A bridge crack detection and identification method and system based on unsupervised learning and big data
Through unsupervised learning and big data technology, automated detection of bridge cracks has solved the problems of low efficiency and high cost of traditional methods, improved bridge safety, and promoted intelligent management and digital transformation.
Patent Information
- Application Number
- CN202410978809.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-22
AI Technical Summary
The traditional bridge crack detection method has problems such as high labor cost, low efficiency, and susceptible to subjective factors, and there is no effective solution yet.
Using an unsupervised learning and big data method, data preprocessing and clustering analysis are performed by obtaining bridge structure data, using principal component analysis, density clustering algorithm, convolutional neural network, isolated forest anomaly detection algorithm and machine learning algorithm. , extract and classify crack features, build a healthy model of bridge structure, and predict crack development trends.
It realizes automated inspection, reduces maintenance costs, improves bridge structure safety, and promotes the digital transformation of intelligent bridge management and transportation infrastructure.
Smart Images

Figure CN118536072B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of bridge crack detection, and in particular to a bridge crack detection and identification method and system based on unsupervised learning and big data. Background Art
[0002] Bridges are a vital component of modern transportation infrastructure, and the problem of bridge cracks is a key challenge in bridge structural health monitoring. The presence of cracks may lead to a decrease in the strength and stability of the bridge structure, thereby endangering the safety of the bridge. Therefore, it is particularly important to accurately and timely detect and identify bridge cracks.
[0003] Traditional crack detection methods usually rely on manual inspections or the use of equipment such as sensors. These methods have problems such as high labor costs, low efficiency, and susceptibility to subjective factors.
[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention
[0005] The embodiments of the present application provide a bridge crack detection and identification method and system based on unsupervised learning and big data to solve the above-mentioned technical problems.
[0006] The present application provides a bridge crack detection and identification method based on unsupervised learning and big data, comprising: acquiring collected structural data of a bridge and preprocessing the structural data; the structural data includes image data and sensor data; preprocessing the structural data using a principal component analysis dimensionality reduction algorithm to retain key information in the structural data; clustering analysis of the preprocessed structural data using a density-based clustering algorithm to determine temporal implicit features in the structural data; locating and segmenting the crack area of the bridge using a convolutional neural network for the image data in the preprocessed structural data; and extracting crack features of the segmented crack area; wherein the crack features include shape features and texture features; based on the crack features of the segmented crack area and the temporal implicit features in the structural data, classifying the segmented crack area using an isolated forest anomaly detection algorithm to obtain crack area analysis results; constructing a health assessment model using a machine learning algorithm, and training a bridge structure health model according to the crack area analysis results; wherein the bridge structure health model is used to perform health assessment on the bridge; and predicting the crack development trend of the bridge based on the health assessment results of the trained bridge structure health model and the big data analysis results.
[0007] Furthermore, the clustering analysis of the preprocessed structural data using a density-based clustering algorithm to determine the temporal implicit features in the structural data includes:
[0008] Clustering the structural data using the density-based clustering algorithm to form measurement clusters;
[0009] Each measurement in the measurement cluster is processed based on a distance calculation formula to obtain a clustering parameter of each measurement; wherein the distance calculation formula is:
[0010] ,
[0011] Among them, the object is a vector in n-dimensional space; Represents the position of object X in the i-th dimension space; object is a vector in the n-dimensional space; Represents the position of object Y in the i-th dimension space; is the distance between object X and object Y;
[0012] A split tree is constructed based on the clustering parameters of each measurement; and feature coding and reconstruction are performed on the leaf node position vectors in the split tree to obtain the temporal implicit features in the structural data.
[0013] Furthermore, based on the crack features of the segmented crack area and the temporal implicit features in the structural data, the segmented crack area is classified using an isolation forest anomaly detection algorithm to obtain a crack area analysis result, including:
[0014] Constructing an isolation forest, comprising: setting the isolation forest to include m isolated trees; randomly selecting b samples from the crack features of the crack area and the temporal implicit features as root nodes of the isolation trees;
[0015] Determine the crack characteristics of the crack area and the path length of the temporal implicit feature in the jth isolated tree in the isolation forest ;in, ; represents the number of edges that the crack feature and the temporal implicit feature pass through from the root node to the leaf node of the j-th isolated tree; represents the average path length from the root node to the leaf node of the j-th isolated tree;
[0016] Based on the path length The segmented crack regions are classified to obtain analysis results of the crack regions.
[0017] Furthermore, the health assessment results of the bridge structure health model obtained through training and the big data analysis results predict the crack development trend of the bridge, including:
[0018] Acquire bridge crack data from a big data platform, and analyze to obtain the big data analysis results;
[0019] Decomposing the health assessment results and the big data analysis results by using a matrix eigenvalue decomposition algorithm to obtain a reduced-dimensional health assessment data set and a reduced-dimensional big data analysis data set;
[0020] Based on the health assessment data set after dimensionality reduction and the big data analysis data set after dimensionality reduction, a time decomposition model is used to predict the crack development trend of the bridge; wherein the time decomposition model includes trend decomposition displacement, random decomposition displacement and fluctuation decomposition displacement;
[0021] The trend decomposition displacement The prediction is based on the following formula:
[0022]
[0023] in, is the t-order trend decomposition displacement; is the t-1 order trend decomposition displacement; B is the lag operator; is a random decomposition displacement sequence; a is the sequence mean; is the autoregression coefficient; is the moving average coefficient.
[0024] Furthermore, the crack area analysis results, the health assessment results and the crack development trend of the bridge are visualized.
[0025] Furthermore, the image data included in the structural data include structural crack image data, non-structural crack image data and no-crack image data; wherein the structural crack image data include transverse crack image data and longitudinal crack image data; and the non-structural crack image data include fragmented crack image data.
[0026] Furthermore, before preprocessing the structural data using the principal component analysis dimensionality reduction algorithm, the method further includes:
[0027] The image data included in the structure data is subjected to random brightness distortion transformation, random contrast distortion transformation and random saturation distortion transformation.
[0028] The present application also provides a bridge crack detection and identification system based on unsupervised learning and big data, including a data collection and preprocessing unit, a data clustering analysis unit, an image segmentation and feature extraction unit, an anomaly detection and crack identification unit, a bridge health assessment unit, and a crack development trend prediction unit, wherein:
[0029] The data collection and preprocessing unit is used to obtain the collected structural data of the bridge and preprocess the structural data; the structural data includes image data and sensor data; the principal component analysis dimensionality reduction algorithm is used to preprocess the structural data to retain key information in the structural data;
[0030] The data clustering analysis unit is used to perform clustering analysis on the preprocessed structure data using a density-based clustering algorithm to determine the temporal implicit features in the structure data;
[0031] The image segmentation and feature extraction unit is used for image segmentation and feature extraction, and for locating and segmenting the crack area of the bridge using a convolutional neural network for the image data in the preprocessed structural data; and extracting crack features of the segmented crack area; wherein the crack features include shape features and texture features;
[0032] The anomaly detection and crack identification unit is used to classify the segmented crack area based on the crack characteristics of the segmented crack area and the temporal implicit characteristics in the structural data using an isolated forest anomaly detection algorithm to obtain a crack area analysis result;
[0033] The bridge health assessment unit is used to construct a health assessment model through a machine learning algorithm, and train the bridge structure health model according to the crack area analysis result; wherein the bridge structure health model is used to perform health assessment on the bridge;
[0034] The crack development trend prediction unit is used to predict the crack development trend of the bridge based on the health assessment results of the bridge structure health model obtained through training and the big data analysis results.
[0035] Furthermore, the bridge crack detection and identification system based on unsupervised learning and big data also includes:
[0036] A visualization display unit is used to visualize the crack area analysis results, the health assessment results and the crack development trend of the bridge.
[0037] Furthermore, the bridge crack detection and identification system based on unsupervised learning and big data also includes:
[0038] The image distortion transformation unit is used to perform random brightness distortion transformation, random contrast distortion transformation and random saturation distortion transformation on the image data included in the structural data before preprocessing the structural data using the principal component analysis dimensionality reduction algorithm.
[0039] Based on the embodiments provided by the present application, in combination with a data collection and preprocessing unit, a data clustering analysis unit, an image segmentation and feature extraction unit, an anomaly detection and crack identification unit, a bridge health assessment unit, and a crack development trend prediction unit. The safety of the bridge structure can be improved, that is, by effectively detecting and identifying bridge cracks, potential safety hazards can be discovered in a timely manner, and then corresponding repair measures can be taken to improve the safety of the bridge structure; the maintenance cost can be reduced, that is, the traditional bridge crack detection method requires a lot of manpower and material resources, while the method based on unsupervised learning and big data can realize automatic detection and reduce maintenance costs; intelligent bridge management can be promoted. Specifically, the introduction of unsupervised learning into the field of bridge crack detection can provide technical support for intelligent bridge management and promote the digital transformation of transportation infrastructure. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0041] Figure 1 is a flowchart of an optional bridge crack detection and identification method based on unsupervised learning and big data according to an embodiment of the present application;
[0042] Figure 2 is a structural diagram of an optional bridge crack detection and identification system based on unsupervised learning and big data according to an embodiment of the present application;
[0043] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0044] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.
[0045] Alternatively, if Figure 1 As shown, the present application provides a bridge crack detection and identification method based on unsupervised learning and big data, comprising:
[0046] S101, acquiring the collected structural data of the bridge and preprocessing the structural data; the structural data includes image data and sensor data;
[0047] S102, preprocessing the structural data using a principal component analysis dimensionality reduction algorithm to retain key information in the structural data;
[0048] S103, performing cluster analysis on the preprocessed structural data using a density-based clustering algorithm to determine temporal implicit features in the structural data;
[0049] In some embodiments of the present application, methods such as K-means clustering and density clustering may be used to classify data into different categories and find the spatial distribution pattern of cracks.
[0050] S104, using a convolutional neural network to locate and segment the crack area of the bridge for the image data in the preprocessed structural data; and extracting crack features of the segmented crack area; wherein the crack features include shape features and texture features;
[0051] S105, based on the crack characteristics of the segmented crack area and the time series implicit characteristics in the structural data, the segmented crack area is classified using an isolation forest anomaly detection algorithm to obtain a crack area analysis result;
[0052] S106, constructing a health assessment model through a machine learning algorithm, and training a bridge structure health model according to the crack area analysis results; wherein the bridge structure health model is used to perform health assessment on the bridge;
[0053] S107, predicting the crack development trend of the bridge based on the health assessment results of the trained bridge structure health model and the big data analysis results.
[0054] Based on the embodiments provided in the present application, the collected structural data of the bridge is obtained, and the structural data is preprocessed. The preprocessed structural data is clustered and analyzed using a density-based clustering algorithm to determine the temporal implicit features in the structural data. For the image data in the preprocessed structural data, a convolutional neural network is used to locate and segment the crack area of the bridge; and the crack features of the segmented crack area are extracted; based on the crack features of the segmented crack area and the temporal implicit features in the structural data, the segmented crack area is classified using an isolated forest anomaly detection algorithm to obtain crack area analysis results; a health assessment model is constructed using a machine learning algorithm, and a bridge structure health model is trained according to the crack area analysis results; based on the health assessment results of the trained bridge structure health model and the big data analysis results, the crack development trend of the bridge is predicted. It can improve the safety of bridge structures, that is, by effectively detecting and identifying bridge cracks, potential safety hazards can be discovered in a timely manner, and then corresponding repair measures can be taken to improve the safety of bridge structures; it can reduce maintenance costs, that is, traditional bridge crack detection methods require a lot of manpower and material resources, while methods based on unsupervised learning and big data can realize automated detection and reduce maintenance costs; it can promote intelligent bridge management. Specifically, introducing unsupervised learning into the field of bridge crack detection can provide technical support for intelligent bridge management and promote the digital transformation of transportation infrastructure.
[0055] Furthermore, a density-based clustering algorithm is used to perform cluster analysis on the preprocessed structural data to determine the temporal implicit features in the structural data, including:
[0056] Cluster the structural data using density-based clustering algorithms to form measurement clusters;
[0057] Each measurement in the measurement cluster is processed based on the distance calculation formula to obtain the clustering parameters of each measurement; wherein the distance calculation formula is:
[0058]
[0059] Among them, the object is a vector in n-dimensional space; Represents the position of object X in the i-th dimension space; object is a vector in n-dimensional space; Represents the position of object Y in the i-th dimension space; is the distance between object X and object Y;
[0060] A split tree is constructed based on the clustering parameters of each measurement; the leaf node position vectors in the split tree are feature encoded and reconstructed to obtain the temporal implicit features in the structural data.
[0061] Furthermore, based on the crack characteristics of the segmented crack area and the temporal implicit features in the structural data, the segmented crack area is classified using the isolation forest anomaly detection algorithm to obtain the crack area analysis results, including:
[0062] Constructing an isolation forest, including: setting an isolation forest including m isolation trees; randomly selecting b samples from crack features and temporal implicit features of the crack region as root nodes of the isolation tree;
[0063] Determine the crack characteristics of the crack area and the path length of the temporal implicit characteristics in the jth isolated tree in the isolation forest ;in, ; Represents the number of edges that the crack features and temporal implicit features pass through from the root node to the leaf node of the jth isolated tree; Represents the average path length from the root node to the leaf node of the jth isolated tree;
[0064] Based on path length The segmented crack areas are classified to obtain crack area analysis results.
[0065] Furthermore, based on the health assessment results of the trained bridge structure health model and the big data analysis results, the crack development trend of the bridge is predicted, including:
[0066] Obtain bridge crack data from the big data platform and analyze it to obtain big data analysis results;
[0067] The health assessment results and big data analysis results are decomposed using the matrix eigenvalue decomposition algorithm to obtain a reduced-dimensional health assessment data set and a reduced-dimensional big data analysis data set;
[0068] Based on the reduced-dimensionality health assessment dataset and the reduced-dimensionality big data analysis dataset, the time decomposition model is used to predict the crack development trend of the bridge; the time decomposition model includes trend decomposition displacement, random decomposition displacement and fluctuation decomposition displacement;
[0069] Trend Decomposition Shift The prediction is based on the following formula:
[0070]
[0071] in, is the t-order trend decomposition displacement; is the t-1 order trend decomposition displacement; B is the lag operator; is a random decomposition displacement sequence; a is the sequence mean; is the autoregression coefficient; is the moving average coefficient.
[0072] Furthermore, the crack area analysis results, health assessment results, and crack development trends of the bridge are visualized.
[0073] The crack area analysis results, health assessment results and crack development trends of bridges are provided to engineers and managers for reference. Based on the big data analysis results, decision making is assisted to achieve intelligent and precise bridge maintenance management.
[0074] Furthermore, the image data included in the structural data include structural crack image data, non-structural crack image data and no-crack image data; wherein the structural crack image data include transverse crack image data and longitudinal crack image data; and the non-structural crack image data include fragmented crack image data.
[0075] Furthermore, before using the principal component analysis dimensionality reduction algorithm to preprocess the structural data, the method further includes:
[0076] The image data included in the structure data is subjected to a random brightness distortion transformation, a random contrast distortion transformation, and a random saturation distortion transformation.
[0077] Alternatively, if Figure 2 As shown, the present application provides a bridge crack detection and identification system based on unsupervised learning and big data, including a data collection and preprocessing unit 201, a data clustering analysis unit 202, an image segmentation and feature extraction unit 203, an anomaly detection and crack identification unit 204, a bridge health assessment unit 205 and a crack development trend prediction unit 206, wherein:
[0078] The data collection and preprocessing unit 201 is used to obtain the collected structural data of the bridge and preprocess the structural data; the structural data includes image data and sensor data; the principal component analysis dimensionality reduction algorithm is used to preprocess the structural data to retain key information in the structural data;
[0079] The data clustering analysis unit 202 is used to perform clustering analysis on the pre-processed structure data using a density-based clustering algorithm to determine the temporal implicit features in the structure data;
[0080] The image segmentation and feature extraction unit 203 is used for image segmentation and feature extraction. The convolutional neural network is used to locate and segment the crack area of the bridge for the image data in the preprocessed structural data; and to extract the crack features of the segmented crack area; wherein the crack features include shape features and texture features;
[0081] The anomaly detection and crack identification unit 204 is used to classify the segmented crack regions based on the crack characteristics of the segmented crack regions and the temporal implicit characteristics in the structural data using an isolation forest anomaly detection algorithm to obtain crack region analysis results;
[0082] The bridge health assessment unit 205 is used to construct a health assessment model through a machine learning algorithm, and train a bridge structure health model according to the crack area analysis results; wherein the bridge structure health model is used to perform health assessment on the bridge;
[0083] The crack development trend prediction unit 206 is used to predict the crack development trend of the bridge based on the health assessment results of the trained bridge structure health model and the big data analysis results.
[0084] Based on the embodiments provided by the present application, in combination with a data collection and preprocessing unit, a data clustering analysis unit, an image segmentation and feature extraction unit, an anomaly detection and crack identification unit, a bridge health assessment unit, and a crack development trend prediction unit. The safety of the bridge structure can be improved, that is, by effectively detecting and identifying bridge cracks, potential safety hazards can be discovered in a timely manner, and then corresponding repair measures can be taken to improve the safety of the bridge structure; the maintenance cost can be reduced, that is, the traditional bridge crack detection method requires a lot of manpower and material resources, while the method based on unsupervised learning and big data can realize automatic detection and reduce maintenance costs; intelligent bridge management can be promoted. Specifically, the introduction of unsupervised learning into the field of bridge crack detection can provide technical support for intelligent bridge management and promote the digital transformation of transportation infrastructure.
[0085] Furthermore, the bridge crack detection and recognition system based on unsupervised learning and big data also includes:
[0086] The visualization display unit is used to visualize the crack area analysis results, health assessment results and crack development trends of the bridge.
[0087] Furthermore, the bridge crack detection and recognition system based on unsupervised learning and big data also includes:
[0088] The image distortion transformation unit is used to perform random brightness distortion transformation, random contrast distortion transformation and random saturation distortion transformation on the image data included in the structural data before using the principal component analysis dimension reduction algorithm to preprocess the structural data.
[0089] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A bridge crack detection and identification method based on unsupervised learning and big data, characterized in that: include: Acquire the collected structural data of the bridge and pre-process the structural data; the structural data includes image data and sensor data; Preprocessing the structural data using a principal component analysis dimensionality reduction algorithm to retain key information in the structural data; Performing cluster analysis on the preprocessed structural data using a density-based clustering algorithm to determine temporal implicit features in the structural data; For the image data in the preprocessed structural data, a convolutional neural network is used to locate and segment the crack area of the bridge; and crack features of the segmented crack area are extracted; wherein the crack features include shape features and texture features; Based on the crack features of the segmented crack area and the temporal implicit features in the structural data, the segmented crack area is classified using an isolated forest anomaly detection algorithm to obtain a crack area analysis result; A health assessment model is constructed by a machine learning algorithm, and a bridge structure health model is trained according to the crack area analysis results; wherein the bridge structure health model is used to perform health assessment on the bridge; Based on the health assessment results of the bridge structure health model obtained through training and the big data analysis results, predict the crack development trend of the bridge; The method of performing cluster analysis on the preprocessed structural data using a density-based clustering algorithm to determine the temporal implicit features in the structural data includes: Clustering the structural data using the density-based clustering algorithm to form measurement clusters; Each measurement in the measurement cluster is processed based on a distance calculation formula to obtain a clustering parameter of each measurement; wherein the distance calculation formula is: , Among them, object X=( ,… ) is a vector in n-dimensional space; Represents the position of object X in the i-th dimension space; object Y=( ,… ) is a vector in the n-dimensional space; Represents the position of object Y in the i-th dimension space; is the distance between object X and object Y; Constructing a split tree based on the clustering parameters of each measurement; performing feature encoding reconstruction on the leaf node position vectors in the split tree to obtain the temporal implicit features in the structural data; Based on the crack features of the segmented crack area and the temporal implicit features in the structural data, the segmented crack area is classified using an isolated forest anomaly detection algorithm to obtain a crack area analysis result, including: Constructing an isolation forest, comprising: setting the isolation forest to include m isolated trees; randomly selecting b samples from the crack features of the crack area and the temporal implicit features as root nodes of the isolation trees; Determine the crack characteristics of the crack area and the path length of the temporal implicit feature in the jth isolated tree in the isolation forest ; in, + ; represents the number of edges that the crack feature and the temporal implicit feature pass through from the root node to the leaf node of the j-th isolated tree; represents the average path length from the root node to the leaf node of the j-th isolated tree; Based on the path length The segmented crack regions are classified to obtain analysis results of the crack regions.
2. The bridge crack detection and identification method based on unsupervised learning and big data according to claim 1 is characterized in that: The health assessment result of the bridge structure health model obtained through training and the big data analysis result are used to predict the crack development trend of the bridge, including: Acquire bridge crack data from a big data platform, and analyze to obtain the big data analysis results; Decomposing the health assessment results and the big data analysis results by using a matrix eigenvalue decomposition algorithm to obtain a reduced-dimensional health assessment data set and a reduced-dimensional big data analysis data set; Based on the health assessment data set after dimensionality reduction and the big data analysis data set after dimensionality reduction, a time decomposition model is used to predict the crack development trend of the bridge; wherein the time decomposition model includes trend decomposition displacement, random decomposition displacement and fluctuation decomposition displacement; The trend decomposition displacement The prediction is based on the following formula: , in, is the t-order trend decomposition displacement; is the t-1 order trend decomposition displacement; B is the lag operator; is a random decomposition displacement sequence; a is the sequence mean; is the autoregression coefficient; is the moving average coefficient.
3. The bridge crack detection and identification method based on unsupervised learning and big data according to claim 2 is characterized in that: The crack area analysis results, the health assessment results and the crack development trend of the bridge are visualized.
4. The bridge crack detection and identification method based on unsupervised learning and big data according to claim 1 is characterized in that: The image data included in the structural data include structural crack image data, non-structural crack image data and no-crack image data; wherein the structural crack image data include transverse crack image data and longitudinal crack image data; and the non-structural crack image data include fragmented crack image data.
5. The bridge crack detection and identification method based on unsupervised learning and big data according to claim 1 is characterized in that: Before preprocessing the structural data using the principal component analysis dimensionality reduction algorithm, the method further includes: The image data included in the structure data is subjected to random brightness distortion transformation, random contrast distortion transformation and random saturation distortion transformation.
6. A bridge crack detection and identification system based on unsupervised learning and big data, the system implements the method according to claim 1, characterized in that: It includes data collection and preprocessing unit, data clustering analysis unit, image segmentation and feature extraction unit, anomaly detection and crack identification unit, bridge health assessment unit and crack development trend prediction unit, among which, The data collection and preprocessing unit is used to obtain the collected structural data of the bridge and preprocess the structural data; the structural data includes image data and sensor data; the principal component analysis dimensionality reduction algorithm is used to preprocess the structural data to retain key information in the structural data; The data clustering analysis unit is used to perform clustering analysis on the preprocessed structure data using a density-based clustering algorithm to determine the temporal implicit features in the structure data; The image segmentation and feature extraction unit is used for image segmentation and feature extraction, and for locating and segmenting the crack area of the bridge using a convolutional neural network for the image data in the preprocessed structural data; and extracting crack features of the segmented crack area; wherein the crack features include shape features and texture features; The anomaly detection and crack identification unit is used to classify the segmented crack area based on the crack characteristics of the segmented crack area and the temporal implicit characteristics in the structural data using an isolated forest anomaly detection algorithm to obtain a crack area analysis result; The bridge health assessment unit is used to construct a health assessment model through a machine learning algorithm, and train a bridge structure health model according to the crack area analysis result; wherein the bridge structure health model is used to perform health assessment on the bridge; The crack development trend prediction unit is used to predict the crack development trend of the bridge based on the health assessment results of the bridge structure health model obtained through training and the big data analysis results.
7. The bridge crack detection and identification system based on unsupervised learning and big data according to claim 6 is characterized in that: The bridge crack detection and identification system based on unsupervised learning and big data also includes: A visualization display unit is used to visualize the crack area analysis results, the health assessment results and the crack development trend of the bridge.
8. The bridge crack detection and identification system based on unsupervised learning and big data according to claim 7 is characterized in that: The bridge crack detection and identification system based on unsupervised learning and big data also includes: The image distortion transformation unit is used to perform random brightness distortion transformation, random contrast distortion transformation and random saturation distortion transformation on the image data included in the structural data before preprocessing the structural data using the principal component analysis dimensionality reduction algorithm.
