Equipment coating aging performance prediction method, application method and system

Abstract

The invention provides an equipment coating aging performance prediction method, an application method and a system, and the method comprises the steps: obtaining the current coating aging performance data of the equipment at the current moment and the historical coating aging performance data of the equipment at the historical moment; wherein the historical moment is a time point sequence before the current moment; and inputting the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by a plurality of long-short-term memory network layers for prediction to obtain coating aging performance data of the equipment at the next moment. According to the invention, the accuracy and real-time performance of predicting the aging performance of the coating of the equipment can be improved.

Description

Technical Field

[0001] The present invention relates to the field of predicting the aging performance of coatings, and particularly to a method, an application method and a system for predicting the aging performance of coatings of a device. Background Art

[0002] Coating technology plays an important role in the fields of materials science and protective engineering, especially in anti-corrosion and surface protection. As a protective material, the basic function of a coating is to shield and isolate and delay the corrosion process of materials. However, coatings will inevitably be affected by various environmental factors in the natural environment, such as: ultraviolet (UV) light, temperature and humidity changes, and chemical erosion, resulting in a gradual decline or even failure of the coating performance. Therefore, in-depth research on the aging performance of coatings is of great significance for improving the durability and reliability of coatings.

[0003] There are mainly two traditional methods for coating aging research: natural environment aging tests and artificial accelerated aging tests in the laboratory. Among them, natural environment aging tests, such as atmospheric aging tests, observe and record the aging process of coating samples by exposing them to typical or severe natural environments. Although the results of this method are closer to the actual application state, it usually takes several years and is inefficient; in addition, limited by the number of samples, it is difficult to comprehensively evaluate the coating performance. Artificial accelerated aging tests in the laboratory simulate the aging process of coatings in a short time by controlling and intensifying specific environmental factors (such as UV irradiation, temperature, humidity, etc.) to quickly obtain data. However, this method may not be able to fully reproduce the aging mechanism under natural conditions, so there are certain limitations. Summary of the Invention

[0004] To solve the problems of the prior art, the present invention proposes a method, an application method and a system for predicting the aging performance of coatings of a device, aiming to improve the accuracy and real-time performance of predicting the aging performance of coatings of a device.

[0005] The object of the present invention is achieved by the following technical solutions:

[0006] On the one hand, the present invention provides a method for predicting the aging performance of coatings of a device, the method comprising:

[0007] Obtaining the current coating aging performance data of the device at the current moment and the historical coating aging performance data at the historical moment; wherein, the historical moment is a time point sequence before the current moment;

[0008] Inputting the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by a plurality of long short-term memory network layers for prediction, to obtain the coating aging performance data of the device at the next moment.

[0009] Optionally, the process of constructing the coating aging performance prediction model includes:

[0010] Obtain the coating aging performance data set corresponding to the device at a historical time point sequence;

[0011] Preprocess the coating aging performance data set to obtain a preprocessed data set;

[0012] Use the preprocessed data set to perform iterative training and performance evaluation on a neural network model constructed by the multiple long short-term memory network layers to obtain the coating aging performance prediction model.

[0013] Optionally, the neural network model sequentially includes: two long short-term memory network layers, a first activation function layer, a dropout layer, a fully connected layer, and a second activation function layer.

[0014] Optionally, the step of using the preprocessed data set to perform iterative training and performance evaluation on a neural network model constructed by the multiple long short-term memory network layers to obtain the coating aging performance prediction model includes:

[0015] Randomly sample the preprocessed data set according to a set ratio to obtain a training set and a test set;

[0016] Use the training set and the test set to perform multiple rounds of iterative training and performance evaluation on the neural network model until the coating aging performance prediction model with performance evaluation indicators meeting preset conditions is obtained.

[0017] Optionally, the step of preprocessing the coating aging performance data set to obtain a preprocessed data set includes:

[0018] Perform outlier processing, data noise reduction, and data normalization on the coating aging performance data set in sequence to obtain an intermediate data set;

[0019] Construct samples for the intermediate data set according to the time sequence relationship between the input and output of the long short-term memory network layer to obtain the preprocessed data set.

[0020] Correspondingly, the present invention also provides a coating aging performance prediction system for a device, and the system includes:

[0021] A first acquisition module, configured to acquire the current coating aging performance data of the device at the current moment and the historical coating aging performance data at the historical moment; wherein, the historical moment is a time point sequence before the current moment;

[0022] The first prediction module is configured to input the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the device at the next moment.

[0023] On the other hand, the present invention provides an application method of a coating aging performance prediction method of a device in a power grid system. The coating aging performance prediction method of the device is any one of the above-mentioned coating aging performance prediction methods of the device. The application method includes:

[0024] The obtained current coating aging performance data and historical coating aging performance data are respectively the current power grid coating aging performance data and historical power grid coating aging performance data of the power grid device at the current moment;

[0025] Input the current power grid coating aging performance data and the historical power grid coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the power grid device at the next moment.

[0026] Correspondingly, the present invention further provides an application system of a coating aging performance prediction method of a device in a power grid system. The coating aging performance prediction method of the device is any one of the above-mentioned coating aging performance prediction methods of the device. The application system includes:

[0027] The second acquisition module is configured to obtain the current coating aging performance data and the historical coating aging performance data, which are respectively the current power grid coating aging performance data and the historical power grid coating aging performance data of the power grid device at the current moment;

[0028] The second prediction module is configured to input the current power grid coating aging performance data and the historical power grid coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the power grid device at the next moment.

[0029] On yet another aspect, the present invention further provides an electronic device, including: at least one processor and a memory; the memory and the processor are connected through a bus;

[0030] The memory is used for storing one or more programs;

[0031] When the one or more programs are executed by the at least one processor, the coating aging performance prediction method of the device as described in any one of the above is implemented, and the application method of the coating aging performance prediction method of the device as described above in a power grid system is implemented.

[0032] Correspondingly, the present invention further provides a readable storage medium with an execution program stored thereon. When the execution program is executed, it implements the coating aging performance prediction method for the device described in any one of the above, and the application method of the coating aging performance prediction method for the device described above in the power grid system.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] The present invention provides a method, an application method and a system for predicting the coating aging performance of a device. By means of a coating aging performance prediction model constructed by multiple LSTM network layers, the coating aging performance of the device is predicted. In this way, on the one hand, by using the coating aging performance prediction model constructed by multiple LSTM network layers, it is possible to achieve the prediction without having to establish a specific aging performance parameter model in advance, thus greatly simplifying the establishment process of the coating aging performance prediction model; on the other hand, due to the long-term and short-term memory ability of the LSTM network layer, it is particularly suitable for processing time series data and can effectively capture the temporal variation characteristics of the device during the coating aging process, thereby improving the prediction flexibility and accuracy of the coating aging performance prediction model.

[0035] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the technical solutions provided by the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings, where:

[0037] Figure 1 It is a schematic flow chart of a method for predicting the coating aging performance of a device provided by an embodiment of the present invention;

[0038] Figure 2 It is a schematic diagram of the unit structure and operation process of the long short-term memory network layer provided by an embodiment of the present invention;

[0039] Figure 3 It is a schematic flow chart of iterative training of a neural network model constructed by an LSTM network layer provided by an embodiment of the present invention;

[0040] Figure 4 It is a schematic diagram of the prediction curve fitting result of the low-frequency impedance modulus value in the damp heat test by the coating aging performance prediction model constructed by the present invention under different training ratios;

[0041] Figure 5Schematic diagram of the fitting result of the prediction curve of the adhesion force in the salt spray test by using the coating aging performance prediction model constructed at different training ratios provided by the present invention;

[0042] Figure 6 Flow schematic diagram of the application method of the coating aging performance prediction method of a device provided by an embodiment of the present invention in the power grid system;

[0043] Figure 7 Composition schematic diagram of a coating aging performance prediction system of a device provided by an embodiment of the present invention;

[0044] Figure 8 Composition schematic diagram of the application system of the coating aging performance prediction method of a device provided by an embodiment of the present invention in the power grid system;

[0045] Figure 9 Composition schematic diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0046] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for explaining the present invention, rather than limiting the protection scope of the present invention.

[0047] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0048] In the following description, the terms "first / second / third" involved are only used to distinguish similar objects, and do not represent a specific order for the objects. It can be understood that "first / second / third" can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present invention described here can be implemented in an order other than that illustrated or described here.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the embodiments of the present invention belong. The terms used herein are only for the purpose of describing the embodiments of the present invention, and are not intended to limit the embodiments of the present invention.

[0050] Coating technology plays an important role in the fields of materials science and protective engineering, especially in corrosion prevention and surface protection. As a protective material, the basic function of a coating is to shield and isolate, and delay the corrosion process of materials. However, coatings are inevitably affected by various environmental factors in the natural environment, such as UV light, temperature and humidity changes, and chemical erosion, which can easily lead to a gradual decline or even failure of the coating's performance. Therefore, in-depth research on the aging performance of coatings is of great significance for improving the durability and reliability of coatings.

[0051] There are mainly two methods for traditional coating aging research: natural environment aging tests and artificial accelerated aging tests in the laboratory. Among them:

[0052] 1. Natural environment aging tests, such as atmospheric aging tests, observe and record the aging process of coating samples by exposing them to typical or harsh natural environments. Although the results of this method are closer to the actual application state, it usually takes several years and is less efficient. In addition, due to the limited number of samples, it is difficult to comprehensively evaluate the coating performance.

[0053] 2. Artificial accelerated aging tests in the laboratory simulate the aging process of coatings in a short time by controlling and intensifying specific environmental factors (such as UV irradiation, temperature, humidity, etc.) to quickly obtain data. However, this method may not be able to fully reproduce the aging mechanism under natural conditions, so there are certain limitations.

[0054] In recent years, with the rapid development of artificial intelligence technology, especially the application of neural networks and deep learning, new technical means have been provided for the research of coating aging performance. Through big data analysis, a large amount of experimental data and actual application data can be integrated to establish an accurate coating aging prediction model. This coating aging prediction model can monitor the state of the coating in real time, predict the aging process, provide a scientific basis for maintenance decisions, achieve predictive maintenance, reduce unexpected downtime, and improve production efficiency. In addition, the development of an intelligent decision support system for coating aging based on deep learning can further optimize the coating maintenance plan and execution strategy, enhance the application performance and economic benefits of the coating. Therefore, the research on a high-precision coating aging parameter prediction model based on deep learning and big data has important practical significance and broad application prospects.

[0055] Based on the above description, the present invention proposes a method, an application method and a system for predicting the coating aging performance of a device, which uses a coating aging performance prediction model constructed by multiple LSTM network layers to predict the coating aging performance of the device. In this way, on the one hand, by using the coating aging performance prediction model constructed by multiple LSTM network layers, it is possible to achieve the establishment of the coating aging performance prediction model without the need to pre-establish a specific aging performance parameter model, thereby greatly simplifying the establishment process of the coating aging performance prediction model; on the other hand, due to the long-term and short-term memory ability of the LSTM network layer, it is particularly suitable for processing time series data and can effectively capture the temporal variation characteristics of the device during the coating aging process, thus improving the prediction flexibility and accuracy of the coating aging performance prediction model.

[0056] Embodiment 1:

[0057] The embodiment of the present invention provides a method for predicting the coating aging performance of a device. Refer to Figure 1 As shown, it is a schematic flowchart of a method for predicting the coating aging performance of a device provided by an embodiment of the present invention. Among them, in combination with Figure 1 The following description is made:

[0058] Step 101, obtain the current coating aging performance data of the device at the current moment and the historical coating aging performance data at the historical moment.

[0059] Among them, the historical moment is a time point sequence before the current moment.

[0060] In some embodiments of the present invention, the device can be any physical device with a coating on its surface. The specific material properties of the coating can be determined according to actual needs, and the present invention does not make any limitations in this regard.

[0061] In some embodiments of the present invention, the current moment can be the current time point for collecting relevant data or predicting the corresponding coating aging performance data, such as: 12:10:00 on xx year x month x day, etc. The corresponding historical moment can be: 00:00:00 on xx year x month x day to 12:09:59 on xx year x month x day, etc.

[0062] It should be noted that the historical moment is a time point sequence before the current moment. The duration corresponding to this time point sequence can be 10 hours, 1 day, 1 month, etc., and the sampling time interval corresponding to this time point sequence can be 1 second, 1 minute, 1 hour, etc. The present invention does not make specific limitations in this regard.

[0063] In some embodiments of the present invention, the current coating aging performance data and the historical coating aging performance data may include but are not limited to: gloss, hardness, adhesion, etc.; that is, the current coating aging performance data include but are not limited to: the gloss, hardness, adhesion, etc. of the coating at the current moment; correspondingly, the historical coating aging performance data include but are not limited to: the gloss, hardness, adhesion, etc. of the coating at historical moments.

[0064] It should be noted that the current coating aging performance data and the historical coating aging performance data may refer to the current coating aging performance characteristic parameters and the historical coating aging performance characteristic parameters, respectively.

[0065] Step 102: input the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the device at the next moment.

[0066] In some embodiments of the present invention, the LSTM network layer is a deep learning model established by improving the RNN, which has great advantages in predicting time series data. Figure 2 As shown in FIG. 1 , a schematic diagram of the unit structure and operation flow of the long short-term memory network layer provided by an embodiment of the present invention is shown; wherein the core of the LSTM network layer is composed of: a forget gate f t , input gate i t , output gate o t and the cell stores the state c t composition. Figure 2 Medium t is the output of the hidden state of this unit, c, t is the current time information, x t is the current input, c t-1 is the state at the previous moment, represents matrix element multiplication, Represents matrix addition.

[0067] Among them, the forget gate f t Determines the forgetting part in the state memory unit. It is done by t and the output h at the previous moment t-1 The information is then passed through the sigmoid activation function σ(x). This function outputs a value f between 0 and 1. t , where a value close to 1 means more historical information is retained, and a value close to 0 means more historical information is forgotten. The corresponding expression can be referred to as formula (1):

[0068] f t = σ(W f [h t-1 , x t + b f ) Formula (1);

[0069] Where, W f , b f are the weight matrix and bias of the forget gate respectively; σ is the sigmoid activation function.

[0070] The input gate i t functions to determine which parts should be retained in the state memory unit and update the unit information. It processes the current input x t and the output h t-1 at the previous moment by passing them into the sigmoid activation function and the tanh activation function respectively. These two functions generate an i t ranging from [0, 1] and a c ranging from [-1, 1], t . i t is used to control the degree to which c t is superimposed on the state quantity, and the larger the value, the more important the information. Finally, the output value of sigmoid is multiplied by the output value of tanh, and new information is selectively recorded into c t according to f t . The corresponding expression can be referred to as shown in Formula (2):

[0071]

[0072] The output gate o t determines the value of the next hidden state, and the corresponding expression can be referred to as shown in Formula (3):

[0073]

[0074] Where, W i , W c , W o are the weight matrices corresponding to each module, and b i , b c , b o are the biases. These parameters are optimized by the backpropagation algorithm during the training process.

[0075] In some embodiments of the present invention, the coating aging performance prediction model constructed by multiple long short-term memory network layers may further include: a dropout layer, a fully connected layer, etc., and the present invention does not make any limitation thereto.

[0076] It should be noted that by synchronously inputting the current coating aging performance data and the historical coating aging performance data into the coating aging performance prediction model for prediction, the coating aging performance data of the device at the next moment adjacent to the current moment can be obtained. Here, the coating aging performance data at the next moment includes, but is not limited to: the gloss, hardness, adhesion, etc. of the coating.

[0077] In some embodiments of the present invention, the construction process of the coating aging performance prediction model can ensure the construction of the coating aging performance prediction model based on the LSTM network layer through data processing and model construction, so as to achieve accurate prediction of subsequent coating performance. Here, it can be implemented in the following manner with reference to steps A1 to A3:

[0078] Step A1: Obtain the coating aging performance data set corresponding to the device at the historical time point sequence.

[0079] In some embodiments of the present invention, the coating aging performance data set corresponding to the device at the historical time point sequence can be collected through outdoor exposure tests or indoor accelerated tests. The coating aging performance data set includes, but is not limited to: data such as gloss, hardness, and adhesion at each time point corresponding to the historical time point sequence. Thus, these data are constructed into a comprehensive coating aging performance data set, and subsequent preprocessing can be performed on the coating aging performance data set as the basic input for subsequent related models.

[0080] Step A2: Preprocess the coating aging performance data set to obtain a preprocessed data set.

[0081] In some embodiments of the present invention, preprocessing the coating aging performance data set, where the preprocessing includes, but is not limited to: outlier processing, data denoising, data normalization, and sample construction, etc.

[0082] In some embodiments of the present invention, the above step A2 can be implemented through the following process:

[0083] First, perform outlier processing, data denoising, and data normalization on the coating aging performance data set in sequence to obtain an intermediate data set.

[0084] In some embodiments of the present invention, the data in the coating aging performance data set can be sequentially subjected to outlier processing, data denoising, and data normalization, thereby obtaining an intermediate data set.

[0085] It should be noted that regarding the outlier processing, due to the influence of objective factors such as measurement equipment and environment, outliers may appear in the coating aging performance dataset, which will inevitably have a significant impact on the prediction accuracy of subsequent related models. Therefore, the absolute median method can be used to identify these outliers, and the linear interpolation method of adjacent non-outlier values can be used to effectively fill these outlier points, so as to ensure the accuracy and consistency of the data in the processed coating aging performance dataset.

[0086] Similarly, regarding data denoising. Here, even after outlier processing, there may still be certain noise in the coating aging performance dataset. If the coating aging performance dataset after outlier processing is directly used for training, it will often prolong the training time of the model and reduce the model accuracy, resulting in a lower generalization ability of the model. Here, the Savitzky-Golay filter can be used to denoise the data. In this way, not only the noise can be effectively removed, but also the important change information in the data can be retained, thereby improving the training accuracy and efficiency of subsequent related models.

[0087] Correspondingly, regarding data normalization, the maximum-minimum normalization method can be used to normalize the coating aging performance dataset. The purpose is to map different types of data, that is, feature parameters, to a specific range, so as to eliminate the influence of dimension on the data without affecting the data characteristics, and to speed up the processing process of subsequent related prediction models and shorten the training time.

[0088] Then, according to the temporal relationship between the input and output of the long short-term memory network layer, the intermediate dataset is constructed into samples to obtain the preprocessed dataset.

[0089] In some embodiments of the present invention, since the coating aging performance prediction model is constructed by multiple long short-term memory network layers, when constructing the training sample data of the coating aging performance prediction model subsequently, the temporal relationship between the input and output of the long short-term memory network layer needs to be considered.

[0090] Correspondingly, the data in the intermediate dataset can be formed into samples corresponding to the structure of: [X t : Y] to obtain the preprocessed dataset. In this way, each sample includes the input X t and the output Y. Among them, X t is the input of a neural network model constructed by multiple LSTM network layers (this neural network model can obtain the coating aging performance prediction model after subsequent training and other operations), and it includes: the coating aging performance data (coating aging performance characteristics) within a certain time window t h before the current moment t, that is: X t = [S t-th , …, S t-1 , St , and S t = [x t , y t , z t , …], where x t , y t , z t etc. are respectively the aging performance data or parameter values of the coating at time t. Correspondingly, Y is the ideal output of the neural network model, which can be called the label of the sample, and its value is the historical aging performance parameter value of the coating at the (t + 1)-th moment, so as to ensure the continuity and dynamics of time series prediction.

[0091] In this way, by sequentially performing outlier processing, data denoising, and data normalization on the coating aging performance data set, and constructing samples according to the time series relationship between the input and output of the long short-term memory network layer, a preprocessed data set that meets the input of the neural network model constructed by multiple long short-term memory network layers is obtained, so as to provide parameter support for the training and performance evaluation of the neural network model.

[0092] Step A3: Use the preprocessed data set to perform iterative training and performance evaluation on the neural network model constructed by the multiple long short-term memory network layers to obtain the coating aging performance prediction model.

[0093] In some embodiments of the present invention, a preprocessed data set can be used to perform multiple rounds of iterative training and performance evaluation on the neural network model constructed by multiple long short-term memory network layers to obtain a coating aging performance prediction model that meets relevant conditions.

[0094] In some embodiments of the present invention, the neural network model constructed by multiple long short-term memory network layers may sequentially include: two long short-term memory network layers, a first activation function layer, a dropout layer (Dropout layer), a fully connected layer, and a second activation function layer.

[0095] In some embodiments of the present invention, the neural network model may mainly include: 2 LSTM network layers, 1 dropout layer, that is, a dropout layer, and 1 fully connected layer. In addition, a Relu activation function layer (that is, the first activation function layer and the second activation function layer mentioned above) can be added after each of the 2 LSTM layers and the fully connected layer.

[0096] In some embodiments of the present invention, two LSTM network layers are the core of the neural network model. Among them, the first LSTM network layer is mainly responsible for capturing the coating aging performance data, that is, the trends and patterns of the performance characteristics changing over time. The second LSTM network layer further analyzes the complex relationships between these trends and patterns, thereby improving the depth and accuracy of the prediction of the neural network model. After the two LSTM network layers, a Relu activation function layer can be added to introduce non-linear characteristics and enhance the learning ability of the neural network model for complex data relationships. After that, a Dropout layer is added to randomly discard a part of the network connections to reduce the overfitting phenomenon of the neural network model during the training process, thereby enhancing the generalization ability of the neural network model. Finally, the fully connected layer is responsible for integrating the learned features, and then a Relu activation function layer is used for non-linear transformation to further improve the ability of the neural network model to capture complex non-linear relationships. Through such a structural design, the model can effectively learn the dynamic changes of the coating performance and provide a reliable basis for prediction.

[0097] In this way, through iterative training and performance evaluation of the neural network model composed of two long short-term memory network layers, the first activation function layer, the Dropout layer, the fully connected layer and the second activation function layer, the prediction accuracy of the subsequent coating aging performance prediction model is improved.

[0098] In some embodiments of the present invention, to effectively implement the prediction of the coating aging performance, a technical solution of training and evaluating a neural network model constructed by multiple long short-term memory network layers can be adopted. It can ensure the efficiency and accuracy of the subsequent coating aging performance prediction model obtained through iterative training and performance evaluation by accurately splitting the data set, iterative training and detailed evaluation.

[0099] In some embodiments of the present invention, the above step A3 can be implemented through the following process:

[0100] First, randomly sample the preprocessed data set according to a set ratio to obtain a training set and a test set.

[0101] In some embodiments of the present invention, the preset data set can be segmented first. Exemplarily, a random sampling method can be adopted to divide the preprocessed data set into a training set and a test set. Further, the preprocessed data set can be randomly sampled according to a set ratio to select a certain proportion of the preprocessed data set as the training set, that is, for the training and parameter adjustment of the neural network model; the remaining preprocessed data set is used as the test set to verify the prediction ability of the neural network model.

[0102] It should be noted that the set ratio can be 30%, 50%, 70%, etc., and the present invention does not make specific limitations thereto.

[0103] Then, using the training set and the test set, the neural network model is subjected to multiple rounds of iterative training and performance evaluation until the coating aging performance prediction model with performance evaluation indicators meeting the preset conditions is obtained.

[0104] In some embodiments of the present invention, the neural network model can be first trained using the training set. That is, during the training process, each group of training samples (coating historical aging performance data) in the training set is input into the neural network model, and the neural network model will predict future performance based on the input coating historical aging performance data, that is, the coating historical aging performance characteristics. After each training iteration ends, the weight and bias parameters in the neural network model are updated through the backpropagation algorithm.

[0105] Correspondingly, the training process of the neural network model can be monitored and output in real time. For each training sample in the training set, the neural network model will output a predicted value Yi', which represents the prediction of the aging performance of the current input sample. These outputs will be compared with the actual aging performance data (labels of the samples) to calculate the error and performance indicators.

[0106] At the same time, the performance evaluation and optimization of the neural network model can be carried out synchronously. Here, in order to comprehensively evaluate the trained model, that is, the prediction effect of the neural network model, relevant evaluation indicators can be borrowed. Among them, the evaluation indicators include but are not limited to: root mean square error, mean absolute percentage error, goodness of fit, etc.

[0107] In some embodiments of the present invention, when evaluating the neural network model using the evaluation indicators, the label of each group of samples is compared with the actual output Y i ’ of this sample. If the error between the output of the label and Y i ’ meets the specified threshold of the evaluation indicator, it indicates that the training of the neural network model ends; otherwise, the hyperparameters of the neural network model (such as: learning rate, number of LSTM layers, number of neurons, etc.) are adjusted. After adjustment, the neural network model will be retrained and evaluated again. This process will be repeated until the evaluation indicators meet the threshold requirements, and the training of the neural network model ends, thereby obtaining the coating aging performance prediction model.

[0108] It should be noted that the performance evaluation indicators meeting the preset conditions can refer to using relevant performance evaluation indicators to evaluate the relevant parameters of the neural network model in the training process. For example, evaluating the accuracy of its output, etc. If the value corresponding to its accuracy is greater than the preset value, it is considered that the neural network model in this training process meets the requirements, that is, the coating aging performance prediction model for which the performance evaluation indicators meet the preset conditions.

[0109] In this way, training the model with the training set and evaluating the performance of the model with the help of the test set can enable the obtained final model to capture the key change characteristics in the aging process, thereby improving the prediction accuracy of the subsequent obtained coating aging performance prediction model.

[0110] Correspondingly, based on the coating aging performance data set corresponding to the device at historical time points, iterative training and performance evaluation are performed on the neural network model constructed by multiple LSTM network layers to construct a coating aging performance prediction model containing multiple LSTM units. In this way, on the one hand, by using the model constructed by the LSTM network layer, there is no need to pre-establish a specific aging performance parameter model, and directly using the coating aging performance data set for learning can greatly simplify the establishment process of the subsequent aging performance prediction model; on the other hand, the long-term and short-term memory ability of the LSTM network layer makes it particularly suitable for processing time series data and can effectively capture the temporal change characteristics in the coating aging process, thereby improving the prediction flexibility and accuracy of the coating aging performance prediction model.

[0111] It should be noted that in the present invention, in order to ensure the actual application effect and long-term effectiveness of the constructed coating aging performance prediction model for predicting the coating aging performance, model maintenance can be performed on the coating aging performance prediction model. In addition, before model maintenance, it is necessary to ensure the efficiency and accuracy of the coating aging performance prediction model in the actual environment through the deployment, actual application, monitoring, and regular calibration of the system for the coating aging performance prediction model. The following is a specific implementation process of a coating aging performance prediction model, that is, the trained LSTM network model, in actual application:

[0112] Step 1. Model deployment. Deploy the coating aging performance prediction model, that is, the coating aging performance prediction model that has been trained and verified, into the production environment. The deployment process includes integrating the coating aging performance prediction model into the existing industrial control system to ensure that the coating aging performance prediction model can receive real-time data and output prediction results. The deployment can use physical servers or cloud infrastructure, which can be specifically selected according to actual business requirements and resource configurations.

[0113] Step 2. Real-time Prediction. Once the model in Step 1 is deployed, the coating aging performance prediction model will start receiving real-time input data, which are automatically collected from the monitoring system of the coating equipment and cover various coating aging performance indicators. The coating aging performance prediction model performs immediate calculations based on the input data and outputs predictions of the future state of the coating to assist the maintenance team in making timely maintenance or replacement decisions.

[0114] Step 3. Performance Monitoring. During operation, the coating aging performance prediction model can continuously monitor its prediction performance. Here, the monitoring metrics include but are not limited to prediction accuracy, response time, etc.

[0115] Step 4. Model Calibration and Iteration. Here, the coating aging performance prediction model can be periodically calibrated based on the monitoring results and feedback collected in actual applications. This may include retraining the coating aging performance prediction model to adapt to new data patterns or adjusting model parameters to improve prediction accuracy. In addition, with the changes in coating technology and environmental conditions, the coating aging performance prediction model can also be iteratively updated periodically to ensure its long-term effectiveness and adaptability.

[0116] Based on the above description, in order to effectively address the numerous deficiencies in the prior art, the embodiments of the present invention propose a time series prediction method based on LSTM, specifically for predicting the coating aging performance parameters. In the prior art, although the natural exposure test is close to real environmental conditions, it has a long cycle and low efficiency and cannot provide accurate information on coating performance in a timely manner; while the accelerated aging test in the laboratory, although having a short time cycle, often cannot fully simulate the complex factors in the natural environment, resulting in the data obtained not being able to comprehensively reflect the aging performance change trend of the coating in actual use.

[0117] Based on this, the embodiments of the present invention effectively overcome the above problems by adopting an advanced LSTM network. The LSTM network can process and predict the time relationships in long sequence data, which is particularly crucial for the gradually changing process of coating performance aging. Through in-depth learning of historical data, this model can not only comprehensively consider the changes in various performance parameters of the coating during use but also dynamically predict the aging performance of the coating in the future for a period of time. This method can update the prediction results in real time, providing a scientific basis for the maintenance and replacement of the coating, thereby significantly improving the efficiency of coating management and reducing the potential risks caused by poor coating performance.

[0118] Reference Figure 3As shown in the figure, it is a schematic flowchart of iterative training of a neural network model constructed by an LSTM network layer provided by an embodiment of the present invention. Among them: First, obtain the original data of 301. Here, the original data can refer to the coating aging performance dataset corresponding to the device at a historical time point sequence. Secondly, the original data of 301 is successively subjected to: outlier processing of 302, SG filtering of 303, data normalization of 304, and sample construction of 305 to obtain a preprocessing dataset for iterative training and performance evaluation of a neural network model constructed by multiple LSTM network layers. Each set of data in this preprocessing dataset includes: labels of 306 and features (data) of 307. Then, the features of 307 are input into a neural network model 310 through 309. This neural network model 310 can successively include: an LSTM layer, an LSTM layer, a Relu layer, a Dropout layer, a fully connected layer, and a Relu layer, and determine the corresponding error, that is, 308, with the output of 311 and the labels of 306. Finally, based on the error of 308, iterative training of the neural network model of 310 is performed, that is, 312, to obtain a final coating aging performance prediction model.

[0119] In summary, a method for predicting coating aging performance by using a coating aging performance prediction model constructed based on an LSTM network layer provided by an embodiment of the present invention can effectively solve the problems of long cycle and incomplete data in the related art, thereby improving the accuracy and real-time performance of prediction, and further greatly optimizing the maintenance and management work of the coating to ensure the continuous performance and safety of the coating in practical applications.

[0120] Here, in order to enable those skilled in the art to better understand the technical solution of the present invention, the preferred solution of the present invention will be described below in combination with the best embodiments of specific applications. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0121] In the present invention, the low-frequency impedance modulus in the damp heat test and the adhesion in the salt spray test can be predicted respectively to verify the effectiveness of the coating aging performance prediction model constructed by multiple LSTM network layers provided by the embodiments of the present invention. The following is a detailed description of the best embodiments of the coating aging performance prediction model in practical applications.

[0122] (1) Experimental setup: In this invention, the low-frequency impedance modulus of a certain coating in the damp heat test and the adhesion in the salt spray test were statistically analyzed. For the damp heat test, referring to GB / T 1740-2007 "Determination of Resistance to Damp Heat, Damp Mildew and Salt Spray of Paints and Varnishes", the test temperature was selected as (47±1)°C and the relative humidity as (96±2)%RH for the experiment. The salt spray test was carried out referring to GB / T 1771-2007 "Determination of Resistance to Neutral Salt Spray of Paints and Varnishes". The low-frequency impedance modulus and adhesion of the coating were observed regularly. Considering that the coating performance changes little in the short term and it is difficult to accurately measure the performance differences, the linear interpolation method was used to supplement some data. The above data were used for the subsequent training and testing of the neural network model constructed by multiple LSTM network layers.

[0123] (2) Data preprocessing: The collected low-frequency impedance modulus and adhesion were first subjected to necessary preprocessing, including denoising, normalization, etc., to meet the input requirements of the coating aging performance prediction model. The preprocessed data were divided into training sets with different proportions, specifically: 20%, 30%, 40% and 50%. By setting the time window (t h ) to 10 hours, these data sets can be used to train the neural network model constructed by multiple LSTM network layers.

[0124] (3) Model training and evaluation: The neural network model constructed by multiple LSTM network layers was trained using the above training data sets with different segmentation ratios. The root mean square error was used as the loss function during the training process. After the model training was completed, it was evaluated to verify its prediction accuracy.

[0125] (4) Prediction result display: As shown in Figure 4 and Figure 5 , the performance of the coating aging performance prediction model constructed by multiple LSTM network layers was displayed through Figure 4 and Figure 5 . Figure 4 and Figure 5 show the prediction results corresponding to the coating aging performance prediction models constructed under different training sample ratios. In particular, when using 20% of the training samples, the training time of the model for predicting the low-frequency impedance modulus of the coating was only 3.769 seconds, and the root mean square error of the prediction was 0.144 Ω·cm 2 ; the training time of the model for predicting the adhesion of the coating was 3.722 seconds, and the root mean square error of the prediction was 0.121 Mpa. As the proportion of the training samples for the model increases, although the training time increases slightly, the prediction error of the neural network model constructed by multiple LSTM network layers decreases significantly, showing the high efficiency and accuracy of the model.

[0126] This embodiment shows that the method for predicting the coating aging performance of the device provided in the embodiment of the present invention is not only fast and efficient in the training process of the model, but also can provide accurate prediction results. Thus, it further verifies the feasibility and effectiveness of the method of the present invention in practical applications, especially suitable for industrial environments that require rapid and accurate prediction of aging performance. Table 1 summarizes the detailed prediction results of the neural network model constructed by multiple LSTM network layers under different settings of training sample ratios, further demonstrating the performance of the neural network model constructed by multiple LSTM network layers under various conditions.

[0127] Table 1 Prediction Results of Low-Frequency Impedance Modulus and Adhesion

[0128]

[0129] Based on the above description, an implementation method for predicting the coating aging performance of a device based on deep learning provided by an embodiment of the present invention mainly predicts the aging performance of the coating of the device by constructing and applying a deep learning network model, that is, a coating aging performance prediction model constructed based on LSTM network layers. The following aspects are mainly involved in the process of constructing the coating aging performance prediction model:

[0130] (1) Data preprocessing: First, obtain a dataset related to the aging performance of the coating and preprocess the data. This includes steps such as outlier processing, filtering and denoising, data normalization, and sample construction to ensure the quality of the data input into the model in the model training process and provide a solid foundation for high-precision prediction.

[0131] (2) Model construction based on LSTM network layers: Combining the advantages of the LSTM network, construct a neural network model containing multiple LSTM network layers. These LSTM network layers can effectively process and learn the long-term dependence relationships in time series data, especially suitable for complex physical processes such as coating aging that evolve over time. In addition, the neural network model also includes a Relu activation layer, a Dropout layer to reduce overfitting, as well as a fully connected layer and an output layer, jointly constituting a complete network learning model.

[0132] (3) Model training and testing: Divide the preprocessed dataset into a training set and a testing set. Use the training set to train the neural network model containing multiple LSTM network layers. By learning the historical performance data of coating aging, the neural network model containing multiple LSTM network layers can capture the key change features in the aging process. The obtained coating aging performance prediction model is then used to predict the real-time or future coating aging performance.

[0133] (4) Practical applications and optimization: The coating aging performance prediction model constructed in the present invention is not only applied to the actual prediction of coating performance, but also the structure and parameters of the coating aging performance prediction model can be continuously optimized according to the feedback in actual applications to adapt to different industrial environments and aging conditions.

[0134] The above key technical contents constitute the core of the present invention. Through these technical solutions, the present invention can provide a method for predicting the aging performance of coatings that is both fast and accurate, thereby significantly improving the accuracy and practicality of coating aging prediction and having important industrial application value.

[0135] In other words, by adopting the coating aging performance prediction model constructed based on the LSTM network layer, the present invention can effectively solve the deficiencies of traditional coating aging performance prediction methods and demonstrate significant technical effects. Among them, traditional coating aging performance prediction usually requires complex physical or chemical models, which are not only computationally cumbersome but also often ineffective in dealing with complex or unknown aging mechanisms. By using the coating aging performance prediction model constructed by the LSTM network layer, the present invention does not need to pre-establish a specific aging performance parameter model and directly uses historical aging performance data for learning, greatly simplifying the model establishment process and at the same time improving the flexibility and accuracy of prediction. And the long-term and short-term memory ability of the LSTM network layer makes it particularly suitable for processing time series data and can effectively capture the temporal variation characteristics in the coating aging process. At the same time, in actual applications, such as in the specific applications above, the coating aging performance prediction model adopted by the present invention shows in the application on the low-frequency impedance modulus dataset and the adhesion dataset that even when the number of training samples is small (such as 20% of the data volume), the model can not only complete training quickly but also accurately predict the trend of aging performance changing with time, and the root mean square error is small. This verifies the superiority of the model in terms of fast response and high-precision prediction. With the increase of the training data volume, the prediction accuracy is further improved, showing the good learning ability and adaptability of the model.

[0136] Thus, the method for predicting the aging performance of the coating of the device provided by the embodiment of the present invention has the following beneficial effects:

[0137] 1. High-efficiency and high-precision prediction: The use of the coating aging performance prediction model constructed by the LSTM network layer can significantly improve the efficiency and accuracy of predicting the aging performance of the device's coating, making the prediction process faster and more accurate. This is of great significance for coating maintenance work, can timely warn of potential aging problems, and effectively avoid major failures.

[0138] 2. Reduce production interruptions: By providing highly accurate aging prediction, production and maintenance personnel of the equipment can formulate effective maintenance measures in a timely manner, thereby reducing production interruptions and economic losses caused by aging problems.

[0139] 3. Strong adaptability: The coating aging performance prediction model constructed by the LSTM network layer of the present invention can adapt to aging data of different types and complexities, has good universality and scalability, and is applicable to a variety of industrial application scenarios.

[0140] 4. Simplify the model establishment process: Compared with traditional methods that require detailed physical or chemical knowledge for modeling, the data-driven coating aging performance prediction model can simplify the establishment and adjustment process of the model, thereby reducing the technical threshold for implementation.

[0141] In summary, the technical solution of the present invention not only solves the limitations of traditional methods, but also brings significant technical effects in practical applications, including improving prediction efficiency and accuracy, reducing production interruptions, and adapting to changing application environments, thereby providing an efficient and reliable technical support for coating maintenance.

[0142] The coating aging performance prediction method for the equipment provided by the embodiment of the present invention predicts the coating aging performance of the equipment by means of a coating aging performance prediction model constructed by multiple LSTM network layers. In this way, on the one hand, by using the coating aging performance prediction model constructed by multiple LSTM network layers, it is possible to achieve the establishment process of the coating aging performance prediction model without having to establish a specific aging performance parameter model in advance, so as to greatly simplify the establishment process of the coating aging performance prediction model; on the other hand, due to the long-term and short-term memory ability of the LSTM network layer, it is particularly suitable for processing time series data and can effectively capture the temporal change characteristics of the equipment during the coating aging process, thereby improving the prediction flexibility and accuracy of the coating aging performance prediction model.

[0143] Embodiment 2:

[0144] Based on the same inventive concept, the embodiment of the present invention also provides an application method of the coating aging performance prediction method for the equipment in the power grid system. Refer to Figure 6 As shown, it is a schematic flowchart of the application method of the coating aging performance prediction method for the equipment provided by the embodiment of the present invention in the power grid system; wherein, the coating aging performance prediction method for the equipment is the coating aging performance prediction method for the equipment described in the above embodiment, and the application method includes:

[0145] Step 601, the currently obtained coating aging performance data and historical coating aging performance data are the current power grid coating aging performance data and historical power grid coating aging performance data of the power grid equipment at the current moment, respectively.

[0146] Step 602: Input the current power grid coating aging performance data and the historical power grid coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, to obtain the coating aging performance data of the power grid equipment at the next moment.

[0147] In some embodiments of the present invention, for the relevant descriptions of the current power grid coating aging performance data and the historical power grid coating aging performance data of the power grid equipment at the current moment in step 601, reference can be made to the description of step 101 above, and the present invention makes no specific limitations thereon.

[0148] Correspondingly, the specific implementation manner of step 602 is similar to the specific implementation manner of step 102 above. Here, reference can also be made to the description of step 102, and the present invention makes no specific limitations thereon.

[0149] The application method of the coating aging performance prediction method for the equipment provided by the embodiments of the present invention in the power grid system uses a coating aging performance prediction model constructed by multiple LSTM network layers to predict the coating aging performance of the power grid equipment. In this way, on the one hand, by using a coating aging performance prediction model constructed by multiple LSTM network layers, it is possible to achieve without pre-establishing a specific aging performance parameter model, thus greatly simplifying the establishment process of the coating aging performance prediction model; on the other hand, due to the long short-term memory ability of the LSTM network layer, it is particularly suitable for processing time series data, and can effectively capture the temporal change characteristics of the power grid equipment during the coating aging process, thereby improving the prediction flexibility and accuracy of the coating aging performance prediction model, so as to timely warn of potential aging problems of the power grid equipment and provide strong support for the subsequent operation of the power grid equipment.

[0150] Embodiment 3

[0151] Based on the same inventive concept, the embodiments of the present invention also provide a coating aging performance prediction system for equipment. Refer to Figure 7 As shown, it is a schematic diagram of the composition of a coating aging performance prediction system for equipment provided by the embodiments of the present invention. In combination with Figure 7 the following description is made. The system 700 includes:

[0152] A first acquisition module 701, configured to acquire the current coating aging performance data of the equipment at the current moment and the historical coating aging performance data at the historical moment; wherein, the historical moment is a time point sequence before the current moment;

[0153] A first prediction module 702, configured to input the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, to obtain the coating aging performance data of the equipment at the next moment.

[0154] In some embodiments of the present invention, the system 700 further includes: a model construction module, and the model construction module includes:

[0155] An acquisition unit, configured to acquire a coating aging performance data set corresponding to the device at a historical time point sequence;

[0156] A preprocessing unit, configured to preprocess the coating aging performance data set to obtain a preprocessed data set;

[0157] A training and evaluation unit, configured to perform iterative training and performance evaluation on a neural network model constructed by the plurality of long short-term memory network layers by using the preprocessed data set, to obtain the coating aging performance prediction model.

[0158] In some embodiments of the present invention, the neural network model sequentially includes: two long short-term memory network layers, a first activation function layer, a dropout layer, a fully connected layer, and a second activation function layer.

[0159] In some embodiments of the present invention, the training and evaluation unit is specifically configured to randomly sample the preprocessed data set according to a set ratio to obtain a training set and a test set;

[0160] Use the training set and the test set to perform multiple rounds of iterative training and performance evaluation on the neural network model until the coating aging performance prediction model whose performance evaluation index meets a preset condition is obtained.

[0161] In some embodiments of the present invention, the preprocessing unit is specifically configured to perform outlier processing, data denoising, and data normalization on the coating aging performance data set in sequence to obtain an intermediate data set;

[0162] Construct samples for the intermediate data set according to the timing relationship between the input and output of the long short-term memory network layer to obtain the preprocessed data set.

[0163] It should be noted that the description of this system side is similar to the description of the above method embodiment, and has beneficial effects similar to those of the method embodiment. For technical details not disclosed in the system embodiment of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

[0164] Embodiment 4:

[0165] Based on the same inventive concept, an application system of a method for predicting the coating aging performance of a device in a power grid system is further provided in an embodiment of the present invention. Refer to Figure 8As shown in the figure, it is a schematic diagram of the composition of an application system of a method for predicting the coating aging performance of a device provided by an embodiment of the present invention in a power grid system. The method for predicting the coating aging performance of the device is the method for predicting the coating aging performance of the device described in the above embodiment. The application system 800 includes:

[0166] A third acquisition module 801, configured to acquire the current coating aging performance data and historical coating aging performance data, which are respectively the current power grid coating aging performance data and historical power grid coating aging performance data of the power grid device at the current moment;

[0167] A second prediction module 802, configured to input the current power grid coating aging performance data and the historical power grid coating aging performance data into a coating aging performance prediction model constructed by a plurality of long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the power grid device at the next moment.

[0168] It should be noted that the description of this application system is similar to the description of the above application method embodiment, and has similar beneficial effects to the application method embodiment. For the technical details not disclosed in the system embodiment of the present invention, please refer to the description of the application method embodiment of the present invention for understanding.

[0169] Embodiment 5:

[0170] Based on the same inventive concept, as Figure 9 shown in the figure, the present invention further provides an electronic device, which may be a computer device, a single-chip microcomputer device, a smart mobile device, etc. The electronic device in this embodiment may include a processor 910, a memory 920, a transceiver component 930, etc. The processor 910, the memory 920, and the transceiver component 930 are connected through a bus 940; the memory 920 may be used to store an execution program, and an exemplary execution program may include instructions; the processor 910 is used to execute the instructions stored in the memory. The memory 920 may also be used to store data, and the data may be called and / or modified when the instructions are executed.

[0171] The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete logic gate devices, or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding function, so as to implement the coating aging performance prediction method of the device in the above embodiments, and the application method of the coating aging performance prediction method of the device in the power grid system.

[0172] Embodiment 6:

[0173] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device-readable storage medium (Memory). The electronic device-readable storage medium is a memory device in the electronic device, used to store programs and data. It can be understood that the storage medium here can include both the built-in storage medium in the electronic device, and of course, can also include the extended storage medium supported by the electronic device. The storage medium provides a storage space, and this storage space stores the operating system of the terminal. And, one or more instructions suitable for being loaded and executed by the processor are also stored in this storage space. These instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory, or a non-volatile memory, such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium to implement the coating aging performance prediction method of the device in the above embodiments, and the application method of the coating aging performance prediction method of the device in the power grid system.

[0174] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0175] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to produce a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0176] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0177] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0178] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for predicting coating aging performance of equipment, characterized in that: The method comprises: Obtaining current coating aging performance data of the equipment at the current moment and historical coating aging performance data at historical moments; wherein the historical moments are a sequence of time points before the current moment; The current coating aging performance data and the historical coating aging performance data are input into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the equipment at the next moment.

2. The method according to claim 1, characterized in that The construction process of the coating aging performance prediction model includes: Obtaining a coating aging performance data set corresponding to a historical time point sequence of the device; Preprocessing the coating aging performance data set to obtain a preprocessed data set; The preprocessed data set is used to iteratively train and evaluate the performance of a neural network model constructed by the multiple long short-term memory network layers to obtain the coating aging performance prediction model.

3. The method according to claim 2, characterized in that The neural network model includes in sequence: two long short-term memory network layers, a first activation function layer, a dropout layer, a fully connected layer and a second activation function layer.

4. The method according to claim 2 or 3, characterized in that: The preprocessed data set is used to iteratively train and evaluate the performance of the neural network model constructed by the multiple long short-term memory network layers to obtain the coating aging performance prediction model, including: Randomly sampling the preprocessed data set according to a set ratio to obtain a training set and a test set; The training set and the test set are used to perform multiple rounds of iterative training and performance evaluation on the neural network model until the coating aging performance prediction model is obtained whose performance evaluation indicators meet preset conditions.

5. The method according to claim 2 or 3, characterized in that: The preprocessing of the coating aging performance data set to obtain a preprocessed data set includes: The coating aging performance data set is sequentially subjected to outlier processing, data noise reduction and data normalization to obtain an intermediate data set; According to the temporal relationship between the input and output of the long short-term memory network layer, sample construction is performed on the intermediate data set to obtain the preprocessed data set.

6. A coating aging performance prediction system for equipment, characterized in that: The system comprises: A first acquisition module is used to acquire the current coating aging performance data of the equipment at the current moment and the historical coating aging performance data at the historical moments; wherein the historical moments are a sequence of time points before the current moment; The first prediction module is used to input the current coating aging performance data and the historical coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the equipment at the next moment.

7. An application method of a coating aging performance prediction method for equipment in a power grid system, characterized in that: The coating aging performance prediction method of the equipment is a coating aging performance prediction method using the equipment described in any one of claims 1 to 5, and the application method comprises: The current coating aging performance data and the historical coating aging performance data obtained are respectively the current grid coating aging performance data and the historical grid coating aging performance data of the grid equipment at the current moment; The current power grid coating aging performance data and the historical power grid coating aging performance data are input into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the power grid equipment at the next moment.

8. An application system of a coating aging performance prediction method for equipment in a power grid system, characterized in that: The coating aging performance prediction method of the equipment is a coating aging performance prediction method of the equipment according to any one of claims 1 to 5, and the application system comprises: The second acquisition module is used to acquire current coating aging performance data and historical coating aging performance data, which are respectively the current grid coating aging performance data and historical grid coating aging performance data of the grid equipment at the current moment; The second prediction module is used to input the current power grid coating aging performance data and the historical power grid coating aging performance data into a coating aging performance prediction model constructed by multiple long short-term memory network layers for prediction, so as to obtain the coating aging performance data of the power grid equipment at the next moment.

9. An electronic device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the coating aging performance prediction method of the equipment as described in any one of claims 1 to 5 and the application method of the coating aging performance prediction method of the equipment as described in claim 7 in the power grid system are implemented.

10. A readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, the coating aging performance prediction method of the equipment as described in any one of claims 1 to 5 and the application method of the coating aging performance prediction method of the equipment as described in claim 7 in the power grid system are implemented.

Images