Method and system for determining surface aging degree of silicone rubber

Abstract

The invention discloses a method and system for determining the surface aging degree of silicone rubber, and belongs to the technical field of insulator performance evaluation. The method comprises the following steps: acquiring Fourier infrared spectrum data of silicone rubber surface composite insulators at different aging levels; training a target neural network model according to the Fourier infrared spectrum data to generate an analysis model; for the target silicone rubber surface composite insulator, obtaining target Fourier infrared spectrum data; and determining the aging degree of the target silicone rubber surface based on the analysis model according to the target Fourier infrared spectrum data. According to the invention, the efficiency of data analysis is effectively improved, the aging degree of the silicone rubber surface can be accurately determined, and the deviation caused by human factors is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of insulator performance evaluation, and more particularly to a method and a system for determining the aging degree of a silicone rubber surface. Background Art

[0002] Composite insulators play a vital role in the power system. They are indispensable components in modern transmission lines and substations. Their main function is to provide electrical insulation, ensuring that during the transmission of high-voltage power, the current only flows along the wire and does not leak into the surrounding environment, or to avoid electrical short circuits between conductors of different voltage levels. At the same time, they are also responsible for physical support, fixing and maintaining the appropriate distance between the wire and the tower or other supporting structures to ensure the safe and stable operation of the transmission line. Compared with traditional porcelain or glass insulators, composite insulators have better resistance to pollution flashover, which is one of the common faults in power systems, especially in dusty and humid environments, where the accumulation of dirty substances will reduce the insulation performance.

[0003] The aging of composite insulators is a complex process caused by multiple factors, mainly involving physical, chemical and electrical aspects. These factors work together to cause the gradual degradation of the performance of composite insulators. Aging can lead to the degradation of the performance of composite insulators, which is manifested in decreased hydrophobicity, reduced insulation performance, and weakened mechanical strength. Therefore, aging monitoring and maintenance of composite insulators are crucial to ensure the safe operation of power systems. Summary of the invention

[0004] In view of the above problems, the present invention proposes a method for determining the degree of aging of the silicone rubber surface, comprising:

[0005] Obtain Fourier infrared spectroscopy data of composite insulators with silicone rubber surfaces at different aging levels;

[0006] Training a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model;

[0007] For the target silicone rubber surface composite insulator, obtain the target Fourier infrared spectrum data;

[0008] Based on the analysis model and the target Fourier transform infrared spectrum data, the aging degree of the target silicone rubber surface is determined.

[0009] Optionally, training a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model includes:

[0010] Preprocessing the Fourier transform infrared spectrum data to generate target data;

[0011] Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set;

[0012] Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model;

[0013] The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

[0014] Optionally, preprocessing the Fourier transform infrared spectrum data includes:

[0015] The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

[0016] Optionally, define the target neural network model based on the type of model input data.

[0017] Optionally, the defined target neural network model is a convolutional neural network model.

[0018] Optionally, the training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting the learning rate, setting the batch size, setting the loss function, and setting the optimizer.

[0019] Optionally, inputting the test set into the neural network model for training includes:

[0020] The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

[0021] Optionally, testing the original analysis model with the test set to determine the performance of the original analysis model includes:

[0022] The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

[0023] Optionally, preset evaluation indicators include at least one of the following: accuracy, recall and F1 score of the original analysis model relative to the test set.

[0024] Optionally, adjusting the original analysis model includes: adjusting a hyperparameter of the original analysis model;

[0025] The hyperparameters include at least one of the following: learning rate and batch size.

[0026] In another aspect, the present invention further provides a system for determining the degree of aging of a silicone rubber surface, comprising:

[0027] The first data acquisition unit is used to obtain Fourier infrared spectrum data of silicone rubber surface composite insulators of different aging levels;

[0028] A training unit, used to train a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model;

[0029] The second data acquisition unit is used to obtain target Fourier infrared spectrum data for the target silicone rubber surface composite insulator;

[0030] An output unit is used to determine the aging degree of the target silicone rubber surface based on the analysis model according to the target Fourier transform infrared spectrum data.

[0031] Optionally, training a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model includes:

[0032] Preprocessing the Fourier transform infrared spectrum data to generate target data;

[0033] Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set;

[0034] Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model;

[0035] The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

[0036] Optionally, preprocessing the Fourier transform infrared spectrum data includes:

[0037] The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

[0038] Optionally, define the target neural network model based on the type of model input data.

[0039] Optionally, the defined target neural network model is a convolutional neural network model.

[0040] Optionally, the training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting the learning rate, setting the batch size, setting the loss function, and setting the optimizer.

[0041] Optionally, inputting the test set into the neural network model for training includes:

[0042] The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

[0043] Optionally, testing the original analysis model with the test set to determine the performance of the original analysis model includes:

[0044] The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

[0045] Optionally, preset evaluation indicators include at least one of the following: accuracy, recall and F1 score of the original analysis model relative to the test set.

[0046] Optionally, adjusting the original analysis model includes: adjusting a hyperparameter of the original analysis model;

[0047] The hyperparameters include at least one of the following: learning rate and batch size.

[0048] In yet another aspect, the present invention further provides a computing device, comprising: one or more processors;

[0049] a processor for executing one or more programs;

[0050] When the one or more programs are executed by the one or more processors, the above-described method is implemented.

[0051] In yet another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the method described above is implemented.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] The present invention proposes a method for determining the aging degree of a silicone rubber surface, comprising: obtaining Fourier infrared spectrum data of silicone rubber surface composite insulators of different aging levels; training a target neural network model according to the Fourier infrared spectrum data to generate an analysis model; obtaining target Fourier infrared spectrum data for a target silicone rubber surface composite insulator; and determining the aging degree of the target silicone rubber surface according to the target Fourier infrared spectrum data based on the analysis model. The present invention effectively improves the efficiency of data analysis, can accurately determine the aging degree of the silicone rubber surface, and reduces the deviation caused by human factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of the method of the present invention;

[0055] Figure 2 A flowchart of an embodiment of the method of the present invention;

[0056] Figure 3 A schematic diagram of a network structure of an embodiment of the method of the present invention;

[0057] Figure 4 It is a structural diagram of the system of the present invention. DETAILED DESCRIPTION

[0058] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0059] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0060] Embodiment 1:

[0061] The present invention proposes a method for determining the degree of aging of the silicone rubber surface. Figure 1As shown, including:

[0062] Step 1, obtaining Fourier infrared spectrum data of silicone rubber surface composite insulators with different aging levels;

[0063] Step 2: training the target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model;

[0064] Step 3, acquiring target Fourier transform infrared spectrum data for the target silicone rubber surface composite insulator;

[0065] Step 4: Determine the aging degree of the target silicone rubber surface based on the analysis model and the target Fourier transform infrared spectrum data.

[0066] The target neural network model is trained according to the Fourier transform infrared spectrum data to generate an analysis model, including:

[0067] Preprocessing the Fourier transform infrared spectrum data to generate target data;

[0068] Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set;

[0069] Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model;

[0070] The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

[0071] The Fourier transform infrared spectrum data is preprocessed, including:

[0072] The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

[0073] Among them, the target neural network model is defined according to the type of model input data.

[0074] Among them, the defined target neural network model is a convolutional neural network model.

[0075] Among them, the training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting the learning rate, setting the batch size, setting the loss function and setting the optimizer.

[0076] The step of inputting the test set into the neural network model for training includes:

[0077] The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

[0078] The step of testing the original analysis model with the test set to determine the performance of the original analysis model includes:

[0079] The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

[0080] Among them, the preset evaluation indicators include at least one of the following: accuracy, recall rate and F1 score of the original analysis model relative to the test set.

[0081] Wherein, adjusting the original analysis model includes: adjusting the hyperparameters of the original analysis model;

[0082] The hyperparameters include at least one of the following: learning rate and batch size.

[0083] The present invention is further described below in conjunction with specific implementation cases:

[0084] Case flow Figure 2 As shown, including:

[0085] (1) First, it is necessary to collect Fourier infrared spectral data of composite insulators at different aging levels. These data should cover samples of different ages, different aging degrees, and different pollution conditions to ensure the diversity and representativeness of the data. The data should also be classified according to aging level and divided into training and test sets.

[0086] (2) Preprocess the raw spectral data, including baseline correction, eliminating background interference, and ensuring data quality. At the same time, smoothing is performed as needed to reduce the impact of noise.

[0087] (3) Define a neural network model in a deep learning framework (TensorFlow or PyTorch). The choice of model can be determined based on the form and scale of the input data. If the input is one-dimensional data, a multi-layer perceptron (MLP) can be established; if the input is a Fourier infrared waveform image, a convolutional neural network model (CNN) can be established to extract features from the image; if the wavelength signal of the Fourier infrared image is regarded as a time series, a long short-term memory (LSTM) or an attention mechanism can be used to extract features from the wavelength signal. These neural network models usually include input preprocessing, feature extraction, and downstream task modules. The model structure with the appropriate scale and running speed can be selected according to the needs.

[0088] (4) After determining the model structure, you need to select the model's training settings, determine the number of neurons in each layer, learning rate, batch size, loss function, and optimizer, etc.

[0089] (5) The composite insulator Fourier transform infrared spectrum data set established in step (2) is used to train the neural network model. During the training process, the model minimizes the classification and regression loss values ​​in step (4), uses the optimizer to optimize the model gradient, and finally determines the weight of the model so that it can obtain the best aging degree classification effect.

[0090] (6) Use the test set to evaluate the trained model. The model outputs the probability distribution of each aging level and selects the level with the highest probability as the prediction result. Calculate the accuracy, recall, F1 score and other indicators of the model on the test set to evaluate the performance of the model.

[0091] (7) Analyze the performance of the model on the test set and adjust the model's hyperparameters, such as learning rate, batch size, etc., based on the test results to improve model performance.

[0092] Furthermore, in step (3), when creating the model, multiple models of different sizes are simultaneously established. In theory, a smaller model can achieve faster running speed and smaller memory requirements, which can reduce the hardware requirements for model operation, and a larger model can achieve higher accuracy. Models of multiple different sizes can be applied in different application scenarios.

[0093] Furthermore, in step (3), the one-dimensional data is stacked into high-dimensional data for processing. One-dimensional data usually represents a single sequence or vector. After stacking it into two-dimensional or multi-dimensional data, the neural network can learn and capture the correlation between different dimensions, which helps the model understand the complex patterns and structures in the data. For image and video data, after stacking into two-dimensional or three-dimensional data, the convolutional layer of the convolutional neural network (CNN) can be used to detect local features and spatial relationships. For time series data, the one-dimensional convolutional layer can also identify temporal correlations on the stacked data. At the same time, high-dimensional data input can increase the capacity of the neural network, allowing the model to fit more complex function mappings. In some cases, converting one-dimensional data into high-dimensional data is a means of feature engineering that can help the model better understand and distinguish different data points, especially when the original data itself lacks sufficient information.

[0094] Furthermore, in step (6), a visualized heat map is used to show which parts of the input Fourier transform infrared spectrum in the deep learning model have a greater impact on the aging degree of the composite insulator. A heat map is created using a data visualization tool such as Matplotlib, Seaborn or Plotly. The horizontal axis represents the input feature, and the vertical axis can be left blank or represent the hidden layer neurons. The color depth represents the importance of the wavelength feature, and generally the darker the color, the greater the impact.

[0095] Furthermore, in step (7), hyperparameters are adjusted by grid search, random search or Bayesian optimization to further improve model performance.

[0096] Use Fourier infrared spectrometer to obtain Fourier infrared spectrum data of composite insulators. Preprocess the existing infrared images and data: Select representative composite insulator samples, covering different aging levels, to ensure that the samples are diverse in terms of aging degree, pollution status, etc. Before using the infrared spectrometer, ensure that the surface of the composite insulator is clean, free of dust, moisture and other contaminants to avoid affecting the measurement results. According to the shape and size of the insulator, decide whether the sample needs to be cut or ground to meet the sampling requirements of the spectrometer. According to the needs of the composite insulator aging degree analysis, set the appropriate wavelength scanning range.

[0097] Preprocess the Fourier transform infrared spectral data. First, confirm that there are no missing or outliers in the spectral data. If there are missing values, fill them in according to the situation (such as using the average, median or interpolation method). Use moving average, median filtering and other methods to reduce random noise in the spectrum. Finally, convert the values ​​of the spectral data to the same range, such as [0, 1] or [-1, 1], such as Z-score standard deviation standardization, as shown below.

[0098]

[0099] Where X is the input data, μ is the mean, σ is the variance, X new is the standardized data.

[0100] Next, a neural network model is needed to treat the Fourier transform infrared spectroscopy data as a time series signal and use the long short-term memory network (LSTM) for feature extraction and analysis. Its structure is as follows: Figure 3 shown.

[0101] LSTM consists of three gating units and memory C, as shown in equations (1)-(4).

[0102] f t =σ(W f ·[H t-1 , x t ]+b f ) (1)

[0103]

[0104]

[0105]

[0106] where x t represents input, represents hidden state, and is the memory state;

[0107] Formula (1) represents the forget gate, and the activation value of the forget gate is determined by the current input x t and the hidden state h at the previous moment t-1 Calculated through a sigmoid activation function.

[0108] Formula (2) represents the input gate, which determines which information of the current input needs to be added to the memory, including the weight i obtained by calculating the input information through sigmoid t And the candidate memory calculated by tanh

[0109] Formula (3) represents the output gate, which determines which information in the memory is passed to the hidden state of the next moment or used as the current output. Its activation value o t From the current input x t and the hidden state h at the previous moment t-1 It is calculated by sigmoid, and then the final hidden state h is obtained by tanh calculation with memory. t .

[0110] Formula (4) represents the memory update, which is calculated from the results of the forget gate and the input gate.

[0111] LSTM is particularly suitable for processing sequence information and long-distance dependencies in spectral data. It arranges Fourier infrared spectral data in wavelength order to form one-dimensional time series data. Each row of data represents a spectral sample, and each column corresponds to the spectral intensity at a specific wavelength. As needed, the continuous wavelength data can be divided into sequences of fixed length as the input sequence of LSTM, while ensuring that the sequence length is suitable for model processing and that the key spectral features can be retained.

[0112] Use the deep learning framework pytorch to build the LSTM model: First, you need to import the libraries required to build and train the model; next, define a multi-layer perceptron model that includes an input layer, an LSTM layer, and an output layer. The number of LSTM layers can be one or more. With each additional layer, the model's expressiveness will theoretically be stronger, but it will also increase the difficulty of training and the computational cost. The relu activation function is added after the input layer and the hidden layer, and a nonlinear part is introduced into the model. The output layer usually uses the Softmax activation function to convert the output of the previous layer into a probability distribution of each category, so that each sample can be assigned to the category with the highest probability.

[0113] Next, we define the loss function and optimizer required for model training. In classification tasks, the commonly used loss function is the cross-entropy loss, as shown in formula (6).

[0114]

[0115] Among them, y and are the true value and the predicted value, respectively, n is the number of samples during training, and the cross entropy loss is used to measure the difference between the probability distribution predicted by the model and the true label. The gradient of the loss function with respect to the model parameters is calculated through the back propagation algorithm, and then these parameters are updated using an optimization algorithm (such as gradient descent, Adam, etc.) to minimize the loss function so that the model gradually learns to classify correctly.

[0116] The model is evaluated using the test set. The evaluation indicators mainly include: precision P, recall R, F1 score, intersection over union (IOU) and average accuracy (AP). The formulas for each indicator are as follows:

[0117]

[0118]

[0119]

[0120] During testing, the above evaluation indicators are used to comprehensively evaluate the model performance.

[0121] Analyze the performance of the model on the test set. According to the test results, adjust the model's hyperparameters, such as learning rate, batch size, etc., to improve the model performance. Use the adjusted hyperparameters to retrain the model and evaluate it on the test set. Continuously iterate the above steps, and further adjust the model's hyperparameters according to the test results and feedback information until satisfactory performance is achieved. Finally, use a visualized heat map to show which parts of the input Fourier transform infrared spectrum in the deep learning model have a greater impact on the aging degree of the composite insulator. Use data visualization tools such as Matplotlib, Seaborn, or Plotly to create a heat map to analyze the relationship between the Fourier transform infrared spectrum waveform of the composite insulator and the aging degree.

[0122] This method combines deep learning and insulator infrared Fourier spectroscopy to automatically extract features from composite insulator Fourier infrared spectroscopy data through a deep learning model, and classify the degree of aging of the composite insulator surface accordingly. Compared with traditional manual analysis methods, the deep learning model provides a more objective and consistent evaluation standard. This automated process significantly improves the efficiency of data analysis and reduces the deviation caused by human factors.

[0123] Embodiment 2:

[0124] The present invention also provides a system 200 for determining the degree of aging of a silicone rubber surface. Figure 4 As shown, including:

[0125] The first data acquisition unit 201 is used to obtain Fourier infrared spectrum data of silicone rubber surface composite insulators of different aging levels;

[0126] A training unit 202, used to train a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model;

[0127] The second data acquisition unit 203 is used to obtain target Fourier infrared spectrum data for the target silicone rubber surface composite insulator;

[0128] The output unit 204 is used to determine the aging degree of the target silicone rubber surface based on the analysis model according to the target Fourier transform infrared spectrum data.

[0129] The target neural network model is trained according to the Fourier transform infrared spectrum data to generate an analysis model, including:

[0130] Preprocessing the Fourier transform infrared spectrum data to generate target data;

[0131] Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set;

[0132] Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model;

[0133] The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

[0134] The Fourier transform infrared spectrum data is preprocessed, including:

[0135] The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

[0136] Among them, the target neural network model is defined according to the type of model input data.

[0137] Among them, the defined target neural network model is a convolutional neural network model.

[0138] Among them, the training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting the learning rate, setting the batch size, setting the loss function and setting the optimizer.

[0139] The step of inputting the test set into the neural network model for training includes:

[0140] The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

[0141] The step of testing the original analysis model with the test set to determine the performance of the original analysis model includes:

[0142] The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

[0143] Among them, the preset evaluation indicators include at least one of the following: accuracy, recall rate and F1 score of the original analysis model relative to the test set.

[0144] Wherein, adjusting the original analysis model includes: adjusting the hyperparameters of the original analysis model;

[0145] The hyperparameters include at least one of the following: learning rate and batch size.

[0146] The present invention effectively improves the efficiency of data analysis, can accurately determine the aging degree of the silicone rubber surface, and reduces the deviation caused by human factors.

[0147] Embodiment 3:

[0148] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute 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 (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of the method in the above embodiment.

[0149] Embodiment 4:

[0150] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both a built-in storage medium in a computer device and an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiment.

[0151] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0152] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0153] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0154] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0155] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0156] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for determining the degree of aging of a silicone rubber surface, characterized in that: include: Obtain Fourier infrared spectroscopy data of composite insulators with silicone rubber surfaces at different aging levels; Training a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model; For the target silicone rubber surface composite insulator, obtain the target Fourier infrared spectrum data; The aging degree of the target silicone rubber surface is determined based on the analysis model and the target Fourier transform infrared spectrum data.

2. The method according to claim 1, characterized in that The target neural network model is trained according to the Fourier transform infrared spectrum data to generate an analysis model, including: Preprocessing the Fourier transform infrared spectrum data to generate target data; Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set; Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model; The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

3. The method according to claim 2, characterized in that The preprocessing of the Fourier transform infrared spectrum data comprises: The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

4. The method according to claim 2, characterized in that: Define the target neural network model based on the type of model input data.

5. The method according to claim 2, characterized in that: The target neural network model is defined as a convolutional neural network model.

6. The method according to claim 2, characterized in that The training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting a learning rate, setting a batch size, setting a loss function, and setting an optimizer.

7. The method according to claim 2, characterized in that The step of inputting the test set into the neural network model for training comprises: The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

8. The method according to claim 2, characterized in that: The testing the original analysis model with the test set to determine the performance of the original analysis model includes: The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

9. The method according to claim 8, characterized in that The preset evaluation indicators include at least one of the following: accuracy, recall rate and F1 score of the original analysis model relative to the test set.

10. The method according to claim 2, characterized in that The adjusting the original analysis model comprises: adjusting a hyperparameter of the original analysis model; The hyperparameters include at least one of the following: learning rate and batch size.

11. A system for determining the degree of aging of a silicone rubber surface, characterized in that: include: The first data acquisition unit is used to obtain Fourier infrared spectrum data of silicone rubber surface composite insulators of different aging levels; A training unit, used for training a target neural network model according to the Fourier transform infrared spectrum data to generate an analysis model; The second data acquisition unit is used to obtain target Fourier infrared spectrum data for the target silicone rubber surface composite insulator; An output unit is used to determine the aging degree of the target silicone rubber surface based on the analysis model according to the target Fourier transform infrared spectrum data.

12. The system according to claim 11, characterized in that The target neural network model is trained according to the Fourier transform infrared spectrum data to generate an analysis model, including: Preprocessing the Fourier transform infrared spectrum data to generate target data; Classifying the target data according to the aging level, and dividing the classified Fourier transform infrared spectrum data into a training set and a test set; Defining a target neural network model and selecting training settings for the target neural network to set the target neural network model. After the settings are completed, inputting the test set into the target neural network model for training to generate an original analysis model; The original analysis model is tested with the test set to determine the performance of the original analysis model. If the performance meets the preset requirements, the original analysis model is used as the analysis model. If the performance does not meet the preset requirements, the original analysis model is adjusted until it meets the preset requirements.

13. The system according to claim 12, characterized in that The preprocessing of the Fourier transform infrared spectrum data comprises: The Fourier transform infrared spectrum data is subjected to baseline correction processing to eliminate background interference. After the background interference is eliminated, the Fourier transform infrared spectrum data is subjected to smoothing processing to remove the influence of noise.

14. The system according to claim 12, characterized in that Define the target neural network model based on the type of model input data.

15. The system according to claim 12, characterized in that The target neural network model is defined as a convolutional neural network model.

16. The system according to claim 12, characterized in that The training settings include at least one of the following: determining the number of neurons in each layer of the convolutional neural network model, setting a learning rate, setting a batch size, setting a loss function, and setting an optimizer.

17. The system according to claim 12, characterized in that The step of inputting the test set into the neural network model for training comprises: The image wavelength signal of the test set is extracted using a convolutional neural network model, and features of the image wavelength signal are extracted based on long short-term memory (LSTM) or attention mechanism (Attention). Based on the extracted features, the silicone rubber surface aging degree classification is output.

18. The system according to claim 12, characterized in that The testing the original analysis model with the test set to determine the performance of the original analysis model includes: The original analysis model is tested with the test set to obtain test results, and the performance of the original analysis model is determined according to preset evaluation indicators.

19. The system according to claim 18, characterized in that The preset evaluation indicators include at least one of the following: accuracy, recall rate and F1 score of the original analysis model relative to the test set.

20. The system according to claim 12, characterized in that The adjusting the original analysis model comprises: adjusting a hyperparameter of the original analysis model; The hyperparameters include at least one of the following: learning rate and batch size.

21. A computer device, characterized in that: include: one or more processors; a processor for executing one or more programs; When the one or more programs are executed by the one or more processors, the method according to any one of claims 1 to 10 is implemented.

22. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the method according to any one of claims 1 to 10 is implemented.

Images