Road hidden disease area identification method and system based on ground penetrating radar signals

Abstract

The invention discloses a road hidden disease area identification method and system based on a ground penetrating radar signal, and the method comprises the steps: obtaining the original data of the ground penetrating radar signal, and carrying out the hidden defect marking; preprocessing the original data of the ground penetrating radar signal to obtain preprocessed data; extracting time domain features and frequency domain features of the preprocessed data; screening important features from the time domain features and the frequency domain features; the screened important features are utilized to train a classifier, and after training is completed, a road hidden defect area recognition model is obtained; and identifying to-be-identified ground penetrating radar signal data by using the road hidden defect area identification model, and determining a road hidden defect area identification result. Accurate identification of the hidden disease area of the road is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road intelligent detection, and in particular to a road hidden disease area identification method and system based on ground penetrating radar signals. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] With the rapid development of transportation infrastructure and the increasing traffic volume, the quality and safety of roads have become a widespread concern in society. During the service of roads, under the coupling effect of multiple factors such as vehicle load, climate environment, etc., especially heavy traffic, temperature change, rainfall erosion, etc., hidden defects may occur inside the road. If not detected and repaired in time, it will seriously affect the structural integrity of the road, shorten the service life of the road, and may even lead to serious traffic accidents. Therefore, it is of great practical significance to quickly and accurately identify the internal hidden defect area of the road.

[0004] Traditional road defect detection methods, such as visual inspection and core analysis, can usually only identify surface or shallow diseases, and are time-consuming and labor-intensive, with low efficiency, which is difficult to meet the needs of large-scale and rapid detection. With the development of ground penetrating radar technology, using high-frequency electromagnetic waves for non-destructive testing of road internal structure has gradually become an effective means. Ground penetrating radar can penetrate the road structure layer and obtain the reflection signal of the underground target object, thereby providing characteristic information of the internal defects of the road.

[0005] Since ground penetrating radar usually needs to scan multiple times and repeatedly in a large area to ensure coverage and detection accuracy, the amount of accumulated raw data is extremely large. Existing ground penetrating radar data processing methods, such as manual analysis and simple image processing algorithms, often have low recognition accuracy, strong subjectivity, and low efficiency, which makes it difficult to fully exploit the potential of ground penetrating radar technology in road hidden defect detection. SUMMARY

[0006] In order to solve the above problems, the present application proposes a road hidden disease area identification method and system based on ground penetrating radar signals, which realizes the rapid and accurate identification of the road hidden disease area.

[0007] To achieve the above purpose, the present application adopts the following technical solutions: In a first aspect, a road hidden disease area identification method based on ground penetrating radar signals is proposed, comprising: Obtaining ground penetrating radar signal raw data and labeling hidden defects; Pretreating the ground penetrating radar signal raw data to obtain pretreated data; Extract the time-domain and frequency-domain features of the preprocessed data; Select important features from time-domain and frequency-domain features; The classifier is trained using the selected key features. Once training is complete, a road hidden defect area identification model is obtained. The hidden defect area identification model of the road is used to identify the ground penetrating radar signal data to determine the identification results of the hidden defects area of ​​the road.

[0008] Further preprocessing of the raw ground-penetrating radar signal data includes interference suppression, time-frequency conversion, background removal, and range gain processing.

[0009] Furthermore, a recursive feature elimination algorithm with fusion cross-validation is adopted to select important features from time-domain and frequency-domain features.

[0010] Furthermore, a random forest classifier was chosen as the classifier.

[0011] Furthermore, during the training process, swarm intelligence optimization algorithms are used to optimize the hyperparameters of the classifier.

[0012] Furthermore, the swarm intelligence optimization algorithm is the whale optimization algorithm.

[0013] Secondly, a road hidden defect area identification system based on ground-penetrating radar signals is proposed, including: The data acquisition unit is used to acquire raw data of ground-penetrating radar signals and mark hidden defects. The preprocessing unit is used to preprocess the raw ground-penetrating radar signal data to obtain preprocessed data; The feature extraction unit is used to extract the time-domain and frequency-domain features of the preprocessed data. The feature filtering unit is used to filter important features from time-domain features and frequency-domain features; The model training unit is used to train the classifier using the selected key features. Once training is complete, a road hidden defect area identification model is obtained. The road hidden defect area identification unit is used to identify the ground-penetrating radar signal data to be identified using the road hidden defect area identification model, and to determine the identification result of the road hidden defect area.

[0014] Thirdly, a computer device is proposed, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the method for identifying hidden road defects based on ground-penetrating radar signals proposed in the first aspect.

[0015] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program is adapted to be loaded and executed by a processor to implement the method for identifying hidden disease area of road based on ground penetrating radar signal according to the first aspect.

[0016] In a fifth aspect, a computer program product is provided, which comprises a computer program, and the computer program is executed by a processor to implement the method for identifying hidden disease area of road based on ground penetrating radar signal according to the first aspect.

[0017] Compared with the prior art, the present application has the following beneficial effects: The method for identifying hidden disease area of road based on ground penetrating radar signal provided by the present application can automatically identify and locate the hidden defect area in the road by deeply analyzing the internal structure data of the road collected by the ground penetrating radar, extracting the time domain and frequency domain features, screening the important features from the time domain and frequency domain features, training the classifier by using the important features, obtaining the road hidden defect area identification model after the training is completed, and then using the road hidden defect area identification model, has the advantages of high identification precision, high automation degree and wide applicability, can effectively early warning the road disease in the early stage, thereby reducing the maintenance cost and improving the road safety. In addition, only the important features extracted from the time domain and frequency domain features are used for classifier training, which can avoid model overfitting, improve model training precision, improve model training precision, and finally improve the identification precision of the hidden defect area in the road.

[0018] The advantages of the additional aspects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be known by the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0019] The drawings accompanying the specification of this application form a part thereof, serve to provide further understanding of the application, and together with the description of the exemplary embodiments of the application and the explanation thereof serve to explain the application, and do not constitute improper limitations on the application.

[0020] Figure 1 The flowchart of the method for identifying hidden disease area of road based on ground penetrating radar signal disclosed by the embodiment is shown in the figure; Figure 2 The importance level diagram of the time domain feature and the frequency domain feature disclosed by the embodiment is shown in the figure; Figure 3 The visualization diagram of the clustering result of the normal signal and the hidden defect signal disclosed by the embodiment is shown in the figure; Figure 4 The iteration optimization traversal point diagram of the whale optimization algorithm disclosed by the embodiment is shown in the figure; Figure 5 An iteration optimal position graph of a parameter to be optimized of a random forest algorithm disclosed as an embodiment; Figure 6 A prediction result graph of a random forest algorithm disclosed as an embodiment; Figure 7 An important feature screening flowchart disclosed as an embodiment. DETAILED DESCRIPTION

[0021] The application will be further described below in conjunction with the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0023] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and furthermore, it should be understood that when the terms "comprise" and / or "include" are used in the specification, there is a feature, step, operation, device, component, and / or combination thereof.

[0024] Embodiment 1 In this embodiment, a road hidden defect area identification method based on ground penetrating radar signal is disclosed, as shown in Figures 1-6 , comprising: S1: Obtain ground penetrating radar signal raw data and label hidden defects.

[0025] In this embodiment, a ground coupling step frequency ground penetrating radar system is used to collect data from the road section to be tested, with the parameters set as bandwidth≥3GHz, time window≥83ns, and longitudinal sampling interval≤5cm. Hidden defect areas are marked during the collection process to obtain ground penetrating radar signal raw data for subsequent processing and identification.

[0026] S2: Preprocess the ground penetrating radar signal raw data to obtain preprocessed data.

[0027] The preprocessing of the ground penetrating radar signal raw data in this embodiment includes interference suppression, time-frequency conversion, background removal, and range gain processing. Specifically: The collected radar data is subjected to interference suppression, and the radar data is converted from frequency domain dimension to time domain dimension using inverse fast Fourier transform, with Kaiser window type, and background removal using sliding window averaging and mean value method in sequence, and gain compensation of radar signal data using exponential gain function.

[0028] After that, the processed data is exported as an n*m dimensional matrix, n is the sum of the number of normal waveform channels and the number of hidden defect waveform channels, m is the number of sampling points of each waveform data, wherein the number of normal waveform channels and the number of hidden defect waveform channels are each n / 2.

[0029] Preferably, n is 300, m is 232, 70% of the total number of waveform channels is used for the training set, and 30% is used for the test set, and the number of normal waveform channels and the number of hidden waveform channels in the training set are each 1 / 2.

[0030] S3: Extract the time domain features and frequency domain features of the preprocessed data to obtain an n*m dimensional matrix, n is the number of features, m is the number of features. l l After that, the extracted time domain features and frequency domain features are normalized.

[0031] Preferably, the number of features n is 16. l

[0032] The time domain features include mean value , standard deviation , skewness , kurtosis , maximum amplitude , minimum amplitude , root mean square , amplitude factor , waveform factor , impact factor , margin factor , energy . s i is the amplitude value of the radar signal sampling point, N is the number of sampling points of each radar signal.

[0033] The frequency domain features include center of gravity frequency , frequency root mean square , average frequency , frequency variance , f k is the frequency value of the kth spectrum, and S(k) is the amplitude value corresponding to the frequency value of the kth spectrum.

[0034] S4: Screen important features from the time domain features and frequency domain features.

[0035] Feature selection is performed on the normalized features obtained in S3 to remove irrelevant or redundant features and select important features that have a significant impact on the identification result.

[0036] ​​The embodiment adopts a recursive feature elimination algorithm with fusion cross-validation to screen important features from time domain features and frequency domain features.

[0037] The classifier of the embodiment selects a random forest classifier, and performs multiple rounds of training on the random forest classifier based on the time domain features and the frequency domain features obtained by S3. The weight of each feature is calculated after each round of training, and the weight can be reflected by the importance of the feature. According to the calculated weight of the feature, the lowest weight feature is screened out to obtain the screened feature. The next round of training is performed based on the screened feature, and the step is repeated to obtain the feature importance ranking of each feature. Then, based on the ranking, feature subsets of n (n = 1, 2, 3, …, 16) features are selected in turn for model training and cross-validation, and finally the feature subset with the highest accuracy is obtained to form an important feature set, as shown in Figure 2 Figure 2 The shaded column chart in the middle is the important feature set screened out.

[0038] Specifically, as shown in Figure 7 , all time domain features and frequency domain features are used to perform the first round of training on the classifier to obtain the weight of each feature. The features are screened according to the weight, and the remaining features are used to perform the second round of training on the classifier after the lowest weight features are screened out. The weight of each feature participating in the second round of training is obtained, and the features participating in the second round of training are screened according to the weight. The remaining features are used to perform the third round of training on the classifier after the lowest weight features are screened out. Until the weight of all time domain features and frequency domain features is obtained, and then the importance ranking of all time domain features and frequency domain features is obtained according to the weight. Based on the ranking, different numbers of features are selected from all time domain features and frequency domain features to form feature subsets. A plurality of feature subsets are obtained. The classifier is trained and cross-validated through all feature subsets to obtain the feature subset corresponding to the optimal performance of the trained classifier as the important feature set.

[0039] Advantages: By gradually removing features, features that contribute less to the model or even have a negative impact are removed, and the model performance is evaluated in combination with cross-validation to robustly screen the optimal feature subset, improve the generalization ability of the model, and reduce overfitting.

[0040] S5: Train the classifier using the screened important features, and obtain the road concealed defect area recognition model after the training is completed.

[0041] In addition to obtaining the road concealed defect area recognition result, the embodiment also maps the important features into a two-dimensional coordinate form to obtain a visual classification result.

[0042] ​Preferably, the important features can be mapped into two-dimensional coordinate form by using the t-SNE function in Matlab, or other functions can be used as long as they can map the important features into two-dimensional coordinate form.

[0043] The classifier of the embodiment selects a random forest classifier. The label value of the normal area radar signal is 1, and the label value of the hidden defect area radar signal is 2. The important features screened out by S4 are input into the random forest classifier, the random forest classifier is trained, and after the training is completed, a road hidden defect area recognition model based on ground penetrating radar signal is obtained.

[0044] In the training process of the random forest classifier, a swarm intelligence optimization algorithm is used to optimize the hyperparameters of the classifier.

[0045] Preferably, the swarm intelligence optimization algorithm is a whale optimization algorithm, and the optimized hyperparameters include the number of decision trees (ntree) and the number of randomly selected features for decision tree node splitting (mtry). The optimization range of ntree is 0-500, the optimization range of mtry is 0-12, the number of whales is 20, and the number of iterations is 200, as shown in Figure 4 The iteration optimization traversal point graph of the whale optimization algorithm is shown in Figure 5 The iteration optimal position graph of the random forest algorithm to be optimized parameters is shown in

[0046] The test set divided in S2 is identified by the road hidden defect area recognition model to verify the recognition effect of the road hidden defect area recognition model, as shown in Figure 6 The accuracy of the road hidden defect area recognition model obtained in the embodiment for recognizing road hidden defects is 95.5556%, which has a high accuracy.

[0047] S6: Use the road hidden defect area recognition model to identify the ground penetrating radar signal data to be identified and determine the road hidden defect area recognition result.

[0048] The process includes: preprocessing the ground penetrating radar signal data to be identified and extracting important features; inputting the extracted important features into the road hidden defect area recognition model to obtain the road hidden defect area recognition result, which is a hidden defect area or a normal area, and achieving accurate identification of the hidden defects of the road section to be tested.

[0049] The embodiment extracts time domain and frequency domain features by deeply analyzing the internal structure data of the road collected by the ground penetrating radar, uses a random forest as a classifier, establishes a road hidden defect area recognition model, can automatically identify and locate the hidden defect area in the road, has the advantages of high recognition accuracy, high automation degree and wide applicability, can effectively early warn the road disease in the early stage, thereby reducing the maintenance cost and improving the road safety; the recursive feature elimination algorithm using fusion cross validation can effectively remove redundant features, avoid overfitting of the model to the training data, make the model more general, and at the same time, the less number of features can reduce the data dimension and speed up the training speed of the model. In addition, the whale optimization algorithm is used to optimize the model parameters, simplify the structure of the model, reduce the complexity of the model, further speed up the training and learning rate of the model, and improve the prediction accuracy.

[0050] Embodiment 2 In this embodiment, a road hidden disease area recognition system based on ground penetrating radar signals is disclosed, comprising: A data acquisition unit is configured to acquire ground penetrating radar signal raw data and perform hidden defect labeling. A preprocessing unit is configured to preprocess the ground penetrating radar signal raw data to obtain preprocessed data. A feature extraction unit is configured to extract time domain features and frequency domain features of the preprocessed data. A feature screening unit is configured to screen important features from the time domain features and the frequency domain features. A model training unit is configured to train a classifier using the screened important features, and obtain a road hidden defect area recognition model after the training is completed. A road hidden disease area recognition unit is configured to recognize the ground penetrating radar signal data to be recognized using the road hidden defect area recognition model, and determine a road hidden disease area recognition result.

[0051] The application also discloses a computer device, which comprises: A processor is adapted to execute a computer program. A computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the road hidden disease area recognition method based on ground penetrating radar signals disclosed in embodiment 1.

[0052] The application also discloses a computer readable storage medium storing a computer program, and the computer program is adapted to be loaded and executed by the processor to implement the road hidden disease area recognition method based on ground penetrating radar signals disclosed in embodiment 1.

[0053] The application further discloses a computer program product, which comprises a computer program, and the computer program, when executed by a processor, realizes the method for identifying a hidden disease area of a road based on a ground penetrating radar signal disclosed in Embodiment 1.

[0054] The method disclosed in Embodiment 1 can be directly embodied by a hardware processor to be executed, or be executed by a combination of hardware and software modules in the processor. The software modules can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register or other mature storage media in the art. The storage media is located in a memory, and the processor reads information in the memory to combine the hardware to complete the steps of the above method. To avoid repetition, no longer detailed description is made herein.

[0055] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solutions. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0056] Although the specific embodiments of the application are described above in combination with the drawings, the description is not a limitation on the protection scope of the application, and those skilled in the art should understand that various modifications or changes made by those skilled in the art on the basis of the technical solutions of the application without creative labor are still within the protection scope of the application.

Claims

1. A method for identifying hidden road defects based on ground-penetrating radar signals, characterized in that, include: Acquire raw ground-penetrating radar signal data and mark hidden defects; The raw ground-penetrating radar signal data is preprocessed to obtain preprocessed data; Extract the time-domain and frequency-domain features of the preprocessed data; Select important features from time-domain and frequency-domain features; The classifier is trained using the selected key features. Once training is complete, a road hidden defect area identification model is obtained. The hidden defect area identification model of the road is used to identify the ground penetrating radar signal data to determine the identification results of the hidden defects area of ​​the road.

2. The method for identifying hidden road defects based on ground-penetrating radar signals as described in claim 1, characterized in that, Preprocessing of raw ground-penetrating radar signal data includes interference suppression, time-frequency conversion, background removal, and range gain processing.

3. The method for identifying hidden road defects based on ground-penetrating radar signals as described in claim 1, characterized in that, A recursive feature elimination algorithm with fusion cross-validation is used to select important features from time-domain and frequency-domain features.

4. The method for identifying hidden road defects based on ground-penetrating radar signals as described in claim 1, characterized in that, The random forest classifier was chosen.

5. The method for identifying hidden road defects based on ground-penetrating radar signals as described in claim 1, characterized in that, During training, swarm intelligence optimization algorithms are used to optimize the hyperparameters of the classifier.

6. The method for identifying hidden road defects based on ground-penetrating radar signals as described in claim 5, characterized in that, The swarm intelligence optimization algorithm is the whale optimization algorithm.

7. A road hidden defect area identification system based on ground-penetrating radar signals, characterized in that, include: The data acquisition unit is used to acquire raw data of ground-penetrating radar signals and mark hidden defects. The preprocessing unit is used to preprocess the raw ground-penetrating radar signal data to obtain preprocessed data; The feature extraction unit is used to extract the time-domain and frequency-domain features of the preprocessed data. The feature filtering unit is used to filter important features from time-domain features and frequency-domain features; The model training unit is used to train the classifier using the selected key features. Once training is complete, a road hidden defect area identification model is obtained. The road hidden defect area identification unit is used to identify the ground-penetrating radar signal data to be identified using the road hidden defect area identification model, and to determine the identification result of the road hidden defect area.

8. An electronic device, characterized in that, The device includes: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the method for identifying hidden road defects based on ground-penetrating radar signals as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed by the method for identifying hidden road defects based on ground-penetrating radar signals as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method for identifying hidden road defects based on ground-penetrating radar signals as described in any one of claims 1-6.

Images