction.

Description

Technical Field

[0001] This invention relates to the field of coating wear life prediction technology, and specifically to a coating wear life prediction method and related equipment. Background Technology

[0002] Wear-resistant coatings, as the surface protection structure of key equipment in complex service environments such as aerospace, rail transportation, and national defense, are prone to performance degradation and failure under the effects of particle erosion, adhesive wear, abrasive wear, fatigue spalling, and thermo-coupling. This seriously threatens the operational safety and service reliability of the equipment, so it is urgent to carry out research on its life prediction and condition early warning.

[0003] Currently, most wear-resistant coating life prediction technologies are based on single information sources to construct life prediction models, mainly relying on single degradation parameters such as friction coefficient, wear amount, and coating thickness variation for life characterization. However, wear-resistant coatings not only exhibit continuous thickness decay during service but also undergo dynamic evolution of surface wear morphology, such as microcrack propagation, localized spalling, wear track deepening, and substrate exposure. Current single-mode prediction techniques cannot fully reflect the actual failure evolution process of coating wear under complex service conditions, and still have shortcomings in prediction accuracy and precision. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method and related equipment for predicting coating wear life. The purpose is to overcome the shortcomings of the prior art, which relies on only a single degradation parameter for life characterization, resulting in insufficient information utilization and incomplete degradation characterization. By simultaneously utilizing two modal data, namely coating surface wear morphology images and coating thickness sequences, a more accurate prediction of the remaining coating life can be achieved.

[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: According to a first aspect of the present invention, a method for predicting coating wear life is provided, comprising: Obtain surface wear morphology images and coating thickness sequences of the coating to be predicted within a specified time window; The surface wear morphology image and the coating thickness sequence are input into a pre-trained coating lifetime prediction model to obtain the remaining lifetime prediction value; wherein, the coating lifetime prediction model is constructed based on a two-stream neural network and is used to fuse the spatial features of the surface wear morphology image and the temporal features of the coating thickness sequence for lifetime prediction.

[0006] In one possible implementation of the first aspect, the coating lifetime prediction model includes a visual flow branch and a geometric flow branch, wherein the visual flow branch is a two-dimensional convolutional neural network used to extract the spatial features; and the geometric flow branch is a one-dimensional convolutional neural network used to extract the temporal features.

[0007] In one possible implementation of the first aspect, when the coating lifetime prediction model fuses the spatial features and the temporal features, it dynamically generates visual flow weights and geometric flow weights based on the information entropy of the surface wear morphology image and the variance of the coating thickness sequence; the sum of the visual flow weights and the geometric flow weights is 1.

[0008] In one possible implementation of the first aspect, the step of fusing the spatial features and the temporal features in the coating lifetime prediction model includes: The spatial and temporal features, which are weighted by the visual flow weight and the geometric flow weight, are combined using a weighted stitching method to obtain the first fused feature; A bilinear fusion method is used to perform an outer product operation on the spatial features and temporal features weighted by the visual flow weight and geometric flow weight to obtain the second fused feature. The first fusion feature and the second fusion feature are mapped to a unified feature space and aggregated to obtain a fusion feature vector for lifetime prediction.

[0009] In one possible implementation of the first aspect, the remaining lifetime prediction value is the remaining lifetime duration between the current time and the coating failure time, or the remaining lifetime percentage corresponding to the current time.

[0010] In one possible implementation of the first aspect, the coating lifetime prediction model is trained in the following manner: Acquire coating surface wear morphology image samples and coating thickness sequence samples at multiple wear times, as well as the corresponding remaining lifetime label values; Using the image samples and thickness sequence samples as input, and the remaining lifetime label value as a supervision signal, the initial model is trained with mean squared error as the loss function to obtain the coating lifetime prediction model.

[0011] In one possible implementation of the first aspect, the end of the remaining lifetime marker value is determined based on at least one of the following: The coating thickness reaches the preset failure threshold, the exposed area of ​​the substrate exceeds the preset threshold, the friction coefficient increases abnormally, and the service performance indicators are lower than the allowable value.

[0012] In one possible implementation of the first aspect, before inputting the surface wear morphology image and the coating thickness sequence into the coating lifetime prediction model, the method further includes: The surface wear morphology image is subjected to grayscale conversion, size unification, histogram equalization, and illumination compensation; and The coating thickness sequence is subjected to noise reduction, outlier removal, and smoothing. The surface wear morphology image and the coating thickness sequence were synchronized in advance according to a unified time reference.

[0013] According to a second aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the aforementioned coating wear life prediction method.

[0014] According to a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned method for predicting coating wear life.

[0015] According to a fourth aspect of the present invention, a computer program product is provided that, when executed by a processor, implements the aforementioned coating wear life prediction method.

[0016] Compared with the prior art, the present invention has at least the following beneficial effects: This invention discloses a method for predicting the wear life of a coating. It acquires surface wear morphology images and coating thickness sequences within the same time window, and simultaneously processes both modal data using a prediction model based on a two-stream neural network. This allows for the extraction of spatial features reflecting surface morphology changes from the images and temporal features reflecting thickness reduction patterns from the thickness sequences. These two types of features are then fused to predict the remaining life. Compared to methods relying solely on single parameters such as friction coefficient, wear amount, or thickness, this method simultaneously captures degradation information in both spatial and temporal dimensions, providing a more comprehensive reflection of the coating's true state changes during wear service and overcoming the shortcomings of insufficient single-modal information representation. Within the two-stream neural network, surface morphology features and thickness temporal features are fused to achieve complementarity. When the representational capability of one modality declines due to operating condition interference or degradation stage characteristics, the other modality can still provide effective degradation evidence, thereby improving the accuracy of remaining life prediction and reducing the risk of misjudgment caused by fluctuations in single information. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the specific embodiments of the present invention, the drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a coating wear life prediction method according to the present invention.

[0019] Figure 2 This is a flowchart of a coating wear life prediction method according to an embodiment of the present invention.

[0020] Figure 3 This is a diagram of the dual-stream heterogeneous feature extraction and fusion network architecture in a coating wear life prediction method of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 Combination Figure 1 and Figure 2 This embodiment demonstrates the overall process of the coating wear life prediction method provided. This embodiment provides a coating wear life prediction method for predicting the remaining life of a wear-resistant coating under the current wear state. The method includes the following steps: S1: Obtain the surface wear morphology image and coating thickness sequence of the coating to be predicted within a specified time window.

[0023] In this step, two modal data of the coating within the same service cycle are simultaneously acquired using image acquisition and thickness measurement devices. The acquired surface wear morphology images reflect spatial morphological information such as coating surface texture, scratches, spalling pits, and abrasive grain distribution. The acquired coating thickness sequence consists of multiple thickness measurements arranged in chronological order, reflecting the degradation trend of the coating thickness over time. The surface wear morphology images and the coating thickness sequence are synchronized and aligned according to a unified time reference to ensure that they correspond to the same wear stage.

[0024] S2: Input the surface wear morphology image and the coating thickness sequence into a pre-trained coating lifetime prediction model to obtain the remaining lifetime prediction value; wherein, the coating lifetime prediction model is constructed based on a two-stream neural network and is used to fuse the spatial features of the surface wear morphology image and the temporal features of the coating thickness sequence for lifetime prediction.

[0025] In this step, the coating lifetime prediction model receives a surface wear morphology image and a coating thickness sequence as input. Through its internal two-stream neural network structure, it extracts spatial features from the image and temporal features from the thickness sequence. After fusing these two types of features, it outputs the predicted remaining lifetime value of the coating at the current moment. This predicted remaining lifetime value represents the remaining service time or percentage of the coating from the current moment to the expected failure time.

[0026] This embodiment acquires two modal data: surface wear morphology images and coating thickness sequences. It then uses a coating lifetime prediction model based on a two-stream neural network to extract features and fuse the two types of data for prediction. This comprehensively reflects the impact of coating thickness degradation trends and surface morphology evolution on coating lifetime. It overcomes the shortcomings of existing technologies that rely solely on a single degradation parameter for lifetime characterization, resulting in insufficient information utilization and incomplete degradation characterization. This improves the accuracy of coating wear lifetime prediction.

[0027] Example 2 The coating lifetime prediction model includes a visual flow branch and a geometric flow branch. The visual flow branch is a two-dimensional convolutional neural network used to extract the spatial features; the geometric flow branch is a one-dimensional convolutional neural network used to extract the temporal features.

[0028] See Figure 3 The coating life prediction model internally constructs two parallel feature extraction branches. The visual flow branch consists of a two-dimensional convolutional neural network, which includes two-dimensional convolutional layers, normalization layers, pooling layers, and fully connected layers. It can extract edge features, texture features, pitting contour features, and local exfoliation features layer by layer from the input preprocessed surface wear morphology image, and finally output a fixed-dimensional spatial feature vector. The geometric flow branch is composed of a one-dimensional convolutional neural network, which includes one-dimensional convolutional layers, pooling layers, and feature compression layers. This network extracts first-order rate of change features, second-order curvature features, and local pulsed degradation features from the preprocessed coating thickness sequence, ultimately outputting a fixed-dimensional temporal feature vector. .

[0029] In the visual flow branch, the process of extracting spatial feature vectors can be represented as:

[0030] In the formula, Indicates time The acquired wear images, Represents a two-dimensional convolutional neural network mapping. Indicates the parameters of the visual flow network. This represents the output spatial topography feature vector.

[0031] In the geometric flow branch, the extraction process of temporal feature vectors can be represented as:

[0032] In the formula, Indicates time The corresponding thickness sequence window, Represents a one-dimensional convolutional neural network mapping. Represents the parameters of the geometric flow network. This represents the output geometrically degenerate eigenvector.

[0033] By designing the two heterogeneous branches mentioned above, the model can use an adapted network structure to extract features based on the characteristics of image data and sequence data, thereby obtaining more discriminative degenerate feature representations.

[0034] Example 3 When fusing the spatial features and the temporal features, the coating lifetime prediction model dynamically generates visual flow weights and geometric flow weights based on the information entropy of the surface wear morphology image and the variance of the coating thickness sequence; the sum of the visual flow weights and the geometric flow weights is 1.

[0035] Within the coating lifetime prediction model, the information entropy is calculated from the grayscale distribution of the surface wear morphology image, and its calculation formula is as follows:

[0036] In the formula, Indicates time The corresponding image information entropy, The grayscale histogram of the image represents the first... The probability of each gray level appearing This represents the total number of gray levels. The higher the information entropy, the more complex the image texture and the more disordered the wear morphology, usually corresponding to more significant crack propagation, increased peeling pits, or abrasive particle accumulation.

[0037] The variance is calculated from the coating thickness sequence data within a specified time window, and the calculation formula is as follows:

[0038] In the formula, Indicates time The corresponding thickness variance, Represents the first in the thickness sequence window Each sample value, This represents the average thickness within the corresponding window. This indicates the window length. A larger variance indicates a more drastic change in thickness, reflecting that the wear process may have transitioned from a steady thinning to a stage of localized sudden drop or unstable degradation.

[0039] After calculating the information entropy and thickness variance, both are first subjected to minimum-maximum normalization. The normalization formula is as follows:

[0040] In the formula, Indicates the indicator to be normalized. and Let these represent the minimum and maximum values ​​in the sample set, respectively. This represents a very small positive number that prevents the denominator from being zero. Based on the normalized information entropy... and normalized thickness variance Generate visual flow weight coefficients and geometric flow weighting coefficient The calculation formula is as follows:

[0041]

[0042] And satisfy .

[0043] This embodiment introduces information entropy and variance as quantitative indicators of the richness of information in each modality, enabling the model to automatically adjust the contribution ratio of visual flow and geometric flow at different wear stages: when the surface texture complexity increases, the weight of visual flow increases; when the thickness sequence shows significant fluctuations, the weight of geometric flow increases. This mechanism avoids the problem that fixed weights are difficult to adapt to changes in wear degradation characteristics at multiple stages, and improves the model's adaptability to complex degradation processes.

[0044] Example 4 The steps of fusing the spatial features and the temporal features in the coating lifetime prediction model include: combining the spatial features and temporal features weighted by the visual flow weight and the geometric flow weight using a weighted splicing method to obtain a first fused feature; performing an outer product operation on the spatial features and temporal features weighted by the visual flow weight and the geometric flow weight using a bilinear fusion method to obtain a second fused feature; mapping the first fused feature and the second fused feature to a unified feature space and aggregating them to obtain a fused feature vector for lifetime prediction.

[0045] Specifically, the coating lifetime prediction model has two parallel fusion paths built inside.

[0046] The first fusion path performs weighted concatenation of the weighted features, and its expression is:

[0047] In the formula, This represents the concatenated feature vector. This represents a vector concatenation operation. This path is used to preserve the first-order statistical information and relative independence of visual and geometric features.

[0048] The second fusion path performs bilinear fusion on the weighted features, and its expression is as follows:

[0049] In the formula, This represents the feature vector after bilinear fusion. This represents the outer product operation. This indicates that the matrix is ​​flattened into a vector. This bilinear channel can explicitly model the multiplicative coupling relationship between visual and geometric features, enhancing the ability to identify abrupt wear stages.

[0050] After obtaining the features of the two paths, and Each feature vector is projected onto a unified feature space using a trainable mapping matrix and then aggregated to obtain the final fused feature vector. Its expression is:

[0051] In the formula, The trainable mapping matrix represents the linearly concatenated channels. This represents the trainable mapping matrix for bilinear channels. Indicates the bias term. This represents a non-linear activation function.

[0052] As an alternative, the bilinear fusion part can be implemented using a compressed bilinear fusion method to reduce the computational burden caused by the increase in feature dimensions.

[0053] This embodiment employs a hybrid fusion strategy combining weighted stitching and bilinear fusion. Weighted stitching preserves the independent first-order statistical information of each modal feature, while bilinear fusion establishes a higher-order interaction between spatial and temporal features. This allows for a more comprehensive capture of the correlation patterns between image morphology changes and thickness degradation during coating wear. After unified mapping and nonlinear aggregation, the two types of features form a more expressive fused feature representation, effectively improving the characterization of complex wear mechanisms and the accuracy of remaining life prediction.

[0054] Example 5 The remaining lifetime prediction value is the remaining lifetime duration between the current time and the coating failure time, or the remaining lifetime percentage corresponding to the current time. In the coating lifetime prediction model, the regression prediction layer receives the final fused feature vector. As input, the remaining life prediction is obtained through a nonlinear regression function mapping, and its expression is:

[0055] In the formula, Indicates time The predicted remaining life of the wear-resistant coating. Indicates that there are trainable parameters The nonlinear regression function. When the output format uses remaining lifetime, It directly represents the remaining service time from the current moment to the coating failure moment. When the output format is a percentage of remaining life,... It represents the proportion of the remaining lifespan at the current moment to the total lifespan.

[0056] This embodiment provides two optional result output formats, enabling the prediction method to adapt to engineering needs in different application scenarios. It can directly obtain the remaining service time to guide maintenance scheduling, or obtain the percentage format to facilitate comparison of life status across devices or operating conditions.

[0057] As an optional application method, after obtaining the predicted remaining lifespan, the predicted value can be compared with a preset threshold. When the predicted remaining lifespan is lower than the preset threshold, the coating is determined to enter an early warning state, thus providing a basis for maintenance decisions.

[0058] Example 6 The coating lifetime prediction model is trained as follows: acquire coating surface wear morphology image samples and coating thickness sequence samples at multiple wear times, as well as corresponding remaining lifetime labels; use the image samples and thickness sequence samples as input, the remaining lifetime labels as supervision signals, and train the initial model with mean square error as the loss function to obtain the coating lifetime prediction model.

[0059] During training, industrial cameras were first used to acquire images of the coating surface wear morphology at different wear stages, and the images were preprocessed. Preprocessing included grayscale conversion, size unification, histogram equalization, and illumination compensation. Grayscale conversion reduced interference from color variations, allowing the system to focus more on structural information such as scratches, spalling pits, cracks, and abrasive particles. Size unification met the input requirements of the convolutional network. Histogram equalization enhanced texture contrast. Illumination compensation reduced the effects of ambient brightness and shadows. Simultaneously, a laser displacement sensor was used to acquire coating thickness variation sequences, and the coating thickness data was preprocessed. Preprocessing included denoising, outlier removal, and smoothing. Denoising employed one or more of moving average filtering, moving median filtering, and linear interpolation to reduce the impact of sensor offset, workpiece sway, and stray interference, and to ensure the thickness data more accurately reflected the coating thinning process. Outlier removal eliminated outliers that significantly deviated from the overall trend, preventing outliers from affecting temporal feature learning. Smoothing processes suppress high-frequency fluctuations while preserving the overall downward trend and local variation information. After preprocessing the image and thickness sequence samples, the surface wear morphology images and coating thickness sequences are synchronized according to a unified time reference to ensure that the two types of samples correspond to the same wear stage. When the sampling frequencies of the surface image and thickness data are inconsistent, nearest neighbor matching, linear interpolation, or repeated sampling methods are used to complete time alignment. The lifetime endpoint of each sample is determined according to preset failure conditions, and the remaining lifetime value is labeled for the full lifetime cycle samples to obtain the corresponding remaining lifetime label value.

[0060] During training, image samples and thickness sequence samples are used as input to the initial model, and remaining lifetime labels are used as supervision signals. Mean squared error (MSE) is used as the loss function to constrain the deviation between the predicted results and the actual labeled results. The expression for the MSE loss function is:

[0061] In the formula, Indicates the number of training samples. Indicates the first The true remaining lifetime labeled value for each sample This represents the model's predicted value.

[0062] The coating lifetime prediction model is obtained by jointly updating all trainable parameters in the visual flow branch, geometric flow branch, adaptive weight module, hybrid fusion layer and regression prediction layer through backpropagation until the loss function converges or the preset termination condition is met.

[0063] This embodiment uses an end-to-end joint training approach to enable the parameters of each module, including feature extraction, weight allocation, feature fusion, and lifetime prediction, to be optimized collaboratively. This ensures that the entire system learns with prediction accuracy as the unified goal, thereby obtaining a coating lifetime prediction model with better performance.

[0064] Example 7 The remaining lifespan value is determined based on at least one of the following conditions: the coating thickness reaches a preset failure threshold, the exposed substrate area exceeds a preset threshold, the friction coefficient increases abnormally, or the service performance index is lower than the allowable value.

[0065] When constructing training samples, one or more of the above criteria can be selected as the basis for determining the end of the lifespan, depending on the actual application scenario and coating type. When the coating state is detected to meet any of the selected criteria, the coating is determined to have reached the failure time. Based on this failure time, the remaining lifespan label value is calculated for samples at each time point throughout the entire lifespan.

[0066] This embodiment provides multiple optional failure criteria, enabling the lifetime labeling during model training to adapt to failure definitions under different coating types and service conditions, thereby improving the applicability of the trained coating lifetime prediction model in different application scenarios.

[0067] Example 8 Before inputting the surface wear morphology image and the coating thickness sequence into the coating lifetime prediction model, the method further includes: performing grayscale processing, size unification processing, histogram equalization processing, and illumination compensation processing on the surface wear morphology image; and performing noise reduction processing, outlier removal processing, and smoothing processing on the coating thickness sequence; the surface wear morphology image and the coating thickness sequence are synchronized and aligned in advance according to a unified time reference.

[0068] Specifically, the image preprocessing method is the same as that described in Example 6. The thickness sequence preprocessing method is the same as that described in Example 6. The time synchronization alignment method is the same as that described in Example 6.

[0069] This embodiment eliminates adverse factors such as noise, illumination differences, and scale inconsistencies in the original data by setting a preprocessing procedure consistent with that used in the training phase during the model inference phase. This ensures that the data quality input to the model remains consistent with that used during training, thereby guaranteeing that the coating lifetime prediction model is stable and accurate under actual deployment conditions. Time synchronization alignment ensures that image samples and thickness samples correspond to the same wear stage, guaranteeing that the features extracted by the dual-flow branch reflect the coating degradation state at the same moment.

[0070] It should be noted that the image preprocessing order, thickness denoising method, weight generation order, fusion layer implementation method, and regression layer structure described in the above embodiments can be adjusted or equivalently replaced according to specific working conditions in practical applications. Without departing from the core concept of this invention based on dual-stream feature extraction, adaptive weight adjustment, and hybrid fusion prediction, these adjustments or replacements all fall within the scope of protection claimed by this invention.

[0071] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used in the operation of a coating wear life prediction method.

[0072] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be Random Access Memory (RAM) or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the coating wear life prediction method in the above embodiments.

[0073] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0074] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] The present invention also provides a computer program product for executing any of the coating wear life prediction methods described above. Since the computer program product provided by the present invention belongs to the same inventive concept as the coating wear life prediction method described above, it possesses all the advantages of the coating wear life prediction method described above. Therefore, the beneficial effects of the computer program product provided by the present invention will not be elaborated further here.

[0078] In this invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0079] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention.

Claims

1. A method for predicting coating wear life, characterized in that, include: Obtain surface wear morphology images and coating thickness sequences of the coating to be predicted within a specified time window; The surface wear morphology image and the coating thickness sequence are input into a pre-trained coating lifetime prediction model to obtain the remaining lifetime prediction value; wherein, the coating lifetime prediction model is constructed based on a two-stream neural network and is used to fuse the spatial features of the surface wear morphology image and the temporal features of the coating thickness sequence for lifetime prediction.

2. The coating wear life prediction method according to claim 1, characterized in that, The coating lifetime prediction model includes a visual flow branch and a geometric flow branch. The visual flow branch is a two-dimensional convolutional neural network used to extract the spatial features. The geometric flow branch is a one-dimensional convolutional neural network used to extract the temporal features.

3. The coating wear life prediction method according to claim 1, characterized in that, When fusing the spatial features and the temporal features, the coating lifetime prediction model dynamically generates visual flow weights and geometric flow weights based on the information entropy of the surface wear morphology image and the variance of the coating thickness sequence; the sum of the visual flow weights and the geometric flow weights is 1.

4. The coating wear life prediction method according to claim 3, characterized in that, The steps of fusing the spatial features and the temporal features in the coating lifetime prediction model include: The spatial and temporal features, which are weighted by the visual flow weight and the geometric flow weight, are combined using a weighted stitching method to obtain the first fused feature; A bilinear fusion method is used to perform an outer product operation on the spatial features and temporal features weighted by the visual flow weight and geometric flow weight to obtain the second fused feature. The first fusion feature and the second fusion feature are mapped to a unified feature space and aggregated to obtain a fusion feature vector for lifetime prediction.

5. The coating wear life prediction method according to claim 1, characterized in that, The remaining lifetime prediction value is the remaining lifetime duration between the current time and the coating failure time, or the remaining lifetime percentage corresponding to the current time.

6. The coating wear life prediction method according to claim 1, characterized in that, The coating lifetime prediction model was trained in the following manner: Acquire coating surface wear morphology image samples and coating thickness sequence samples at multiple wear times, as well as the corresponding remaining lifetime label values; Using the image samples and thickness sequence samples as input, and the remaining lifetime label value as a supervision signal, the initial model is trained with mean squared error as the loss function to obtain the coating lifetime prediction model.

7. The coating wear life prediction method according to claim 6, characterized in that, The end of the remaining lifetime value is determined based on at least one of the following: The coating thickness reaches the preset failure threshold, the exposed area of ​​the substrate exceeds the preset threshold, the friction coefficient increases abnormally, and the service performance indicators are lower than the allowable value.

8. The coating wear life prediction method according to claim 1, characterized in that, Before inputting the surface wear morphology image and the coating thickness sequence into the coating lifetime prediction model, the method further includes: The surface wear morphology image is subjected to grayscale conversion, size unification, histogram equalization, and illumination compensation; and The coating thickness sequence is subjected to noise reduction, outlier removal, and smoothing. The surface wear morphology image and the coating thickness sequence were synchronized in advance according to a unified time reference.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a coating wear life prediction method as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a coating wear life prediction method as described in any one of claims 1 to 8.

Images