Tunnel disease monitoring data enhancement method and system based on CTGAN

Abstract

The invention provides a tunnel disease monitoring data enhancement method and system based on CTGAN, and relates to the technical field of data enhancement, and the method comprises the steps: carrying out the disease type marking and disease feature preprocessing of a real data table of a tunnel site disease; extracting time sequence features and spatial position features from the preprocessed disease features; taking the time sequence features, the spatial position features and the disease types as conditional information to train a CTGAN, and using the CTGAN to generate synthetic data having similar distribution with the preprocessed real data table to obtain a new sample data table; after the new sample data table is generated, verifying a synthesis effect by adopting a mode of combining a statistical method and a support vector machine; according to the method, the confrontation network is generated by using the condition table, tunnel disease monitoring data is supplemented and optimized based on a real data table of tunnel field diseases, the problems of data scarcity and category imbalance are solved, and the training effect of a machine learning model is improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of data enhancement, and in particular to a tunnel disease monitoring data enhancement method and system based on CTGAN. Background Art

[0002] During long-term operation, tunnels often experience structural defects such as lining cracks, water leakage, lining back voids, and lining deformation. These defects directly affect the safety and stability of tunnels. Therefore, tunnel health assessment is the key to ensuring safe tunnel operation. The assessment process relies on multiple defect characteristic variables, such as crack length, water leakage status, lining strength ratio, and void depth, and classifies tunnel conditions by health level.

[0003] However, actual real data tables usually have problems of data scarcity and category imbalance; due to the high cost of tunnel monitoring and complex monitoring conditions, there are fewer data sets with complete feature indicators; in addition, the types and severity of tunnel diseases vary greatly, and some categories of data in the data set are relatively scarce, making the training effect of the machine learning model unsatisfactory and difficult to accurately predict the health status of the tunnel.

[0004] In order to solve these problems, data enhancement technology is needed; tunnel disease monitoring data is mainly numerical tabular data, and traditional image or text data enhancement methods are not suitable for such data, resulting in poor data enhancement effect for tunnel disease monitoring, which in turn affects the accuracy of tunnel health status assessment. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a tunnel disease monitoring data enhancement method and system based on CTGAN, which uses a conditional table generative adversarial network (CTGAN) to supplement and optimize the tunnel disease monitoring data based on the real data table of tunnel on-site diseases, solve the problems of data scarcity and category imbalance, and improve the training effect of the machine learning model.

[0006] According to some embodiments, the present disclosure adopts the following technical solutions:

[0007] A tunnel disease monitoring data enhancement method based on CTGAN, comprising:

[0008] Obtain the real data table of tunnel site diseases, and mark the disease types and pre-process the disease characteristics;

[0009] Extracting time series features and spatial location features from the preprocessed disease features;

[0010] The CTGAN is trained with time series features, spatial location features and disease types as conditional information, and the trained CTGAN is used to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table;

[0011] After generating the new sample data table, the generated synthetic data effect is verified by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

[0012] According to some embodiments, the present disclosure adopts the following technical solutions:

[0013] A tunnel disease monitoring data enhancement system based on CTGAN, including an acquisition module, an extraction module, a synthesis module and a verification module:

[0014] The acquisition module is configured to: obtain the real data table of tunnel site diseases, and perform disease type labeling and disease feature preprocessing;

[0015] The extraction module is configured to: extract time series features and spatial location features from the preprocessed disease features;

[0016] The synthesis module is configured to: train the CTGAN using the time series features, spatial location features and disease types as conditional information, and use the trained CTGAN to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table;

[0017] The verification module is configured to: after generating a new sample data table, verify the effect of the generated synthetic data by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

[0018] According to some embodiments, the present disclosure adopts the following technical solutions:

[0019] A computer program product includes a computer program, and when the computer program is executed by a processor, the method for enhancing tunnel disease monitoring data based on CTGAN is implemented.

[0020] According to some embodiments, the present disclosure adopts the following technical solutions:

[0021] A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, a tunnel disease monitoring data enhancement method based on CTGAN is implemented.

[0022] According to some embodiments, the present disclosure adopts the following technical solutions:

[0023] An electronic device includes: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the tunnel disease monitoring data enhancement method based on CTGAN.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] The present invention effectively supplements scarce samples through the CTGAN generation mechanism and significantly improves the imbalance problem of category distribution; CTGAN not only generates diversified synthetic data through adversarial training and conditional generation mechanism, but also ensures the distribution consistency of synthetic data and original data, avoiding the disadvantages of traditional methods in generating redundant or invalid data; the synthetic data has high quality and stability, and can significantly improve the generalization ability of machine learning models in complex monitoring tasks; specifically, the synthetic data generated by CTGAN enhances the model's prediction ability under new data and different scenarios, reduces overfitting, and makes the model more robust in practical applications.

[0026] In addition, the present invention further proposes a highly automated data enhancement and verification scheme, which integrates the entire process of data preprocessing, generation and verification, so that the entire data enhancement and verification process is intelligent and adaptively adjusted. It can not only handle the complex features involved in tunnel monitoring data, but also ensure the integrity and consistency of the original data through automated preprocessing steps; data generation can automatically generate an appropriate amount of synthetic data according to the scarcity of different categories of data by dynamically adjusting the generation strategy of CTGAN, thereby improving the pertinence of data generation; data verification ensures the effectiveness and accuracy of generated data at multiple levels through multi-dimensional verification means, and has broad application potential, suitable for a variety of monitoring scenarios and other fields involving data scarcity or imbalance problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0028] Figure 1 The present invention is a method flow chart of an embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of the principle of outlier detection using a box plot according to an embodiment of the present disclosure.

[0030] Figure 3 FIG. 4 is a diagram showing the working principle of CTGAN according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0031] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0032] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0033] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0034] Because tunnel disease monitoring data is mainly numerical tabular data, and the Conditional Table Generative Adversarial Network (CTGAN) can process tabular data, the enhancement method based on CTGAN (Conditional Table Generative Adversarial Network) has become an effective solution. It can generate new numerical data samples to supplement the real data set and improve the model's adaptability to unbalanced data.

[0035] To ensure the validity of the enhanced data, the generated data needs to be verified. By using the support vector machine model, the enhanced data and the real data are trained separately, and the accuracy of the model's prediction of the tunnel health status is verified, thereby ensuring that the generated data not only increases in quantity but also improves the model's predictive performance in quality. This data enhancement and verification mechanism can effectively improve the generalization ability of the tunnel health assessment model and overcome the challenges of insufficient data and class imbalance.

[0036] Example 1

[0037] In one embodiment of the present disclosure, a tunnel disease monitoring data enhancement method based on CTGAN is provided, comprising:

[0038] Obtain the real data table of tunnel site diseases, and mark the disease types and pre-process the disease characteristics;

[0039] Extracting time series features and spatial location features from the preprocessed disease features;

[0040] The CTGAN is trained with time series features, spatial location features and disease features as conditional information, and the trained CTGAN is used to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table;

[0041] After generating the new sample data table, the generated synthetic data effect is verified by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

[0042] As an embodiment, a tunnel disease monitoring data enhancement method based on CTGAN is disclosed in the present invention. It uses a conditional table generation adversarial network (CTGAN) based on the real data table of tunnel field diseases to solve the problems of data scarcity and imbalance and improve the training effect of the machine learning model. The specific implementation process is as follows:

[0043] S1: Data preparation and preprocessing: collect tunnel site disease monitoring data, form a real data table, and label the disease type and preprocess the disease characteristics of the real data.

[0044] The real data table consists of several real data, and each column of data is a disease feature, as shown in Table 1:

[0045] Table 1 Real data table

[0046]

[0047] Processing of real data, including disease type labeling and disease feature preprocessing:

[0048] The disease type annotation is to add a disease type annotation to each real data, including lining cracks, water leakage and cavities behind the lining, etc. The disease characteristics are used to describe the specific indicators or characteristics of each disease type, including multiple characteristic indicators such as crack length, water leakage status and cavity depth, which are used to comprehensively reflect the severity and distribution of tunnel diseases.

[0049] The preprocessing is to perform data encoding, outlier detection and processing, missing value filling, normalization, etc. on the data of each disease characteristic, specifically:

[0050] S11: Convert non-numerical indicator data into numerical indicator data using one-hot encoding or label encoding. For categorical indicators that do not have a sequential relationship, one-hot encoding is usually used to ensure that each category is represented in the form of an independent binary vector. For indicators with a sequence, label encoding is used to maintain the relative order between categories.

[0051] S12: Detect outliers through box plots. The principle of box plots is as follows Figure 2 As shown in the figure, the structure of the box plot consists of the lower quartile (Q1), median (Q2), upper quartile (Q3), upper limit and lower limit.

[0052] The median represents the middle value of the data set and is located in the middle of the box; the lower quartile represents the location of the lower 25% of the data in the data set; the upper quartile represents the location of the lower 25% of the data in the data set; the interquartile range (IQR) is the difference between the upper quartile and the lower quartile, which is used to detect the range of outliers; data points outside the range of 1.5 times the interquartile range are generally considered to be outliers, that is, data points that are less than the lower limit and greater than the upper limit; the formula for calculating the range of outliers is as follows:

[0053] Lower limit=Q1-1.5IQR, upper limit=Q3+1.5IQR

[0054] For the detected outliers, we first conduct manual review, retain the important outliers, and delete the obviously wrong data; then replace the remaining outliers with the limit values, that is, the data less than the lower limit is replaced by the lower limit, and the data greater than the upper limit is replaced by the upper limit;

[0055] S13: After deleting outliers, for missing values ​​in the indicator data, first draw a data distribution histogram for each indicator with missing values ​​to intuitively display its data distribution and analyze the distribution characteristics of the missing values; then, select an appropriate method to fill in the missing values ​​based on the actual distribution and characteristics of the data; commonly used methods include interpolation or filling methods, through which the missing data are reasonably filled to ensure the integrity and consistency of the data;

[0056] S14: Normalize the values ​​of the indicator data using Min-Max normalization to map all values ​​to the range of [0,1] to avoid the large differences in the value range of the indicator data affecting model training.

[0057] S2: Extract time series features and spatial location features from the preprocessed disease features.

[0058] The time series characteristics are the trend information of crack length, width and depth changing over time. By arranging multiple observation records of the same location and the same disease, time series data are formed in chronological order. Linear regression, differential analysis and other methods are used to analyze the changing trends of these characteristics, and key indicators such as average growth rate, maximum growth rate and fluctuation range are extracted to describe the changing characteristics of the disease over time.

[0059] The spatial position feature is the specific position information in the tunnel structure. The specific position information is extracted through methods such as coordinate positioning, segment and position coding, and the relative spatial relationship between the disease features is analyzed to construct features reflecting the spatial distribution of the disease.

[0060] S3: Using CTGAN data augmentation technology, train the conditional generator G and discriminator D to generate synthetic data, that is, a new sample data table.

[0061] CTGAN generates synthetic data with similar distribution to real data through adversarial training of conditional generators and discriminators to solve the problems of small number of samples and unbalanced category distribution in real data.

[0062] CTGAN is used to generate synthetic data, such as Figure 3 As shown, the following steps are included:

[0063] Step S31: Generate a small batch of random vectors from the random noise distribution, and input them into the condition generator G together with the condition information. The generator generates corresponding synthetic data through the input.

[0064] In order to improve the spatiotemporal consistency of the generated data, the condition generator G is designed to receive the time series characteristics and spatial position characteristics in the tunnel disease monitoring data as conditional information; specifically, the crack monitoring data not only includes parameters such as crack length and width, but also introduces its trend over time and specific location information in the tunnel structure.

[0065] Tunnel disease monitoring data contains a variety of disease types, which can be used as conditional information input for the generator to ensure that the generated data is consistent with the real data in terms of category distribution.

[0066] In order to deal with the problem of class imbalance, the conditional generator G can adaptively adjust the class weights of the generated samples according to the data distribution, ensuring that the generated data not only covers the main categories but also generates a sufficient number of minority category samples.

[0067] The specific generation steps are:

[0068] (1) The condition generator G receives time series features, spatial location features, and disease types as condition information input;

[0069] (2) sampling a set of random vectors from a random noise distribution as input to the conditional generator to generate synthetic data;

[0070] (3) Due to the class imbalance problem in real data sets, the generator adaptively adjusts the class distribution of generated samples to ensure that minority disease data is fully generated.

[0071] Step S32: Sample a small batch of samples from the real data and input them into the discriminator D together with the data synthesized by the condition generator G. The discriminator is responsible for judging the authenticity of the data, and by comparing the distribution of the real data and the synthetic data, the loss function of the discriminator is calculated and the model parameters are updated.

[0072] Specifically, the discriminator combines time series and spatial position to construct a more complex identification criterion, thereby more accurately distinguishing real data from synthetic data. By comparing the distributions of the two, the discriminator's loss function is calculated and the model parameters are updated.

[0073] In addition to identifying the authenticity of the generated data, the discriminator can also learn the classification level of the disease. Specifically, the discriminator is designed as a multi-task learning architecture, which includes two tasks: one is used to determine whether the data is real or synthetic, and the other is used to classify the disease level. During the training process, a corresponding disease level label is provided for each real data sample. These labels indicate the severity of the disease for each sample in the real data, so that the discriminator can learn to distinguish different disease levels.

[0074] After receiving feedback from the discriminator, the conditional generator calculates the generator's loss function and updates its parameters to optimize the generated synthetic data to make it closer to the real data.

[0075] Tunnel disease data usually have clear physical limitations. Specifically, the crack length and width cannot exceed a certain threshold, the pH value of the leaking water should be within a certain range, etc. Physical constraints are added to the loss function of the generator to ensure that the generated data conforms to physical reality.

[0076] Specifically, a physical loss function L can be defined p , perform physical verification on each sample in the generated data; if a sample does not conform to these physical rules, the loss value will increase, otherwise it will decrease; such constraints can ensure that the generated data is more consistent with the actual tunnel monitoring data. The physical loss function can be expressed as:

[0077]

[0078] Among them, C(x i ) is the physical constraint for each generated sample.

[0079] Step S33: The conditional generator G and the discriminator D perform multiple rounds of iterations through adversarial training. In each round of iteration, the generator continuously generates new synthetic data, and the discriminator continuously optimizes its discrimination ability and updates the model parameters until the generated data distribution is very close to the real data.

[0080] The specific meanings of the model parameters are:

[0081] Epochs: indicates the number of training iterations of the generator and discriminator. The performance of the generator and discriminator is optimized through multiple iterations.

[0082] Batch size (batch_size): The number of samples randomly drawn from the dataset at each training time for one round of training process;

[0083] The learning rate of the generator and discriminator: controls the step size of each parameter update of the generator and discriminator, affecting the speed of weight adjustment of the model;

[0084] The hidden layer dimensions of the generator and discriminator: define the size of each neural network layer in the generator and discriminator, specifically the number of neurons contained in each layer;

[0085] Number of generated samples: The total number of synthetic data samples generated based on the above parameters to supplement the existing monitoring dataset.

[0086] S4: Preliminary verification of the synthetic data, statistical analysis of the synthetic data, and k-means cluster analysis.

[0087] Specifically, S41: Statistical analysis of real data and synthetic data first calculates the key statistical indicators of each set of data, including minimum, maximum, mean, median, quartiles and standard deviation, to comprehensively evaluate the distribution characteristics of the two sets of data.

[0088] S42: Draw the distribution histogram of real data and synthetic data through visualization to intuitively show the similarities and differences in data distribution between the two.

[0089] S43: Perform k-means clustering analysis on the real data and the synthetic data, and cluster the data using a preset number of clusters.

[0090] Calculate the silhouette coefficient of the two sets of data. The silhouette coefficient evaluates the quality of clustering by measuring the closeness of data points in the same cluster and the separation from other clusters. The silhouette coefficient range is [-1,1]. The closer the value is to 1, the better the clustering effect of the data point. By calculating the silhouette coefficient, the rationality of the clustering effect of the synthetic data and its consistency with the real data are verified.

[0091] S5: Use real data and synthetic data to train and evaluate the support vector machine (SVM) model respectively, optimize the hyperparameters through grid search combined with cross-validation, evaluate the model performance, compare the model training effects of synthetic data and real data, and verify the effectiveness of synthetic data.

[0092] Specifically, S51: perform support vector machine (SVM) model training and hyperparameter optimization, combine grid search and cross validation, select the best model and parameter combination, and improve the generalization ability of the model, specifically:

[0093] (1) A support vector machine (SVM) model was constructed using real data and synthetic data. The hyperparameters of the SVM model were optimized using a grid search combined with a 10-fold cross-validation method. The hyperparameter combination included the penalty parameter C and the kernel function.

[0094] (2) The generalization ability of the model was evaluated through 10-fold cross-validation, and the optimal hyperparameter combination was selected to train the model on real data and synthetic datasets respectively.

[0095] S52: Real-time monitoring of data generation and model training performance, dynamic adjustment of data generation strategies and model parameters, to ensure efficient operation of the entire process. Through a closed-loop feedback mechanism, ensure that the data generation and model optimization process always maintain adaptability, reduce overfitting, and improve the generalization ability of the model in different data scenarios. Specifically:

[0096] (1) After training, the performance of the model is tested and evaluated using evaluation indicators such as accuracy, precision, recall, and F1 value, and the test results of real data and synthetic data are compared.

[0097] (2) By comparing the performance of real data and synthetic data in SVM model training and testing, the effectiveness and rationality of synthetic data are evaluated, and its role in improving model performance and solving data scarcity and imbalance problems is verified.

[0098] The purpose of this embodiment is to solve the problem of scarce and unbalanced tunnel health monitoring data and improve the model training effect and prediction performance of tunnel health assessment; first, the tunnel on-site disease monitoring data was collected, and comprehensive data preprocessing was carried out for outliers, missing values ​​and inconsistency problems in real data, including feature conversion, outlier detection and processing, missing value filling and normalization. On this basis, the conditional table generative adversarial network (CTGAN) was used for data enhancement to generate synthetic data similar to the real data distribution to supplement the number of samples and optimize the problem of uneven category distribution. Finally, the support vector machine (SVM) model was trained and evaluated using real data and synthetic data respectively, and the training and testing effects of the two were compared to verify the effectiveness of synthetic data in improving the accuracy of tunnel health assessment and solving the problem of data imbalance. This embodiment not only improves the accuracy and reliability of tunnel health assessment, but also effectively solves the problems of data scarcity and imbalance, and provides a scientific basis for the early prediction and operation and maintenance management of tunnel diseases.

[0099] Example 2

[0100] In one embodiment of the present disclosure, a tunnel disease monitoring data enhancement system based on CTGAN is provided, including an acquisition module, an extraction module, a synthesis module and a verification module:

[0101] The acquisition module is configured to: obtain the real data table of tunnel site diseases, and perform disease type labeling and disease feature preprocessing;

[0102] The extraction module is configured to: extract time series features and spatial location features from the preprocessed disease features;

[0103] The synthesis module is configured to: train the CTGAN using the time series features, spatial location features and disease types as conditional information, and use the trained CTGAN to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table;

[0104] The verification module is configured to: after generating a new sample data table, verify the effect of the generated synthetic data by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

[0105] Example 3

[0106] In one embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method for enhancing tunnel disease monitoring data based on CTGAN is implemented.

[0107] Example 4

[0108] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the tunnel disease monitoring data enhancement method based on CTGAN is implemented.

[0109] Example 5

[0110] In one embodiment of the present disclosure, an electronic device is provided, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the tunnel disease monitoring data enhancement method based on CTGAN.

[0111] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0112] 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.

[0113] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A tunnel disease monitoring data enhancement method based on CTGAN, characterized in that: include: Obtain the real data table of tunnel site diseases, and mark the disease types and pre-process the disease characteristics; Extracting time series features and spatial location features from the preprocessed disease features; The CTGAN is trained with time series features, spatial location features and disease types as conditional information, and the trained CTGAN is used to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table; After generating the new sample data table, the generated synthetic data effect is verified by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

2. The tunnel disease monitoring data enhancement method based on CTGAN according to claim 1, characterized in that: The real data table is composed of a number of real data; The disease type labeling is to label each piece of real data with the disease type; The disease characteristics are specific indicators or characteristics used to describe each disease type; The preprocessing includes data encoding, outlier detection and processing, missing value filling, and normalization.

3. The tunnel disease monitoring data enhancement method based on CTGAN according to claim 1, characterized in that: The time series feature is the trend information of the length, width and depth of the cracks changing with time; The spatial position feature is the specific position information in the tunnel structure.

4. The tunnel disease monitoring data enhancement method based on CTGAN according to claim 1, characterized in that: The CTGAN includes a conditional generator and a discriminator; The condition generator takes a small batch of random vectors generated by random noise as input and generates corresponding synthetic data in combination with the condition information; The discriminator takes real data and synthetic data as input to judge whether the data is true or false.

5. The tunnel disease monitoring data enhancement method based on CTGAN according to claim 1, characterized in that: The statistical method is specifically: Compare and analyze the statistical indicators of real data and synthetic data, including minimum, maximum, mean, quartile and standard deviation, to ensure that the statistical distribution of synthetic data is similar to that of real data; Through visualization, the distribution histograms of the two sets of data are drawn to further confirm their similarity; Perform k-means clustering analysis and verify the clustering effect of synthetic data by calculating the silhouette coefficient to ensure consistency with the real data.

6. The tunnel disease monitoring data enhancement method based on CTGAN according to claim 1, characterized in that: The support vector machine uses real data and synthetic data to train and evaluate the support vector machine model, compares the model training and test effects of the two, and verifies the effectiveness of the synthetic data.

7. A tunnel disease monitoring data enhancement system based on CTGAN, characterized in that: Including acquisition module, extraction module, synthesis module and verification module: The acquisition module is configured to: obtain the real data table of tunnel site diseases, and perform disease type labeling and disease feature preprocessing; The extraction module is configured to: extract time series features and spatial location features from the preprocessed disease features; The synthesis module is configured to: train the CTGAN using the time series features, spatial location features and disease types as conditional information, and use the trained CTGAN to generate synthetic data with a similar distribution to the preprocessed real data table to obtain a new sample data table; The verification module is configured to: after generating a new sample data table, verify the effect of the generated synthetic data by combining statistical methods and support vector machines to evaluate the similarity between the synthetic data and the real data.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for enhancing tunnel disease monitoring data based on CTGAN according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the tunnel disease monitoring data enhancement method based on CTGAN is implemented as described in any one of claims 1 to 6.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes a tunnel disease monitoring data enhancement method based on CTGAN as described in any one of claims 1 to 6.

Images