Non-contact intelligent detection method for working state of energy dissipater of slope flexible protection system

Through the significance target detection deep neural network and morphological image processing technology, the problem of poor accuracy and safety risks of the energy consumption detection of the flexible protection system is solved, and the contactless intelligent detection of the working state of the energy consumption is realized, which improves the detection accuracy and security have promoted the intelligent process of the protection industry.

CN119048750AActive Publication Date: 2024-11-29SOUTHWEST JIAOTONG UNIV

Patent Information

Application Number
CN202411042065.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-29
Estimated Expiration
2044-07-31

AI Technical Summary

Technical Problem

In the prior art, the working status detection of the energy consumption device of the flexible protection system relies on manual inspection, which poses poor accuracy and safety risks. Especially in dangerous areas such as mountains and steep cliffs, it is difficult to achieve high-precision contactless detection.

Method used

The deep neural network with significance target detection combined with morphological image processing technology is used to establish the image data set of the energy-consuming operation process through simulated static tensile test, foot-size impact test and field engineering actual measurement, and train the model to realize the working state of the energy-consuming operation. Contactless intelligent detection, automatically segment the binary graph and calculate the remaining energy consumption capacity.

Benefits of technology

It realizes high-precision contactless detection of the working state of the energy-consuming device, reduces the safety risks and errors of manual detection, improves the accuracy of post-disaster assessment, and promotes the intelligence and automation process of the protection industry.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048750A_ABST
    Figure CN119048750A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of slope geological disaster protection, and particularly discloses a slope flexible protection system energy dissipater working state non-contact intelligent detection method which comprises the steps that an energy dissipater quasi-static tensile test, a full-scale impact test and field engineering actual measurement are carried out, and an energy dissipater working process image data set is established; training a saliency target detection deep neural network by adopting a data set to realize automatic segmentation of a binary image of the working state of the energy dissipater; a morphological image processing technology is adopted to obtain an energy dissipater skeleton, and a contour detection technology is adopted to extract the inner and outer contours of the energy dissipater skeleton to automatically calculate the residual energy dissipation capacity. According to the method, the problem of non-contact intelligent recognition of the working state of the key energy consumption component of the flexible protection system is solved through the artificial intelligence technology, the accuracy is high, and integrated application of appearance visual monitoring and internal performance judgment of the key component of the protection engineering is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of slope geological disaster protection, and in particular to a contactless intelligent detection method for the working state of an energy absorber of a slope flexible protection system. Background Art

[0002] Flexible protection systems are widely used in the protection of strong impact natural disasters such as landslides, debris flows, sand flows, and snow flows. The system often faces multiple impacts during operation and requires regular maintenance. At present, it mainly relies on manual inspections to determine the system's post-disaster residual protection capacity. When the system is working, it mainly consumes impact energy through energy absorbers. The system's residual protection capacity is closely related to the working status of the energy absorbers. One of the most critical tasks of manual inspections is to determine the working status of the energy absorbers and evaluate the remaining energy consumption capacity of the energy absorbers, so as to determine whether the structure needs to be repaired, strengthened, or replaced. Due to the highly nonlinear characteristics of the energy absorbers, the inspectors often have large errors in evaluating their remaining energy consumption capacity by manually measuring the stretching amount of the energy absorbers. At the same time, the working environment of the flexible protection system is mostly dangerous areas such as high mountains and steep cliffs, and it is very dangerous to collect data at close range after a disaster. Therefore, a high-precision non-contact detection method is urgently needed.

[0003] The salient target detection technology is a branch of the target detection field, which is mainly used to segment the target from the background and output a binary image, that is, the target pixel value is 1 and the background pixel value is 0. Morphological image processing technology is a branch of the computer vision field, which is mainly used to process binary images to obtain features. The trained salient target detection deep neural network can automatically segment the binary image of the energy consumer from the image, and further use morphological image processing operations such as filtering, medial axis transformation, opening operation and contour detection to obtain the inner and outer contours of the binary image skeleton of the energy consumer. The remaining energy consumption capacity of the energy consumer can be automatically calculated through the inner and outer contours of the binary image skeleton of the energy consumer, and the working status of the energy consumer can be accurately and contactlessly intelligently detected. It is of great significance to improve the accuracy of disaster damage assessment of flexible protection systems and reduce safety risks. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a contactless intelligent detection method for the working status of the energy absorber of the slope flexible protection system, which can realize accurate, contactless and intelligent detection of the working status of the energy absorber, thereby reducing the indirect losses caused by the slow emergency rescue response of the flexible protection structure, and solving the problems of poor accuracy and high risk of the existing manual detection methods mentioned in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a non-contact intelligent detection method for the working state of an energy dissipator of a slope flexible protection system, comprising the following steps:

[0006] S1. Conduct simulated static tensile test, full-scale impact test and field engineering measurement of energy absorber to establish image data set of working process of energy absorber;

[0007] S2, using the data set to train a deep neural network for salient target detection to achieve automatic segmentation of the binary image of the working state of the energy consumer;

[0008] S3. Use morphological image processing technology to obtain the skeleton of the energy absorber and use contour detection technology to extract the inner and outer contours of the skeleton of the energy absorber to automatically calculate the remaining energy consumption capacity.

[0009] Preferably, in step S1, the following is specifically included:

[0010] S11. Conduct a simulated static tensile test on the energy absorber to obtain a video of the entire tensile deformation process, and extract key frames to obtain images of the entire working process;

[0011] S12. Carry out full-scale impact tests on protective structures and field engineering measurements to obtain images of the energy absorber's status before and after working with multiple backgrounds;

[0012] S13. The images of energy absorbers obtained from the quasi-static tensile test, full-scale impact test and field engineering measurements are annotated with salient target detection labels, and data is augmented and divided into training and validation sets.

[0013] Preferably, in order to obtain sufficient working images of the energy absorber, images are collected by conducting quasi-static tensile tests on the energy absorber, full-scale impact tests on the protective structure, field engineering measurements, etc. The test methods, loading equipment, loading speed, etc. should comply with relevant specifications.

[0014] Preferably, in order to improve the generalization performance of the deep neural network model for salient target detection, stretched and deformed images of the energy consumer during the entire working process and multi-background images covering daily usage scenarios should be obtained as much as possible.

[0015] Preferably, in step S2, the following is specifically included:

[0016] S21. Determine the structure of the deep neural network for salient object detection and write code;

[0017] S22, using the training set to train the deep neural network for salient object detection and using the validation set for testing;

[0018] S23, monitoring the training error, verifying the error and adjusting the network structure until the error meets the accuracy requirement, obtaining a trained neural network model, and using the trained neural network model to realize automatic segmentation of the binary graph of the working state of the energy consumer.

[0019] Preferably, in step S3, the following is specifically included:

[0020] S31, filtering and medial axis transformation of the energy dissipator binary image output by the salient target detection deep neural network to obtain a clear skeleton image;

[0021] S32, performing an opening operation and contour detection on the skeleton image to obtain the inner and outer contour lengths of the energy consumer skeleton;

[0022] S33, calculating the remaining energy consumption capacity of the energy consumer according to the inner and outer contour lengths of the energy consumer skeleton.

[0023] Preferably, in step S13, in order to perform semantic segmentation and annotation on the energy consumer image, the significant target detection label annotation is performed by selecting Labelme, EISeg or LabelImg annotation software for annotation. In order to ensure the training effect, a unified standard is used during annotation to ensure the accuracy and consistency of the labels. In order to improve the training accuracy of the deep neural network model for significant target detection, when forming an image data set, the original image and label are rotated, scaled, and blurred to achieve image data amplification.

[0024] Preferably, in step S21, in order to ensure the training effect of the salient target detection deep neural network model, the salient target detection deep neural network is U 2 -Net, F 3 Any of the mature deep neural network structures for salient object detection such as -Net, BASNet, etc.

[0025] Preferably, in step S22, when training the deep neural network for salient object detection, the loss function L is calculated as follows:

[0026]

[0027] Where: M represents the number of feature maps, m represents the mth feature map, represents the weight coefficient of the mth sub-feature map, represents the loss value of the mth sub-feature map, w fuse Represents the total feature map weight coefficient after feature fusion, l fuse Represents the total feature map loss value after feature fusion;

[0028] In order to evaluate the training effect of the deep neural network for salient object detection, the F β , MAE is a key evaluation index to determine the model performance, and is calculated as follows:

[0029]

[0030] Where: β is the weight factor, Precision is the accuracy, Recall is the recall, H and W are the length and width of the image respectively, r and c are the pixel coordinates in the length and width directions of the image respectively, P(r,c) and G(r,c) are the model prediction value and the true value of the label respectively.

[0031] Preferably, in order to automatically calculate the remaining energy consumption capacity of the energy consumer, the binary image output by the trained neural network is further operated by morphological image processing technology to obtain the skeleton image of the energy consumer, which specifically includes three steps: filtering, medial axis transformation, and opening operation. In step S31, in order to improve the quality of the skeleton image output by the medial axis transformation, the binary image of the energy consumer output by the deep neural network for salient target detection is filtered by Python language and OpenCV software library. In order to obtain the skeleton image of the energy consumer, the MedialAxis function in the Scikit-image library is used for medial axis transformation; Gaussian filtering is selected during filtering, and the calculation is performed according to the following formula:

[0032]

[0033] Where: x g and g They represent the horizontal and vertical coordinate distances from the center of the Gaussian kernel, and δ represents the variance.

[0034] Preferably, in step S32, in order to reduce the calculation error caused by the thickness of the skeleton, an open operation is used to unify the thickness of the energy consumer skeleton based on the Python language and the OpenCV software library; in order to obtain the inner and outer contours of the energy consumer skeleton image, the Canny algorithm is used to perform contour detection on the energy consumer skeleton image, and the calculation is performed according to the following formula:

[0035] θ[i,j]=arctan(P y [i,j] / P x [i,j])

[0036]

[0037] Where: M[i,j] and θ[i,j] represent the gradient magnitude and direction of the pixel in the image, respectively; Px[i,j] and Py[i,j] represent the partial derivatives of the pixel in the x and y directions, respectively.

[0038] Preferably, in step S33, the remaining energy consumption capacity R of the energy consumer is calculated by the length of the inner and outer contours output by the skeleton image contour detection, as follows:

[0039]

[0040] L0=η(L R +L C );

[0041] Where: L R Indicates the length of the energy consumer before it is started after it actually works, L C It indicates the elongation of the energy absorber after actual operation. L0 indicates the length of the energy absorber that cannot be started under the limit state. L Outer Represents the outer contour length of the energy consumer skeleton diagram, L Inner represents the inner contour length of the energy consumer skeleton graph, and η is the incomplete starting coefficient of the energy consumer.

[0042] The beneficial effects of the present invention are:

[0043] 1) The present invention proposes a non-contact intelligent detection method for the working status of the energy absorber of the slope flexible protection system. For the first time, deep learning technology and morphological image processing technology are used in the detection of energy absorbers of flexible protection structures, realizing non-contact intelligent detection of the working status of the energy absorber, which provides a revolutionary intelligent alternative to the current manual detection method;

[0044] 2) The method of the present invention realizes non-contact detection of the working status of the energy consumer, without close contact with the disaster source, which greatly improves the safety of manual detection after the disaster; the method of the present invention realizes automatic calculation of the remaining energy consumption capacity of the energy consumer, and improves the calculation accuracy by accurate proportional calculation;

[0045] 3) The non-contact intelligent detection method for the working status of the energy absorber of the slope flexible protection system described in the present invention has clear logic, sufficient theoretical support, and high operability, providing a revolutionary intelligent alternative to the current industry status quo of the protection field that strongly relies on manual detection. It has promoted the process of "intelligence", "automation", and "de-labor" in the protection industry, explored the application of artificial intelligence technology in the protection industry, and is of great significance to promoting the progress of the industry. The present invention has substantial characteristics and progress, has a very broad market application prospect, and is very suitable for promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic flow chart of a non-contact intelligent detection method for the working state of an energy absorber of a flexible slope protection system of the present invention;

[0047] Figure 2 Schematic diagram of the quasi-static tensile test, full-scale impact test and field engineering measurement of the energy dissipator according to the embodiment of the present invention, a is the quasi-static tensile test, b is the full-scale impact test, and c is the field engineering measurement;

[0048] Figure 3 This is a schematic diagram of a portion of images obtained by a pseudo-static tensile test of an energy dissipator according to an embodiment of the present invention;

[0049] Figure 4This is a schematic diagram of a portion of the images obtained by a full-scale impact test according to an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram of some images obtained by field engineering measurement according to an embodiment of the present invention;

[0051] Figure 6 Schematic diagram of some images and labels in the energy consumption device image data set according to an embodiment of the present invention, a is an energy consumption device image, b is a label;

[0052] Figure 7 A schematic diagram of the structure of a deep neural network U2-Net for salient object detection used in an embodiment of the present invention;

[0053] Figure 8 This is a schematic diagram of the intelligent detection and segmentation results of some images in the verification set of an embodiment of the present invention;

[0054] Fig. 9 This is a schematic diagram of the process of the morphological image processing technology of the binary image of the energy consumer according to an embodiment of the present invention;

[0055] Fig.10 Schematic diagram of a method for automatically calculating the remaining energy consumption capacity of an energy consumer according to an embodiment of the present invention;

[0056] Fig.11 It is a schematic diagram of a quasi-static tensile test equipment for an energy dissipator according to an embodiment of the present invention;

[0057] Fig.12 This is a schematic diagram of key steps of using the method of the present invention to intelligently detect the working state of the energy absorber in a quasi-static tensile test image of the energy absorber according to an embodiment of the present invention;

[0058] Fig.13 This is a schematic diagram comparing the remaining energy dissipation capacity of the output energy dissipator of the simulated static tensile test in an embodiment of the present invention with the intelligent detection result of the method of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] See also Figure 1-Figure 13 The embodiment of the present invention provides a technical solution: a non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system performs non-contact intelligent detection on the remaining energy consumption capacity of the key image energy absorber during a quasi-static tensile test of a GS8002 model energy absorber, such as Figure 1 As shown, follow the steps below:

[0061] Step 1: Conduct quasi-static tensile test, full-scale impact test and field engineering measurement of the energy absorber to establish an image dataset of the working process of the energy absorber, such as Figure 2 shown.

[0062] a. Carry out a simulated static tensile test on the energy absorber to obtain a video of the entire tensile deformation process, and extract key frames to obtain images of the entire working process;

[0063] b. Carry out full-scale impact tests on protective structures and field engineering measurements to obtain images of the energy absorber's status before and after working with multiple backgrounds;

[0064] c. Label the energy dissipator images obtained from the quasi-static tensile test, full-scale impact test, and field engineering measurements with salient target detection labels, and perform data augmentation and divide them into training and validation sets;

[0065] Step 2: Use image datasets to train a deep neural network for salient object detection to achieve automatic segmentation of energy consumers

[0066] a. Determine the structure of the deep neural network for salient object detection and write the corresponding code;

[0067] b. Use the training set to train the deep neural network for salient object detection and use the validation set for testing;

[0068] c. Monitor the training error, verification error, and adjust the network structure until the error meets the accuracy requirements.

[0069] Step 3: Use morphological image processing technology to obtain the energy absorber skeleton and use contour detection technology to extract the inner and outer contours of the energy absorber skeleton to automatically calculate the remaining energy consumption capacity

[0070] a. Write code to filter and transform the binary image of the energy dissipator output by the deep neural network for salient target detection to obtain a clear skeleton image;

[0071] b. Write code to perform opening operation and contour detection on the skeleton image to obtain the inner and outer contour lengths of the energy consumer skeleton;

[0072] c. Calculate the remaining energy consumption capacity of the energy absorber based on the inner and outer contour lengths of the energy absorber frame.

[0073] The specific operation details are as follows:

[0074] In order to obtain sufficient working images of the energy absorber, 10 sets of quasi-static tensile tests of the energy absorber were carried out according to the specifications and standards, and the video of the whole process of tensile deformation was obtained. The key frames were extracted to obtain the images of the whole working process, such as Figure 3As shown in Figure 2, a total of 240 images were obtained; twenty full-scale impact tests of the flexible protective structure were carried out according to the specifications and standards, and images of the energy absorber before and after working under the full-scale impact test background were obtained, such as Figure 4 As shown in the figure, a total of 210 images were obtained; 5 field engineering tests were carried out to obtain the images of the energy dissipator before and after working in multiple backgrounds in actual engineering, such as Figure 5 A total of 50 images of energy dissipators were obtained through three methods.

[0075] In order to perform semantic segmentation and annotation on the energy consumer image, Labelme annotation software is selected for annotation. The standard should be unified during annotation to ensure the accuracy and consistency of the label, separate the energy consumer and the background, and write code to convert the annotation output json file into a binary image label for training, such as Figure 6 shown.

[0076] In order to improve the training accuracy of the deep neural network model for salient object detection, the images and labels were amplified by rotation when forming the image dataset. They were rotated 30°, 60°, 90°, 120°, 150°, and 180° respectively, and a total of 3500 images and their corresponding labels were obtained. 80% of them were divided into training sets for model training, and the remaining 20% ​​were divided into validation sets for model verification.

[0077] In order to ensure the training effect of the deep neural network model for salient target detection, U 2 -Net salient target detection deep neural network structure is trained, the network structure is as follows Figure 7 As shown, the disclosed code is modified for training in this embodiment.

[0078] When training a deep neural network for salient object detection, the loss function L can be calculated as follows:

[0079]

[0080] Where: M represents the number of feature maps, and the U used in this embodiment is 2 -Net structure M=6, m represents the mth feature map, represents the weight coefficient of the mth sub-feature map, and the value is 1. represents the loss value of the mth sub-feature map, calculated using the standard binary cross entropy, w fuse Represents the total feature map weight coefficient after feature fusion, with a value of 1, l fuse Represents the total feature map loss value after feature fusion, calculated using standard binary cross entropy.

[0081] In order to evaluate the training effect of the deep neural network for salient object detection, the maximum F β, MAE and other key evaluation indicators to determine the model performance, which can be calculated as follows:

[0082]

[0083] Where: β is the weight factor, and the U used in this embodiment is 2 -Net structure β 2 =0.3, Precision is the accuracy, Recall is the recall, H and W are the length and width of the image respectively, r and c are the pixel coordinates in the length and width directions of the image respectively, P(r,c) and G(r,c) are the model prediction value and the true value of the label respectively.

[0084] The aforementioned validation set is used to test the performance of the salient object detection neural network trained in this embodiment, and the qualitative evaluation results of some images in the validation set are shown as follows: Figure 8 As shown in the figure, we can see that the trained model can automatically segment the binary image of the energy consumer more accurately; the quantitative evaluation is calculated according to the above indicators, and the maximum F β =0.953, MAE=0.0067, the indicators reflect that the model has high accuracy.

[0085] In order to automatically calculate the remaining energy consumption capacity of the energy consumer, the binary image output by the trained neural network is further operated by morphological image processing technology (specifically including filtering, medial axis transformation, and opening operation) to obtain the skeleton image of the energy consumer, and the contour of the skeleton image is detected, such as Fig. 9 shown.

[0086] In order to improve the quality of the skeleton image output by the medial axis transformation, Python language and OpenCV software library are used to filter the binary image of the energy dissipator output by the deep neural network for salient target detection. Gaussian filtering is selected for filtering, and the calculation is based on the following formula:

[0087]

[0088] Where: x g and g They represent the horizontal and vertical coordinate distances from the center of the Gaussian kernel respectively, and δ represents the variance value of 0.8.

[0089] In order to obtain the skeleton diagram of the energy dissipator, the Python language and the MedialAxis function in the Scikit-image library were used to perform medial axis transformation.

[0090] In order to reduce the calculation error caused by the thickness of the skeleton, Python language and the open operation in the OpenCV software library are used to process the skeleton image to unify the skeleton thickness.

[0091] In order to obtain the inner and outer contours of the energy consumer skeleton image, the Canny algorithm is used to detect the contour of the energy consumer skeleton image based on Python language and OpenCV software library, and the calculation is based on the following formula:

[0092] θ[i,j]=arctan(P y [i,j] / P x [i,j])

[0093]

[0094] Where: M[i,j] and θ[i,j] represent the gradient magnitude and direction of the pixel in the image, respectively. x [i,j] and P y [i,j] represents the partial derivatives of the pixel in the x and y directions respectively.

[0095] In order to automatically calculate the remaining energy consumption capacity R of the energy consumer, the inner and outer contour lengths output by the skeleton contour detection can be used for calculation, such as Fig.10 As shown in the figure, this calculation method is equivalent to directly calculating the ratio of the potential start-up length to the total length after the energy consumer actually works, as shown in the following formula:

[0096]

[0097] L0=η(L R +L C )

[0098] Where: L R Indicates the length of the energy consumer before it is started after it actually works, L C Indicates the elongation of the energy absorber after actual operation. L0 indicates the length of the energy absorber that cannot be started under the limit state. For GS8002 energy absorber, L0 = 416.68 mm. Outer Indicates the outer contour length of the energy consumer skeleton diagram, L Inner It represents the inner contour length of the energy absorber skeleton diagram, η is the incomplete starting coefficient of the energy absorber, and GS8002 energy absorber η=0.298.

[0099] In order to verify the effectiveness of the method of the present invention, a quasi-static tensile test of the energy absorber is carried out to perform non-contact intelligent detection of the working state of the energy absorber after the test. The test equipment is as follows: Fig.11 shown.

[0100] During the test, ten pictures with the remaining energy consumption capacity of the test equipment outputting 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, and 10% were selected and numbered R1 to R10. They were input into the trained salient target detection neural network to obtain their binary images. The obtained binary images were subjected to Gaussian filtering, medial axis transformation, opening operation, and contour detection to obtain the inner and outer contours of the skeleton image, as shown in Figure 2. Fig.12 Finally, write the code to automatically calculate the remaining energy consumption capacity of the energy consumer through the inner and outer contour lengths of the energy consumer skeleton diagram.

[0101] In order to illustrate the accuracy of the method of the present invention, the remaining energy consumption capacity of the energy consumer detected by intelligent detection is compared with the output results of the test equipment. Fig.13 The error between the intelligent detection and the test result is less than 3%, which proves the accuracy of the method of the present invention.

[0102] The method of the present invention solves the problem of non-contact intelligent identification of the working state of key energy-consuming components of flexible protection systems through artificial intelligence technology, has high accuracy, and realizes the integrated application of visual monitoring of the appearance and morphology of key components of protection projects and judgment of their internal performance.

[0103] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A non-contact intelligent detection method for the working status of energy dissipators in a slope flexible protection system. It is characterized in that The steps include: S1. Conduct simulated static tensile test, full-scale impact test and field engineering measurement of energy absorber to establish image data set of working process of energy absorber; S2, using the data set to train a deep neural network for salient target detection to achieve automatic segmentation of the binary image of the working state of the energy consumer; S3. Use morphological image processing technology to obtain the skeleton of the energy absorber and use contour detection technology to extract the inner and outer contours of the skeleton of the energy absorber to automatically calculate the remaining energy consumption capacity.

2. According to claim 1, the non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system, Features: In step S1, the specific steps include: S11. Conduct a simulated static tensile test on the energy absorber to obtain a video of the entire tensile deformation process, and extract key frames to obtain images of the entire working process; S12. Carry out full-scale impact tests on protective structures and field engineering measurements to obtain images of the energy absorber's status before and after working with multiple backgrounds; S13. The images of energy absorbers obtained from the quasi-static tensile test, full-scale impact test and field engineering measurements are annotated with salient target detection labels, and data is augmented and divided into training and validation sets.

3. According to claim 1, the non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system, Features: In step S2, the specific steps include: S21. Determine the structure of the deep neural network for salient object detection and write code; S22, using the training set to train the deep neural network for salient object detection and using the validation set for testing; S23, monitoring the training error, verifying the error and adjusting the network structure until the error meets the accuracy requirement, obtaining a trained neural network model, and using the trained neural network model to realize automatic segmentation of the binary graph of the working state of the energy consumer.

4. According to claim 1, the non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system, Features: In step S3, the specific steps include: S31, filtering and medial axis transformation of the energy dissipator binary image output by the salient target detection deep neural network to obtain a clear skeleton image; S32, performing an opening operation and contour detection on the skeleton image to obtain the inner and outer contour lengths of the energy consumer skeleton; S33, calculating the remaining energy consumption capacity of the energy consumer according to the inner and outer contour lengths of the energy consumer skeleton.

5. According to claim 2, the non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system, Features: In step S13, the notable target detection label annotation is performed by selecting Labelme, EISeg or LabelImg annotation software for annotation, and a unified standard is used during annotation to ensure the accuracy and consistency of the labels; when forming an image data set, the original image and label are rotated, scaled, and blurred to achieve image data amplification.

6. The non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system according to claim 3, Features: In step S21, the salient object detection deep neural network is U 2 -Net, F 3 -Net, BASNet or any other.

7. The non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system according to claim 3, Features: In step S22, when training the deep neural network for salient object detection, the loss function L is calculated as follows: Where: M represents the number of feature maps, m represents the mth feature map, represents the weight coefficient of the mth sub-feature map, represents the loss value of the mth sub-feature map, w fuse Represents the total feature map weight coefficient after feature fusion, l fuse Represents the total feature map loss value after feature fusion; When training a deep neural network for salient object detection, we use F β , MAE is a key evaluation index to determine the model performance, and is calculated as follows: Where: β is the weight factor, Precision is the accuracy, Recall is the recall, H and W are the length and width of the image respectively, r and c are the pixel coordinates in the length and width directions of the image respectively, P(r,c) and G(r,c) are the model prediction value and the true value of the label respectively.

8. According to claim 4, the non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system, Features: In step S31, Python language and OpenCV software library are used to filter the binary image of energy dissipators output by the deep neural network for salient target detection, and then the MedialAxis function in the Scikit-image library is used to perform medial axis transformation; Gaussian filtering is selected during filtering, and the calculation is performed according to the following formula: Where: x g and g They represent the horizontal and vertical coordinate distances from the center of the Gaussian kernel, and δ represents the variance.

9. The non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system according to claim 4, Features: In step S32, based on Python language and OpenCV software library, open operation is used to unify the thickness of the energy consumer skeleton, and then the Canny algorithm is used to detect the contour of the energy consumer skeleton image, and the calculation is based on the following formula: θ[i,j]=arctan(P y [i,j] / P x [i,j]) Where: M[i,j] and θ[i,j] represent the gradient magnitude and direction of the pixel in the image, respectively; Px[i,j] and Py[i,j] represent the partial derivatives of the pixel in the x and y directions, respectively.

10. The non-contact intelligent detection method for the working state of the energy absorber of the slope flexible protection system according to claim 4, Features: The remaining energy consumption capacity R of the energy consumer is calculated by the inner and outer contour lengths output by the skeleton image contour detection, as follows: L 0 =η(L R +L C ); Where: L R Indicates the length of the energy consumer before it is started after it actually works, L C Indicates the elongation of the energy absorber after actual operation, L 0 Indicates the length of the energy absorber that cannot be activated under the limit state, L Outer Indicates the outer contour length of the energy consumer skeleton diagram, L Inner represents the inner contour length of the energy consumer skeleton graph, and η is the incomplete start-up coefficient of the energy consumer.

Citation Information

Patent Citations

  • Image salient target detection method combined with deep learning

    CN108898145A

  • Saliency-guided weak supervision target detection method

    CN110598609A

  • Multi-organ instance segmentation method and system

    CN114092439A

  • Flexible protection system intelligent rapid calculation method based on multilayer perceptron

    CN116822146A

  • Bridge crack intelligent identification and measurement method

    CN117911371A

Cited By

  • Intelligent evaluation method for residual performance of passive flexible protection system based on target detection

    CN122200044A