Intelligent pile foundation low strain detection analysis method and system

Abstract

The invention discloses an intelligent pile foundation low-strain detection analysis method, which comprises the following steps: constructing a pile foundation engineering cloud database, storing to-be-detected design pile foundation parameters corresponding to engineering in the cloud database, comparing data acquired by different sensing devices, and judging whether complex calculation is needed to determine the quality analysis of an actual pile foundation. The closer the data obtained by different sensing devices is, the simpler the calculation and verification work needs to be carried out, the size of an object in an image is determined through the image recognition and processing technology, the second pile length data is calculated according to the propagation velocity mapping relation between the pile body material in the design pile foundation parameters and the first vibration force, and the second pile length data is calculated according to the second vibration force. And when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is larger than a threshold value, a trained pile foundation low-strain detection analysis model is used for matching and analyzing the data, a judgment result of the pile foundation quality is automatically given out according to a matching result, and meanwhile additional analysis verification is not carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of detection and analysis, and in particular to a method and system for intelligent pile foundation low-strain detection and analysis. Background Art

[0002] At present, in construction projects, pile structures are mainly made of concrete. If the pouring speed is not controlled during pouring, it is easy to have missing pile structures. Therefore, when inspecting the quality of pile structures, its integrity is also part of the basic inspection content. The pile foundation structure is located underground, and when inspecting its integrity, it cannot be inspected by appearance observation. Combined with previous inspections, ultrasonic flaw detection inspection methods are generally selected, such as low-strain dynamic detection method, ultrasonic detection method, etc. When analyzing data, interference factors need to be eliminated in advance to improve the accuracy of the analysis results. The existing detection technology mainly uses manual on-site measurement area data and graphics, and then exports them to the computer for manual counting and graphic analysis to issue corresponding results. The current method is inefficient and it is believed that the probability of error is relatively high. The construction of an intelligent pile foundation quality analysis system can improve the efficiency and accuracy of analysis and reduce the error factor of human intervention, but the technical training of the system requires a large amount of data and computing resources, so it requires a lot of cost and time. At the same time, for the low-strain detection method, the measurement error also needs to be verified to ensure the accuracy of the detection and analysis. However, the existing technical solutions either do not perform corresponding verification or use complex data modeling for direct verification, resulting in large amounts of calculation and long calculation time. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention discloses a method for intelligent pile foundation low strain detection and analysis, the method comprising:

[0004] Step 1, constructing a cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the pile foundation;

[0005] Step 2, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by a first sensing device, applying a first vibration force to the top of the actual pile foundation by a second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile;

[0006] Step 3, the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation. The actual pile foundation is analyzed according to the internal and external parameters determined by the camera device and the fixed position information, and the first pile length data and the first pile diameter data obtained by the first sensing device are determined, that is, the size of the object in the image is determined by image recognition and processing technology, and the second pile length data is calculated according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force;

[0007] Step 4: when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, the newly input waveform graphic data is matched and analyzed using the trained pile foundation low strain detection and analysis model, and a determination result of the pile foundation quality is automatically issued based on the matching and analysis results, without performing additional analysis and verification;

[0008] Step 5: When the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is less than a first threshold value, the newly input waveform graphic data is matched and analyzed according to the trained pile foundation low-strain detection and analysis model. After obtaining the analysis results, a pile-soil solid finite element model is established according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database. The response speed of the pile top during transient excitation is solved by the ANSYS / LS-DYNA program. The analysis results are verified and quantitatively analyzed by numerical simulation of the foundation piles.

[0009] Furthermore, the training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and divide the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, pile foundation cracking or broken piles, and use different pile foundation quality classifications as evaluation and analysis results, annotate the collected graphic data, and use the annotated data to train the pile foundation low strain detection and analysis model.

[0010] Furthermore, the pile foundation low-strain detection and analysis model is a convolutional neural network model, wherein, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational complexity of data processing and improving the convergence speed.

[0011] Furthermore, the second sensing device includes sensors placed at the pile top and the pile bottom for recording the propagation time and amplitude of the strain wave when the pile top is subjected to vibration force, and calculating the wave velocity in the pile foundation through the propagation time and amplitude.

[0012] Furthermore, the loss function of the pile foundation low strain detection analysis model is:

[0013]

[0014] Among them, L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the n-th input waveform image, xn represents the overall feature vector of the n-th input waveform image in the input image set, xp represents the overall feature vector of all images in the input image set that have the same pile foundation quality classification label as the n-th input waveform image, xq represents the overall feature vector of all images in the input image set that have different pile foundation quality classification labels from the n-th input waveform image, D represents the distance between two vectors, and [·]+ represents the maximum value between 0 and the value in the brackets.

[0015] The present invention also discloses an intelligent pile foundation low strain detection and analysis system, the system comprising:

[0016] A cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the foundation pile;

[0017] first and second sensing devices, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by the first sensing device, applying a first vibration force to the top of the actual pile foundation by the second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile;

[0018] The computing unit, wherein the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation, analyzes the actual pile foundation according to the internal and external parameters determined by the camera device and the fixed position information, determines the first pile length data and the first pile diameter data obtained by the first sensing device, that is, determines the size of the object in the image through image recognition and processing technology, and calculates the second pile length data according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force;

[0019] a first comparison and verification unit, when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, using the trained pile foundation low strain detection and analysis model to match and analyze the newly input waveform graphic data, and automatically issuing a determination result of the pile foundation quality according to the matching and analysis results, without performing additional analysis and verification;

[0020] The second comparison and verification unit, when the similarity between the value of the second pile length data minus the embedding depth and the first pile length data is less than a first threshold, matches and analyzes the newly input waveform graphic data according to the trained pile foundation low strain detection and analysis model, and after obtaining the analysis results, establishes a pile-soil solid finite element model according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database, uses the ANSYS / LS-DYNA program to solve the response speed of the pile top during transient excitation, verifies the analysis results through numerical simulation of the foundation piles and performs quantitative analysis.

[0021] Furthermore, the training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and divide the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, pile foundation cracking or broken piles, and use different pile foundation quality classifications as evaluation and analysis results, annotate the collected graphic data, and use the annotated data to train the pile foundation low strain detection and analysis model.

[0022] Furthermore, the pile foundation low-strain detection and analysis model is a convolutional neural network model, wherein, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational complexity of data processing and improving the convergence speed.

[0023] Furthermore, the second sensing device includes sensors placed at the pile top and the pile bottom for recording the propagation time and amplitude of the strain wave when the pile top is subjected to vibration force, and calculating the wave velocity in the pile foundation through the propagation time and amplitude.

[0024] Furthermore, the loss function of the pile foundation low strain detection analysis model is:

[0025]

[0026] Among them, L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the n-th input waveform image, xn represents the overall feature vector of the n-th input waveform image in the input image set, xp represents the overall feature vector of all images in the input image set that have the same pile foundation quality classification label as the n-th input waveform image, xq represents the overall feature vector of all images in the input image set that have different pile foundation quality classification labels from the n-th input waveform image, D represents the distance between two vectors, and [·]+ represents the maximum value between 0 and the value in the brackets.

[0027] Compared with the prior art, the present invention has the following beneficial effects: the present invention combines the existing research methods, and on this basis proposes a method for hierarchically judging whether it is necessary to verify the result of the low-strain detection method, which greatly reduces the amount of calculation and the calculation time. When there may be a large error in the calculation result, the finite element analysis is further performed, the pile-soil solid model is established by using UG software, and the velocity response of the pile top during transient excitation is solved by using ANSYS / LS-DYNA program. By calculating the pile length and the defect position, the effectiveness of the numerical simulation of the pile-soil finite element model and the reliability of the calculation result are verified, and the foundation pile can be effectively numerically simulated. Based on the MATLAB platform, a quantitative analysis program for pile foundation is compiled by adopting the least square method according to the mathematical model between pile and soil, and the pile foundation parameters are quantitatively analyzed by fitting the model curve with the measured curve. When the error of the calculation result may be small, no second verification is performed. At the same time, the present invention integrates data annotation by entering a large amount of existing detected graphic data and corresponding evaluation and analysis results into the system, and the existing large amount of data annotation can be automatically matched and analyzed with the measured graphics behind to automatically issue the judgment result of the pile foundation quality. In order to speed up the training convergence speed and accuracy of the model, the first detection parameter judgment method is adopted to optimize whether the parameters of the neural network need to be iteratively modified, thereby further reducing the amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but the emphasis is placed on illustrating the principles of the embodiments. In the figures, the same reference numerals designate corresponding parts in different views.

[0029] Figure 1 It is a flow chart of a method for intelligent pile foundation low strain detection and analysis of the present invention. DETAILED DESCRIPTION

[0030] Embodiment 1

[0031] like Figure 1As shown, this embodiment provides a method for intelligent pile foundation low strain detection and analysis, the method comprising:

[0032] Step 1, constructing a cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the pile foundation;

[0033] Step 2, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by a first sensing device, applying a first vibration force to the top of the actual pile foundation by a second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile;

[0034] Step 3, the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation. The actual pile foundation is analyzed according to the internal and external parameters determined by the camera device and the fixed position information, and the first pile length data and the first pile diameter data obtained by the first sensing device are determined, that is, the size of the object in the image is determined by image recognition and processing technology, and the second pile length data is calculated according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force;

[0035] Step 4: when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, the newly input waveform graphic data is matched and analyzed using the trained pile foundation low strain detection and analysis model, and a determination result of the pile foundation quality is automatically issued based on the matching and analysis results, without performing additional analysis and verification;

[0036] Step 5: When the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is less than a first threshold value, the newly input waveform graphic data is matched and analyzed according to the trained pile foundation low-strain detection and analysis model. After obtaining the analysis results, a pile-soil solid finite element model is established according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database. The response speed of the pile top during transient excitation is solved by the ANSYS / LS-DYNA program. The analysis results are verified and quantitatively analyzed by numerical simulation of the foundation piles.

[0037] Furthermore, the training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and divide the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, pile foundation cracking or broken piles, and use different pile foundation quality classifications as evaluation and analysis results, annotate the collected graphic data, and use the annotated data to train the pile foundation low strain detection and analysis model.

[0038] Furthermore, the pile foundation low-strain detection and analysis model is a convolutional neural network model, wherein, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational complexity of data processing and improving the convergence speed.

[0039] Furthermore, the second sensing device includes sensors placed at the pile top and the pile bottom for recording the propagation time and amplitude of the strain wave when the pile top is subjected to vibration force, and calculating the wave velocity in the pile foundation through the propagation time and amplitude.

[0040] In this embodiment, the wave velocity of the strain wave is obtained according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation velocity of the first vibration force. Optionally, a second sensing device includes sensors placed at the top and bottom of the pile to record the propagation time and amplitude of the strain wave when the pile top is subjected to the vibration force. The wave velocity in the pile foundation is calculated according to the propagation time and amplitude. By comparing the calculated wave velocity with the wave velocity in the mapping relationship, it is determined whether the construction material meets the expected quality requirements. When the difference between the calculated wave velocity and the wave velocity obtained according to the mapping relationship is less than a preset value, the construction material is determined to be preliminarily qualified.

[0041] Due to the problem of unqualified construction materials (for example, the materials are unqualified but the wave velocity of the corresponding specific wave is similar), the vibration force propagation velocity is inaccurate in the second pile length data calculated according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation velocity of the first vibration force, and the calculated pile length data has a large error. Therefore, another beneficial effect of this embodiment compared with the prior art is: when the similarity between the value of the second pile length data minus the embedding depth and the first pile length data is greater than or equal to the first threshold, it is compared with the second threshold, wherein the second threshold is greater than the first threshold, and when the similarity is greater than the second threshold, a second vibration force is applied to the pile top of the actual pile foundation through a second sensing device, and a dynamic strain waveform corresponding to the second vibration force propagating downward from the pile top is obtained, and the second vibration force is a vibration force with an amplitude and frequency different from the first vibration force, and then jump back to step 1 for re-detection and analysis.

[0042] Compared with the prior art, no additional calculation propagation speed is required to exclude calculation errors that may be caused by unqualified construction materials.

[0043] Furthermore, the loss function of the pile foundation low strain detection analysis model is:

[0044]

[0045] Among them, L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the nth input waveform image, x n represents the overall feature vector of the nth input waveform image in the input image set, x p represents the overall feature vector of all images in the input image set that have the same pile quality classification label as the nth input waveform image, x q represents the overall feature vector of all images in the input image set that have different pile quality classification labels from the nth input waveform image, D represents the distance between the two vectors, [·] + Indicates the maximum value between 0 and the value in the brackets.

[0046] Embodiment 2

[0047] From the perspective of hardware description, the intelligent pile foundation low strain detection and analysis system in this embodiment is an intelligent detection system designed specifically for pile foundation engineering, with functions such as real-time remote monitoring, no need for measurement and setting out, export of monitoring records as reports, and real-time transmission of construction data to the background. The system can upload data to the engineering management platform in real time through the mobile communication network built into the control terminal, thereby realizing remote management of pile foundation construction. The present invention also discloses an intelligent pile foundation low strain detection and analysis system, the system comprising:

[0048] A cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the foundation pile;

[0049] first and second sensing devices, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by the first sensing device, applying a first vibration force to the top of the actual pile foundation by the second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile;

[0050] The computing unit, wherein the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation, analyzes the actual pile foundation according to the internal and external parameters determined by the camera device and the fixed position information, determines the first pile length data and the first pile diameter data obtained by the first sensing device, that is, determines the size of the object in the image through image recognition and processing technology, and calculates the second pile length data according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force;

[0051] a first comparison and verification unit, when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, using the trained pile foundation low strain detection and analysis model to match and analyze the newly input waveform graphic data, and automatically issuing a determination result of the pile foundation quality according to the matching and analysis results, without performing additional analysis and verification;

[0052] The second comparison and verification unit, when the similarity between the value of the second pile length data minus the embedding depth and the first pile length data is less than a first threshold, matches and analyzes the newly input waveform graphic data according to the trained pile foundation low strain detection and analysis model, and after obtaining the analysis results, establishes a pile-soil solid finite element model according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database, uses the ANSYS / LS-DYNA program to solve the response speed of the pile top during transient excitation, verifies the analysis results through numerical simulation of the foundation piles and performs quantitative analysis.

[0053] Furthermore, the training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and divide the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, pile foundation cracking or broken piles, and use different pile foundation quality classifications as evaluation and analysis results, annotate the collected graphic data, and use the annotated data to train the pile foundation low strain detection and analysis model.

[0054] Furthermore, the pile foundation low-strain detection and analysis model is a convolutional neural network model, wherein, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational complexity of data processing and improving the convergence speed.

[0055] Furthermore, the second sensing device includes sensors placed at the pile top and the pile bottom for recording the propagation time and amplitude of the strain wave when the pile top is subjected to vibration force, and calculating the wave velocity in the pile foundation through the propagation time and amplitude.

[0056] Furthermore, the loss function of the pile foundation low strain detection analysis model is:

[0057]

[0058] Among them, L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the nth input waveform image, x n represents the overall feature vector of the nth input waveform image in the input image set, x p represents the overall feature vector of all images in the input image set that have the same pile quality classification label as the nth input waveform image, x q represents the overall feature vector of all images in the input image set that have different pile quality classification labels from the nth input waveform image, D represents the distance between the two vectors, and [·]+ represents the maximum value between 0 and the value in the brackets.

[0059] The hardware equipment of the system includes Beidou high-precision positioning and orientation terminal, depth sensor, inclination sensor, hammer count sensor and other equipment installed on the pile foundation drilling rig. The data collected by the sensor can be processed and analyzed to obtain information such as the construction quality, depth, verticality, etc. of the pile foundation. At the same time, these data can also be transmitted to the background management system in real time through the mobile communication network, which is convenient for managers to monitor and guide the construction process.

[0060] The software application system of the system can process and analyze the collected data, generate construction progress reports, construction quality reports, pile foundation condition reports, etc., to help managers better understand the construction situation and find and solve problems in a timely manner. At the same time, the software system can also generate various reports and charts as needed to facilitate managers to analyze data and make decisions.

[0061] Data collection layer: This layer is responsible for collecting data from various sensors and equipment, including various parameters in the construction process, such as construction progress, construction quality, pile foundation conditions, etc.

[0062] Data processing layer: This layer is responsible for processing and analyzing the collected data, including data cleaning, data conversion, data mining, etc.

[0063] Data labeling: Label the collected graphic data based on the existing evaluation and analysis results. For example, the pile foundation quality can be divided into insufficient pile foundation bearing capacity, excessive pile foundation inclination, cracked pile foundation or broken piles, etc.

[0064] Model training: Use labeled data to train the model, and improve the accuracy and generalization ability of the model by adjusting model parameters and optimizing algorithms.

[0065] Automatic matching and analysis: Use the trained model to match and analyze the newly input graphic data, and automatically issue the judgment result of the pile foundation quality based on the matching and analysis results.

[0066] Further realize real-time remote monitoring: the intelligent pile foundation low-strain detection and analysis system can transmit data to the background management system in real time through the mobile communication network. Managers can check the construction status anytime and anywhere through computers or mobile phones to realize remote monitoring and management.

[0067] Furthermore, the monitoring records are exported as reports: the intelligent pile foundation low-strain detection and analysis system can automatically generate various reports and charts from the collected data, making it convenient for managers to conduct data analysis and decision-making.

[0068] At the same time, construction data is transmitted to the background in real time: the system can transmit the collected construction data to the background management system in real time, and managers can view the construction data at any time to discover and solve problems in time.

[0069] High degree of automation: The intelligent pile foundation low strain detection and analysis system has a high degree of automation and can automatically complete data collection, processing and analysis, reducing the possibility of manual intervention and errors.

[0070] Multifunctional application: The system not only has pile foundation detection and analysis functions, but can also be expanded as needed, such as adding personnel management, material management, construction progress management and other functions.

[0071] Strong data processing capability: The intelligent pile foundation low-strain detection and analysis system integrates a large number of existing detection graphics and data into data control, and can quickly match, identify, process and analyze the detected graphic data, thereby improving data processing efficiency and accuracy.

[0072] High security: The system adopts a variety of security measures, including data encryption transmission, data backup and recovery functions, to ensure the security and integrity of data.

[0073] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0074] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0075] Although the present invention has been described above with reference to various embodiments, it should be understood that many changes and modifications can be made without departing from the scope of the present invention. Therefore, it is intended that the above detailed description is considered to be illustrative rather than restrictive, and it should be understood that the following claims (including all equivalents) are intended to limit the spirit and scope of the present invention. The above embodiments should be understood to be only used to illustrate the present invention and not to limit the scope of protection of the present invention. After reading the content of the record of the present invention, the technician can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall into the scope defined by the claims of the present invention.

Claims

1. A method for intelligent pile foundation low strain detection and analysis, characterized in that: The method comprises: Step 1, constructing a cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the pile foundation; Step 2, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by a first sensing device, applying a first vibration force to the top of the actual pile foundation by a second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile; Step 3, the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation. The actual pile foundation is analyzed according to the internal and external parameters determined by the camera device and the fixed position information, and the first pile length data and the first pile diameter data obtained by the first sensing device are determined, that is, the size of the object in the image is determined by image recognition and processing technology, and the second pile length data is calculated according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force; Step 4: when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, the newly input waveform graphic data is matched and analyzed using the trained pile foundation low strain detection and analysis model, and a determination result of the pile foundation quality is automatically issued based on the matching and analysis results, without performing additional analysis and verification; Step 5: When the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is less than a first threshold value, the newly input waveform graphic data is matched and analyzed according to the trained pile foundation low-strain detection and analysis model. After obtaining the analysis results, a pile-soil solid finite element model is established according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database. The response speed of the pile top during transient excitation is solved by the ANSYS / LS-DYNA program. The analysis results are verified and quantitatively analyzed by numerical simulation of the foundation piles.

2. The method for low-strain detection and analysis of intelligent pile foundation according to claim 1, characterized in that: The training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and classify the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, cracked pile foundation or broken piles. According to different pile foundation quality classifications as the evaluation and analysis results, the collected graphic data is labeled, and the labeled data is used to train the pile foundation low strain detection and analysis model.

3. The method for intelligent pile foundation low strain detection and analysis as claimed in claim 2, characterized in that: The pile foundation low strain detection and analysis model is a convolutional neural network model, in which, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational work in data processing and improving the convergence speed.

4. The method for low-strain detection and analysis of intelligent pile foundation according to claim 1, characterized in that: The second sensing device includes sensors placed at the top and bottom of the pile for recording the propagation time and amplitude of the strain wave when the top of the pile is subjected to vibration force, and the wave velocity in the pile foundation is calculated according to the propagation time and amplitude.

5. The method for intelligent pile foundation low strain detection and analysis as claimed in claim 1, characterized in that: The loss function of the pile foundation low strain detection analysis model is: Where L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the nth input waveform image, x n represents the overall feature vector of the nth input waveform image in the input image set, x p represents the overall feature vector of all images in the input image set that have the same pile quality classification label as the nth input waveform image, x q represents the overall feature vector of all images in the input image set that have different pile quality classification labels from the nth input waveform image, D represents the distance between the two vectors, [·] + Indicates the maximum value between 0 and the value in the brackets.

6. An intelligent pile foundation low strain detection and analysis system, characterized in that: The system comprises: A cloud database for pile foundation engineering, wherein the cloud database stores the designed pile foundation parameters to be tested corresponding to the engineering, wherein the designed pile foundation parameters include pile length, pile diameter, pile body material, embedding depth, and soil type around the foundation pile; first and second sensing devices, obtaining image data of an actual pile foundation corresponding to the designed pile foundation to be detected by the first sensing device, applying a first vibration force to the top of the actual pile foundation by the second sensing device, and obtaining a dynamic strain waveform corresponding to the first vibration force propagating downward from the top of the pile; The computing unit, wherein the first sensing device is a camera device fixed at a preset position on the construction site and capable of collecting a complete image of the actual pile foundation, analyzes the actual pile foundation according to the internal and external parameters determined by the camera device and the fixed position information, determines the first pile length data and the first pile diameter data obtained by the first sensing device, that is, determines the size of the object in the image through image recognition and processing technology, and calculates the second pile length data according to the mapping relationship between the pile body material in the designed pile foundation parameters and the propagation speed of the first vibration force; a first comparison and verification unit, when the similarity between the value obtained by subtracting the embedding depth from the second pile length data and the first pile length data is greater than or equal to a first threshold, using the trained pile foundation low strain detection and analysis model to match and analyze the newly input waveform graphic data, and automatically issuing a determination result of the pile foundation quality according to the matching and analysis results, without performing additional analysis and verification; The second comparison and verification unit, when the similarity between the value of the second pile length data minus the embedding depth and the first pile length data is less than a first threshold, matches and analyzes the newly input waveform graphic data according to the trained pile foundation low strain detection and analysis model, and after obtaining the analysis results, establishes a pile-soil solid finite element model according to the data collected by the second sensing device and the design pile foundation parameters in the cloud database, uses the ANSYS / LS-DYNA program to solve the response speed of the pile top during transient excitation, verifies the analysis results through numerical simulation of the foundation piles and performs quantitative analysis.

7. The intelligent pile foundation low strain detection and analysis system according to claim 6, characterized in that: The training of the pile foundation low strain detection and analysis model further includes: preprocessing and analyzing the dynamic strain waveform data collected by the second sensing device, wherein the preprocessing is to clean the collected waveform image to reduce the amount of calculated data, and at the same time convert the format of the cleaned data to obtain a data format corresponding to the data analysis, and classify the pile foundation quality into insufficient pile foundation bearing capacity, excessive pile foundation inclination, cracked pile foundation or broken piles. According to different pile foundation quality classifications as the evaluation and analysis results, the collected graphic data is labeled, and the labeled data is used to train the pile foundation low strain detection and analysis model.

8. The intelligent pile foundation low strain detection and analysis system as claimed in claim 7, characterized in that: The pile foundation low strain detection and analysis model is a convolutional neural network model, in which, during the training process, convolution kernels whose weights do not change with back propagation are set in the convolution layer. The closer the value of the second pile length data minus the embedding depth is to the first pile length data, the greater the number of convolution kernels that do not change with back propagation, and the greater the corresponding fixed value weights, thereby reducing the amount of computational work in data processing and improving the convergence speed.

9. The intelligent pile foundation low strain detection and analysis system according to claim 6, characterized in that: The second sensing device includes sensors placed at the top and bottom of the pile for recording the propagation time and amplitude of the strain wave when the top of the pile is subjected to vibration force, and the wave velocity in the pile foundation is calculated according to the propagation time and amplitude.

10. The intelligent pile foundation low strain detection and analysis system according to claim 6, characterized in that: The loss function of the pile foundation low strain detection analysis model is: Where L represents the loss function, represents the first threshold, P represents the number of pile foundation quality classifications in the set of input waveform images, K represents the number of waveform images under each foundation quality classification in the set of input waveform images, n represents the nth input waveform image, x n represents the overall feature vector of the nth input waveform image in the input image set, x p represents the overall feature vector of all images in the input image set that have the same pile quality classification label as the nth input waveform image, x q represents the overall feature vector of all images in the input image set that have different pile quality classification labels from the nth input waveform image, D represents the distance between the two vectors, [·] + Indicates the maximum value between 0 and the value in the brackets.

Images