Multi-modal signal acquisition and defect detection system for foundation pile integrity evaluation

Abstract

The invention provides a multi-modal signal acquisition and defect detection system for foundation pile integrity evaluation, and relates to the technical field of foundation pile integrity evaluation. The system is composed of a signal acquisition device and a defect detection system. The signal acquisition device comprises a supporting system, a lifting system and an auxiliary system, and is used for mounting an acceleration sensor, an ultrasonic transducer or an optical sensor required by defect detection. The defect detection system comprises a defect judgment module and a multi-modal data fusion unit. And the defect judgment module is used for performing low-strain defect detection, ultrasonic defect detection and optical defect detection on signals acquired by the sensors. And the multi-modal data fusion unit fuses the defect detection results, and performs comprehensive judgment on the foundation pile defects by adopting a priority classification and weighting calculation mechanism. According to the system, the defects of the foundation pile can be accurately detected, and integrity evaluation of the foundation pile is completed.

Description

Technical Field

[0001] This invention relates to the field of pile integrity assessment technology, and in particular to a multimodal signal acquisition and defect detection system for pile integrity assessment. Background Technology

[0002] Offshore wind farm overhead line foundation piles are subjected to long-term erosion and corrosion from seawater, making it difficult for them to meet integrity requirements. To assess the safe operating condition of the piles throughout their service life, low-strain dynamics and ultrasonic transduction methods are commonly used to detect the location of internal defects. Furthermore, cracks exposed on the pile surface can severely impact the pile's service life and durability. Failure to promptly detect and address cracks will lead to more serious damage to the structure operating with these defects for an extended period. Therefore, in practical applications, low-strain dynamics, ultrasonic testing, and crack detection are often combined for pile defect detection.

[0003] In low-strain testing, the testing point for solid piles is typically located at 2 / 3 of the radius from the pile center. A sensor needs to be attached to the testing point, and a manual hand-held hammer is used to generate a stress wave, which is then received by the sensor. However, there is currently a lack of sensor fixing devices capable of precisely controlling the testing point position. Furthermore, when using low-strain testing results to determine pile defects, manual interpretation is often used. This involves analyzing wave signal characteristics (including waveform, wave velocity, frequency, amplitude, and resonant peak frequency difference) for manual judgment, requiring a high level of experience from the personnel.

[0004] Currently, the main method for detecting pile integrity using ultrasonic transmission is to pre-embed acoustic logging tubes and use a transmitter and receiver to gradually move up and down along the tubes to detect internal defects in the pile. However, for the pile foundations of overhead lines in offshore wind farms, exposed piles are prone to quality problems due to construction quality, periodic tidal effects, and changes in temperature and humidity. Ultrasonic testing has also been used to detect defects in exposed sections of the pile. For this application scenario, it requires manual installation of the signal transmitter, receiver, and optical sensors at different heights on the pile surface. Two people hold the optical sensors until the test is completed, lacking a precise sensor fixing device to control the measurement point position. When judging pile defects based on ultrasonic testing results, the characteristics of the wave signal (including waveform, sound velocity, amplitude, dominant frequency, PSD, etc.) are analyzed and distributed along the pile length. When there are many cross-sections to be measured, the data from all different measurement lines on all cross-sections is enormous, making analysis and processing time-consuming and labor-intensive, with low automation.

[0005] Crack detection commonly relies on visual inspection or the aid of tools. For high-quality pile foundation surfaces, image detection systems can improve accuracy. However, regardless of the crack detection tool used, a device to secure the detector is often lacking. Crack detection in pile appearance inspection typically uses visual inspection or crack width gauges to determine crack width. These methods lack precision and require manual location of the crack before determining its dimensions. Human error is inevitable in both locating and interpreting cracks.

[0006] In summary, current technological advancements in pile foundation testing equipment lack efficient, integrated sensor mounting devices. Furthermore, there is a lack of integrated defect detection and analysis systems that combine low-strain, ultrasonic, and crack detection multimodal analysis for pile foundation testing data processing. Summary of the Invention

[0007] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows: According to a first aspect of this application, a multimodal signal acquisition and defect detection system for pile integrity assessment is provided, the system comprising: a signal acquisition device and a defect detection system; The signal acquisition device is used to install accelerometers, ultrasonic transducers, or optical sensors. The signal acquisition device includes a support system, a lifting system, and an auxiliary system. The support system includes a spiral support frame and an angle steel support frame, which are connected and fixed by welding. The lifting system includes: a scissor-type folding frame, a pallet, hinge bolts, and an electric control box. The scissor-type folding frame is constructed by connecting steel plates with hinge bolts, and the scissor-type folding frame and the pallet are connected by hinge bolts. The electric control box includes: boom, motor system and pump station system; The auxiliary system includes an optical sensor mounting track and a probe mounting ring, which are fixed by magnetic attraction between a magnet and an adsorbent. The defect detection system includes a defect discrimination module and a multimodal data fusion unit. The defect discrimination module is used to perform the following steps: S100: Based on the initial acceleration signal collected by the accelerometer, the pile is subjected to low-strain defect detection to obtain the low-strain defect detection results. S200: Based on the initial ultrasonic signal collected by the ultrasonic transducer, ultrasonic defect detection is performed on the foundation pile to obtain the ultrasonic defect detection result. S300 performs optical defect detection on the foundation pile based on the initial optical signal collected by the optical sensor, and obtains the optical defect detection result; The multimodal data fusion unit is used to perform the following steps: The S400 integrates low-strain defect detection results, ultrasonic defect detection results, and optical defect detection results, and uses a priority classification and weighted calculation mechanism to comprehensively identify pile defects.

[0008] The present invention has at least the following beneficial effects: This invention provides a multimodal signal acquisition and defect detection system for pile integrity assessment. By employing low-strain detection, ultrasonic detection, and optical signal detection, it improves the accuracy of defect identification. Furthermore, it utilizes a Support Vector Machine (SVM) classification model to intelligently analyze the detection data, enhancing the accuracy of pile integrity judgment and promoting the transformation of pile detection technology from experience-dependent to data-driven. In summary, this invention proposes a multimodal signal acquisition and defect detection system for pile integrity assessment. The system comprises two parts: a signal acquisition device and a defect detection system. The device collects low-strain, ultrasonic, and crack detection data, while the defect detection system processes and analyzes the data. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the structure of the spiral support provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the optical sensor fixed track provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the magnetic suction surface of the probe fixing ring provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the bottom surface of the probe fixing ring provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the angle steel support frame provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the magnetic suction surface of the tenon and mortise splicing block of the probe fixing ring provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the bottom surface of the tenon and mortise joint block of the probe fixing ring provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of the mortise and tenon joint block for fixing the optical sensor track provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the magnet and adsorber structure provided in an embodiment of the present invention; Figure 10 This is a three-dimensional structural schematic diagram of the lifting system provided in an embodiment of the present invention; Figure 11 This is a top view of the lifting system provided in an embodiment of the present invention; Figure 12 A framework diagram of a defect detection system provided in an embodiment of the present invention; Figure 13 A flowchart for defect priority judgment provided in embodiments of the present invention; Symbol explanation: 1. Spiral support frame; 2. Optical sensor mounting track; 3. Probe mounting ring; 4. Scissor folding frame; 5. Electric control box; 6. Angle steel support frame; 7. Tray; 8. Hinge bolt; 9. Steel plate; 10. Probe mounting ring tenon and mortise joint; 11. Magnet; 12. Adsorber; 13. Optical sensor mounting track tenon and mortise joint. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] It should be noted that, based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Furthermore, this device and / or practice the method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.

[0013] The following section will introduce the multimodal signal acquisition and defect detection system used for pile integrity assessment: like Figures 1 to 11 As shown, there are four spiral supports 1, arranged at equal intervals around the pile. Both the spiral supports 1 and the optical sensor mounting track 2 have identical threaded surfaces, interlocking through protrusions and grooves in the threads to prevent slippage. The connection between the spiral supports 1 and the probe mounting ring 3 is similar. The optical sensor mounting track 2 and the probe mounting ring 3 are connected by magnetic attraction devices on their two contact surfaces. The probe mounting ring 3 has magnetic holes on its top surface and four pre-drilled holes on its bottom surface for clamping the accelerometer in low-strain detection. The optical sensor mounting track 2 has two parallel steel rails inside for easy assembly and fixation.

[0014] The scissor folding frame 4 and the electric control box 5 are supporting facilities. The electric control box 5 can operate the scissor folding frame 4 to lift and move. There are four lifting systems in this system, which are arranged at the center of the side length of the quadrilateral formed by the spiral upright frame 1. The angle steel support frame 6 is tightly connected to the spiral upright frame 1 by welding, so that the spiral upright frame 1 stands stably on the ground. The steel support plate 7 is connected to the scissor folding frame at the top layer of the scissor folding frame by hinge bolts 8.

[0015] The electric control box contains a drive system, a power transmission system, a control system, and safety devices, which control the automatic lifting of the scissor lift. The four lifting systems are controlled by the same operating system, enabling simultaneous lifting.

[0016] The signal acquisition device includes a support system, a lifting system, and an auxiliary system. The support system consists of a spiral support frame 1 and an angle steel support frame 6, which are connected and fixed by welding.

[0017] The lifting system consists of a scissor folding frame 4, a pallet 7, hinge bolts 8, and an electric control box 5. The scissor folding frame 4 is connected to the steel plate 9 by the hinge bolts 8, and the scissor folding frame 4 and the pallet 7 are also connected by the hinge bolts 8.

[0018] The electric control box 5 includes the boom, motor system, pump station system, etc., and functions as an electrically controlled lifting scissor folding frame. The auxiliary system consists of an optical sensor fixing track 2 and a probe fixing ring 3. The two are fixed together by the magnetic attraction between the magnet 11 at the top of the probe fixing ring 3 and the suction device 12 at the bottom of the optical sensor fixing track 2. They can be disassembled for operation as needed.

[0019] Preferably, the optical sensor fixing track 2 and the probe fixing ring 3 both adopt a mortise and tenon splicing design, which can be flexibly disassembled and is easy to carry.

[0020] Preferably, the spiral support 1 is designed with scale markings to monitor the lifting range in real time.

[0021] Preferably, the angle steel support frame 6 is designed as a three-way fixed structure, which greatly improves the stability of the spiral support frame without hindering the pile arrangement.

[0022] Preferably, the optical sensor fixing track 2 and the probe fixing ring 3 are designed to be magnetically connected, which can be flexibly assembled to realize one or both simultaneous detection of sound wave transmission and crack detection.

[0023] Preferably, the probe retaining ring 3 has a pre-drilled hole at its bottom. The diameter and depth of the hole meet the requirements for clamping low-strain sensors. During low-strain testing, the tops of the four sensors can be clamped in the hole at the bottom of the probe retaining ring. Therefore, the probe retaining ring can meet the requirements for clamping both ultrasonic sensors and low-strain sensors, achieving dual-purpose functionality.

[0024] Preferably, the optical sensor fixing track 2 and the probe fixing ring 3 are designed with a reserved notch structure. The notch curvature is the same as that of the spiral stand 1, so that the two can be completely spliced ​​together. Their outer surfaces have the same characteristics as the threaded surface of the spiral stand. The two surfaces are interlocked by the protrusions and grooves of the threads to prevent the components from sliding, while not hindering the lifting and lowering movement under external force.

[0025] The defect detection system includes a defect discrimination module and a multimodal data fusion unit. The defect discrimination module is used to perform the following steps: S100 performs low-strain defect detection on the foundation pile based on the initial acceleration signal collected by the accelerometer, and obtains the low-strain defect detection results.

[0026] S200 performs ultrasonic defect detection on the foundation pile based on the initial ultrasonic signal collected by the ultrasonic transducer, and obtains the ultrasonic defect detection results.

[0027] S300 performs optical defect detection on the foundation pile based on the initial optical signal collected by the optical sensor, and obtains the optical defect detection result.

[0028] The multimodal data fusion unit is used to perform the following steps: The S400 integrates low-strain defect detection results, ultrasonic defect detection results, and optical defect detection results, and uses a priority classification and weighted calculation mechanism to comprehensively identify pile defects.

[0029] Furthermore, step S100 includes the following steps: S110 collects the voltage signals corresponding to the unexcited background signal and the initial acceleration signal after excitation, and performs anomaly detection on the voltage signal fluctuation, voltage threshold and noise level.

[0030] S120 divides the pile body into 30 equal segments, performs Fourier transform on the signal, calculates the RMS value of each segment, and normalizes several parameters of the pile with the RMS value.

[0031] S130, the defect level is determined based on the minimum RMS value and the change rate of adjacent segments, so as to obtain the location of the defect segment corresponding to the low strain defect detection.

[0032] Furthermore, step S200 includes the following steps: S210 collects ultrasonic signals before and after excitation and converts them into electrical signals for sensor, voltage, and signal anomaly detection.

[0033] S220 divides the pile body into 30 equal segments, arranges the test profiles according to the specifications, calculates the ultrasonic signal energy and mean energy of each test profile segment after Fourier transform of the signal, as well as the minimum and maximum energy ratio, and normalizes several parameters and energy characteristics of the foundation pile.

[0034] S230, the defect level is determined based on the ratio of the average energy value to the average energy value of the complete pile and the energy ratio, so as to obtain the location of the defect segment corresponding to the ultrasonic defect detection.

[0035] Furthermore, step S300 includes the following steps: S310 acquires optical signals from the surface of the pile segment through an optical sensor and converts them into voltage signals. After Gaussian filtering to remove noise and histogram equalization to enhance crack features, it acquires background images and calculates the difference in light intensity through adaptive binarization.

[0036] S320 uses the Canny operator to extract the crack edge contour, uses the Hough transform to calculate the maximum crack width in the current frame, filters the maximum crack width of each pile segment and normalizes it.

[0037] S330 determines the defect level based on the ratio of the crack width to the maximum value in the database, thus obtaining the location of the pile segment with the largest crack.

[0038] Furthermore, step S400 includes the following steps: S410: First mark the severe defects identified by low strain or ultrasonic testing, and the obvious defective pile segments identified by both.

[0039] For S420, the weighted score is calculated with a low strain weight of 0.4 and an ultrasonic weight of 0.6 for the other cases.

[0040] S430, determine the integrity level of the entire pile and locate the defect based on the weighted score.

[0041] Furthermore, in step S420, when calculating the weighted score, if there are defects in adjacent pile segments, the compensation coefficient is increased to 1.2. The same rule is used for exposed pile segments: crack weight 0.4 and ultrasonic weight 0.6. Finally, the defect location is locked and the integrity level of the entire pile is determined based on the weighted score.

[0042] like Figure 12 and Figure 13 As shown, the defect detection system mainly consists of three types of sensors and five modules. The three types of sensors are: an optical sensor, an accelerometer, and an ultrasonic transducer; the five modules are: a signal acquisition module, an anomaly detection module, a feature extraction module, a defect discrimination module, and a parameter input module.

[0043] The signal acquisition module includes image detection, low strain detection, ultrasonic detection, analog-to-digital conversion unit, and function selection unit.

[0044] The anomaly detection module includes a sensor detection unit, a voltage detection unit, a signal anomaly detection unit, and a fault indication unit.

[0045] The feature extraction module includes a data preprocessing unit and a feature extraction unit.

[0046] The defect identification module uses the collected and processed data to train and predict the SVM model, and includes a low strain identification unit, an ultrasonic identification unit, and a crack identification unit. The parameter input module includes a field parameter input unit and a test result visualization unit.

[0047] The functions of each module and unit are as follows: Optical sensors are used for crack width measurement. They capture and convert optical signals, then convert them into electrical signals for subsequent processing and analysis. This technology enables high-precision, non-contact measurement.

[0048] Accelerometers are used for low-strain detection. By measuring the vibration response of the pile body after hammer excitation and analyzing the reflected waveform, it can determine whether there are defects in the pile body.

[0049] Ultrasonic transducers analyze the acoustic signal characteristics of sound wave propagation by transmitting and receiving ultrasonic signals, thereby identifying defects in the pile body.

[0050] The signal acquisition module includes optical detection, low-strain detection, ultrasonic detection, an analog-to-digital converter (ADC) unit, and a function selection unit. The function selection unit allows selection of different detection functions, and the ADC unit converts the signals acquired by each sensor into discrete digital signals. Data from different detection units can be combined and analyzed as needed. The corresponding detection data can be selected through the function selection unit and used as a multi-source criterion for pile defects in the subsequent defect discrimination module.

[0051] The sensor detection unit of the anomaly detection module is used to detect whether image information is acquired normally and whether the sensor is activated normally. The voltage detection unit checks whether the input signal triggers the acquisition threshold and can acquire data normally; the signal anomaly detection unit detects whether the signal is too weak, lost, or drifting. The fault indication unit provides corresponding error codes and displays fault information in the visualization unit of the parameter input module.

[0052] The feature extraction module includes a data preprocessing unit and a feature extraction unit.

[0053] Data preprocessing unit: The data preprocessing unit includes three types: low strain data preprocessing, ultrasonic data preprocessing, and image data preprocessing.

[0054] Low-strain data: Wavelet filtering removes signal noise and enhances its characteristics.

[0055] Ultrasonic data: Wavelet filtering is used to remove signal noise and enhance its characteristics. A high-pass filter is used to remove the trend term from the signal.

[0056] Image data: High-pass filtering is used to remove noise and enhance image features; edge detection algorithms are used to enhance image edge features. Image pixels are normalized to the range [0,1] and cropped into regions of interest (ROIs).

[0057] Feature extraction unit: The feature extraction unit primarily transforms the signal into physical features usable for analysis, which are then used for subsequent defect assessment. For low-strain signals, its RMS value is calculated. ; Calculating the energy of ultrasonic signals For optical signals, calculate the crack width. .

[0058] The defect discrimination module uses the Support Vector Machine (SVM) algorithm to discriminate the extracted feature vectors, establishing a defect discrimination model for the foundation piles and determining the defect category and approximate location. The SVM model is trained on a remote server using a large amount of balanced sample data obtained from a previous foundation pile inspection database.

[0059] The parameter input module includes a field parameter input unit and a test result visualization unit. The field parameter input unit provides an interface for inputting parameters such as sampling frequency, pile length, pile diameter, concrete grade, and pile number, which are used for subsequent analysis and calculation of pile defects. The visualization unit provides a user interface for real-time monitoring of pile defects and querying test parameters, fault error information, historical data, and test reports.

[0060] Signal acquisition device: (1) Assemble the probe fixing ring 3 along the pile diameter direction. The specific method is as follows: Connect the tenon and mortise splicing blocks 10 of the two probe fixing rings around the pile to be measured. The connection should make the tenon and mortise correspond to each other. In this way, splice the rings around the pile diameter in sequence. A total of 4 splicings are required. After splicing, the probe fixing ring 3 can fit around the pile to be measured.

[0061] (2) Assemble the optical sensor fixing track 2 along the pile diameter. The specific method is as follows: Install the magnet 11 in the groove on the cross-section of the top ring of the probe fixing ring 3. A total of 8 magnets need to be installed. Similarly, install the adsorbent 12 in the groove on the cross-section of the bottom ring of the optical sensor fixing track 2. Similar to step (1), connect the optical sensor fixing track tenon and mortise splicing blocks 13 around the pile to be tested. When splicing, the magnet 11 and the adsorbent 12 should be aligned so that the two ring structures are tightly connected. Splice the ring around the pile diameter in this way. A total of 4 splicings are required. The spliced ​​optical sensor fixing track 2 and the probe fixing ring 3 form a ring-shaped whole. The two fit together on the pile to be tested (for two testing requirements, they can be disassembled for single testing).

[0062] (3) Arrange the four spiral supports 1 evenly around the pile to be measured. The four arrangement points at the bottom of the spiral supports 1 form a square plane, and each pair of adjacent spiral supports 1 forms a square side length. Then assemble the support system and auxiliary system: make the spiral supports 1 fit into the reserved gaps of the probe fixing ring 3 and the optical sensor fixing track 2 around the pile to be measured. A total of 4 assembly steps are required. After assembly, fine-tune the position to ensure that the acoustic optical sensor is in the detection position.

[0063] (4) Arrange the four scissor folding frames 4 at the midpoint of each side of the square above, and stably place the probe fixing ring 3 and the optical sensor fixing track 2 on the support plate 7 on the top of the scissor folding frame 4 to facilitate subsequent lifting and detection.

[0064] Defect detection system: Input parameters such as sampling frequency, pile length, pile diameter, concrete grade, and pile number into the parameter input module. Select different detection functions (low-strain testing, ultrasonic testing, crack detection) according to the pile testing requirements. After setting the initial parameters, install and fix the corresponding sensors in the signal acquisition device to prepare for subsequent data acquisition and analysis.

[0065] The specific implementation methods of each detection module are as follows: Low strain testing: 1) Acquire the initial signal.

[0066] After the detection equipment is turned on, signal acquisition is divided into two stages: (a) The background signal continuously collected after the device is turned on is the signal when it is not excited (including environmental noise, sensor zero drift, etc.), which is converted into a voltage signal vpre_a=[va1…vap], where p is the number of pre-signals.

[0067] (ii) A vertical excitation force is applied at the detection point. At the start of the hammering, the initial form of the low-strain signal is the vibration time-domain signal transmitted from the pile top after the hammering. When the signal exceeds the trigger threshold, the vibration simulation voltage signal is collected. After excitation, the signal exceeding the signal trigger threshold is collected. The trigger signal at this time corresponds to the voltage signal (including effective reflected waves and noise) obtained after the initial collected signal is converted from analog to digital (A / D).

[0068] 2) Signal anomaly detection.

[0069] If the unexcited signal is abnormal (e.g., sensor failure), subsequent excitation signals are not worth analyzing and do not require further analysis. The automated anomaly detection designed in this invention can quickly filter invalid data, reducing manual review time. This module detects the acquired voltage signals, checks for outliers, and ensures the validity of subsequent data processing.

[0070] Abnormal Scenario 1: Abnormal fluctuation signal (background signal before excitation) Possible manifestations include signal values ​​approaching zero, signal drift, significant instability, signal amplitude remaining at zero, and irregular, large fluctuations in the signal.

[0071] Abnormal Scenario 2: Abnormal Output Voltage The main detection target is vpre_a=[va1…vap], specifically whether the input signal exceeds the signal trigger threshold and whether the vibration sensor output voltage exceeds the threshold.

[0072] Abnormal Scenario 3: Abnormal Noise Level The primary detection targets are va = [va1…van]. Noise pollution is severe, and the signal-to-noise ratio is below the threshold. The signal is severely interfered with, exhibiting a significantly abnormal range of power frequency.

[0073] Test Results and Measures: If the above situations occur, it indicates poor contact or damage to the sensor or circuit, abnormal environmental noise, or improper vibration application. In this case, the fault indication function will be triggered, and the user will be prompted through the visual module to troubleshoot the problem and re-enter the signal.

[0074] If the signal does not have the above problems, proceed to step 3).

[0075] 3) Signal preprocessing and feature extraction.

[0076] The pile section is divided into 30 segments, pile_section=30, and the length of each segment is delta_L=L / 30. The total amount of data collected is n, and the number of data points collected for each pile segment is n / 30.

[0077] 3.1 Calculate the number of data points The acquired time-domain signal (acceleration signal, i.e., continuous signal) is subjected to a Fourier transform (FFT) to obtain the frequency-domain signal, which facilitates subsequent calculation of the RMS value. ; In the formula: X(f) is the frequency domain signal, representing the complex amplitude of the signal at frequency f; x(t) is the time domain signal, representing the amplitude of the signal at time t; f is the frequency, in Hz.

[0078] Calculate the total time from the moment the hammer strikes until the vibration wave travels to the bottom of the pile. t ; In the formula: L is the pile length, and c is the wave velocity; both are parameters input by the human-computer interaction module.

[0079] The number of low-strain data points obtained throughout the entire data acquisition process is calculated using the propagation time t and the acquisition frequency f. This number is equal to the propagation time multiplied by the acquisition frequency.

[0080] 3.2 Calculate the RMS value The root mean square (RMS) value of low-strain testing of foundation piles is a statistical measure of signal amplitude, reflecting the strength of signal energy, which is closely related to the integrity of the foundation pile. The RMS value of an intact pile decreases with increasing waveform propagation depth, and the overall RMS value distribution is relatively uniform. Abnormally strong or weak RMS values ​​in defective piles may be due to wave reflection at the defect location, causing abrupt waveform changes.

[0081] Based on the frequency domain signal, calculate the RMS value of each pile segment (i.e., the signal segment): ; In the formula: k is the pile segment number, and is the amplitude value of the signal at each time point. N It represents the total number of signal sampling points.

[0082] 3.3 Eigenvalue Normalization To eliminate the influence of different feature numerical scales and avoid imbalance of feature weights during SVM training, maximum value normalization is performed for each dataset to ensure that all values ​​in the feature vector are within (0, 1).

[0083] The pile length (pile_length), pile diameter (pile_diameter), wave velocity (wave_velosity), and concrete grade (concrete_grade) input from the parameter input module are normalized along with the calculated RMS values. All data are normalized using the maximum value within each dataset.

[0084] For example: Normalized value of data X ; The normalized low-strain eigenvector is X=[pile_length, pile_diameter, wave_velosity, concrete_id, RMS1, RMS2,…RMS 30 ].

[0085] 4) Identify defects in the pile body.

[0086] The piles are divided into 30 sections, RMS i This represents the energy value of the i-th segment. The segment with the smallest energy value among the 30 segments is indexed, and its energy value (RMS) is used as the reference value. min One of the criteria for judging pile defects: i min =argmin(RMS i ) To improve accuracy and avoid errors caused by a single criterion, the RMS rate of change between two adjacent pile segments is calculated based on the eigenvectors: dRMS i =RMS i -RMS i-1 .

[0087] Introducing the RMS change rate between two adjacent pile segments as a second criterion, the following pile defect judgment index is established: The criteria for judging the integrity of each segment pile are as follows: Complete: RMS min >0.7; dRMS i >-0.05 Minor defects: 0.5 < RMS min <0.7; -0.1<dRMS i <-0.05 Obvious defects: 0.3 < RMS min <0.5; -0.2<dRMS i <-0.1 Critical Defect: RMS min <0.3; dRMS i <-0.2 Based on the above four categories, the integrity of the entire pile is determined. Segments with minor, obvious, or serious defects are selected from all pile segments. If a pile segment has a serious defect, it is classified as a seriously defective pile and assigned the label Y=4. If a pile segment definitely has an obvious defect, may have a minor defect, but has no serious defects, it is classified as an obviously defective pile and assigned the label Y=3. If a pile segment has a minor defect, but has no serious or obvious defects, it is classified as a slightly defective pile and assigned the label Y=2. If all pile segments are complete and there are no other types of pile segments, it is classified as a complete pile.

[0088] 5) Calculation of the location of defects in the pile body.

[0089] Based on the defect assessment value, the defect location is located. min The defect range is The center location of this pile segment is: Ultrasonic testing: 1) Acquire the initial signal.

[0090] The initial signal collected includes two parts: the unexcited signal and the voltage signal obtained after analog-to-digital conversion following excitation. The excitation signal is an acoustic vibration signal, and the ultrasonic transducer converts the mechanical vibration signal into an electrical signal.

[0091] 2) Signal anomaly detection.

[0092] The content of signal anomaly detection is the same as that for low strain.

[0093] 3) Signal preprocessing and feature extraction.

[0094] The ultrasonic signal after excitation is a time-domain acoustic signal obtained by the transducer's transmission and reception. The receiving circuit amplifies the electrical signal and transmits it to the data acquisition system, converting it into a digital signal. After Fourier transform (FFT), the frequency domain signal is obtained.

[0095] 3.1 Calculate the number of signals An ultrasonic testing surface is placed at the center of each small pile segment. The ultrasonic signal from each testing surface represents the ultrasonic characteristics of that pile segment. Measuring points are placed on each testing surface, and the ultrasonic waves propagate along the path between two measuring points; this path constitutes the testing profile. The number of testing profiles is arranged according to the standard "Technical Specification for Testing of Building Foundation Piles" (JGJ 106-2014): for pile diameters less than 1000 mm, two pipes are buried, forming one profile; for pile diameters between 1000-1500 mm, three pipes are buried, forming three profiles; and for larger pile diameters, four pipes may be buried, forming six profiles. The following explanation uses four testing profiles per testing surface.

[0096] Similar to the segmented low-strain testing, the pile is divided into 30 segments. Ultrasonic testing requires collecting acoustic parameters segment by segment at intervals of 10-20 cm. Based on specifications and the actual pile length, m testing profiles are set up within each pile segment. The distance between two profiles is L / (30m), where m satisfies 0.1 < L / (30m) < 0.2. A total of 180m signals are collected along the entire pile length. The sampling frequency set for ultrasonic testing determines the time resolution of the measuring points; setting an appropriate sampling frequency ensures normal waveform acquisition.

[0097] Next, the ultrasonic signal energy and average energy of all survey lines on each test profile of each of the 30 pile segments were calculated. Based on the frequency domain signal, the ultrasonic signal energy value of the k-th detection profile of the i-th pile segment is: In the formula: i is the pile segment number, i=1~30; k is the inspection profile number within that pile segment, k=1~m; x r is the amplitude of the ultrasonic signal on the r-th acoustic measurement line. For the 1st to 6th measurement lines, r = 1 to 6. N is the total number of measurement lines in the detection profile, N = 6.

[0098] The average energy of each measuring line in the k-th detection profile of the i-th pile segment: (E) min / E max ) ik Let be the ratio of the minimum energy to the maximum energy of all measuring lines in the k-th detection profile of the i-th pile segment, where i = 1~30 and k = 1~m.

[0099] 3.2 Eigenvalue Normalization Each segment of ultrasonic signal is filtered and denoised, and combined with the on-site input parameters: pile length, pile diameter, wave velocity, and concrete grade, a (60m+4) dimensional feature vector is formed. The number of segmented pile detection profiles (m) and the number of measuring lines per profile (m) need to be set in advance in the function selection unit of the signal acquisition module and the ultrasonic detection unit.

[0100] The maximum value in each dataset is used for normalization, ensuring that all values ​​in the feature vector fall within the range [0, 1]. This normalization method is the same as that used in the low-strain detection feature extraction module. The normalized feature vector is X = [pile_length, pile_diameter, wave_velosity, concrete_id, E...]. 11a E 12a E 13a E 1（m-1）a E 1ma E 21a E 22a …E 30ma (E) min / E max ) 11 ,(E min / E max ) 12 ,(E min / E max )13 …(E) min / E max ) 1（m-1） ,(E min / E max ) 1m …(E) min / E max ) 30m ] 4) Determine defects in the pile body.

[0101] The feature vectors are input into the SVM training model to determine pile defects.

[0102] Energy mean E ika Let E be the average energy of the ultrasonic signals from each acoustic logging line at the k-th detection profile of the i-th pile segment, representing the integrity of that pile segment. The amplitude of the ultrasonic signal attenuates when it encounters a defect. ia Consequently, it decreases; to avoid judgment errors caused by extreme unevenness in energy values ​​and to reduce the false judgment rate, this invention patent also uses E min / E max As a defect criterion, this value represents the ratio of the minimum to the maximum energy of all acoustic logging lines in the k-th detection profile of the i-th pile segment. It indicates the non-uniformity of signal energy. The smaller the value, the more severe the energy attenuation of a certain logging line in the detection profile, and the greater the possibility of a defect in that pile segment. The average energy E of each detection profile of a complete pile is... norm The average energy of each detection profile of the complete pile segment, obtained based on previous data, is obtained from a pre-trained SVM model on a remote platform.

[0103] The pile is divided into 30 sections, E ika (E) min / E max ) ik To determine the indicators of defects in each pile segment, a discrimination result is obtained based on these indicators. Therefore, all data used is the average energy E of the k-th detection profile of the i-th pile segment. ika The ratio of minimum to maximum energy (E) min / E max ) ik The average energy E of a complete pile norm .

[0104] The ultrasonic testing criteria for determining the integrity of each segmented pile are as follows: Complete: E ika >0.92E norm (E) min / E max ) ik >0.85 Minor defects: 0.72 E norm <E ika<0.92 E norm 0.65 < (E min / E max ) ik <0.85 Obvious defect: 0.55 E norm <E ika <0.72 E norm 0.38 < (E min / E max ) ik <0.65 Critical defect: 0.25 E norm <E ika ≤0.55E norm 0.15 < (E min / E max ) ik ≤0.38 Based on the above four categories, the integrity of the entire pile is determined. Segments with minor, obvious, and serious defects are screened. If a segment with serious defects is found, the pile is classified as a seriously defective pile and assigned the label Y=4. If a segment with obvious defects is found, and possibly some segments with minor defects but no segments with serious defects, the pile is classified as an obviously defective pile and assigned the label Y=3. If a segment with minor defects is found, and no segments with serious or obvious defects are found, the pile is classified as a slightly defective pile and assigned the label Y=2. If all segments are complete and there are no other types of segments, the pile is classified as a complete pile.

[0105] 5) Calculation of the location of defects in the pile body.

[0106] According to E ika Discriminant and (E) min / E max ) ik The judgment result pinpoints the defect location. d The defect range is The center location of this pile segment is: Crack detection: 1) Acquire the initial signal.

[0107] Optical imaging sensors acquire images using a non-contact method, and the acquisition of complete signals mainly depends on lighting conditions and the stability of the imaging system. Optical imaging sensors acquire optical signals from the surface of the measured pile segment, which are then converted into voltage signals via a special circuit.

[0108] 2) Signal anomaly detection.

[0109] The content of signal anomaly detection is the same as that for low strain.

[0110] 3) Signal preprocessing and feature extraction.

[0111] 3.1 Data Preprocessing Image preprocessing is used to improve crack contrast, Gaussian filtering is used to remove noise, and histogram equalization is used to enhance crack features.

[0112] 3.2 Crack Identification and Image Acquisition Before acquiring the crack image, first acquire a background image without cracks. I bg (x, y). After formal data acquisition, based on the changes in light intensity, brightness, and contrast at the crack, an adaptive binarization (Otsu thresholding) method is used to acquire images of the crack area. By calculating the difference between the real-time image and the background image, it is determined whether a crack exists in the frame and whether it needs to be acquired. The determination formula is as follows: In the formula, This is a light intensity difference image, reflecting pixel information about changes in light intensity. A threshold T is set to determine whether the current frame contains cracks. If... If so, then record the image.

[0113] 3.3 Crack Maximum Width Frame Determination To optimize the optical crack detection process and ensure that the final output image frame contains the maximum crack width, a real-time filtering step for the maximum crack width is added to the existing adaptive trigger acquisition.

[0114] Extract the current frame using the Canny operator. I t Crack edge profile at (x, y): In the formula: E t This is the crack edge image for the current frame. I t This refers to the currently acquired image information.

[0115] The maximum crack width of the current frame is calculated using the Hough transform. In the formula: W t Indicates the crack edge image E in the current frame. t The maximum crack width is calculated in the formula, where d(x,y) represents the local width of the crack.

[0116] 3.4 Filter the maximum crack width of the current pile segment If the current crack width W t Greater than the recorded maximum valueW max Then update the maximum crack width W. max =W t Update the image I corresponding to the maximum crack width. max =I t Otherwise, do not update, and continue processing the next frame until all frames have been processed. After all acquisition processes, the final I... max The output corresponds to the widest crack location in the i-th pile segment. I i .

[0117] Data was collected for each of the 30 pile segments using the above method, and the maximum crack width of the 30 segments was recorded and stored. W =[ W 1, W 2, W 3,…… W i (i=1~30) 3.5 Crack width normalization The pile length, pile diameter, wave velocity, concrete grade, and crack width are normalized using the maximum value method to ensure that the crack width of each segment is within the range of [0,1].

[0118] In the formula: W is the original crack width value, which is the maximum value among the 30 crack width datasets.

[0119] The normalized feature vector X = [pile_length, pile_diameter, wave_velosity, concrete_id, w1, w2, ... w 30 ] 4) Identify defects in the pile body.

[0120] The criteria for judging the integrity (cracks and defects) of each segment pile are as follows: whole: Minor defects: Obvious defects: Critical defects: In the formula, w i This represents the maximum width of the crack in the i-th pile segment. w max This represents the maximum crack width in the pile crack detection database.

[0121] Based on the above four categories, the integrity of the entire pile is determined. Segments with minor, obvious, or serious defects are selected from all pile segments. If a pile segment has a serious defect, it is classified as a seriously defective pile and assigned the label Y=4. If a pile segment definitely has an obvious defect, may have a minor defect, but has no serious defects, it is classified as an obviously defective pile and assigned the label Y=3. If a pile segment has a minor defect, but has no serious or obvious defects, it is classified as a slightly defective pile and assigned the label Y=2. If all pile segments are complete and there are no other types of pile segments, it is classified as a complete pile.

[0122] 5) Calculation of the location of defects in the pile body.

[0123] Output the index of the pile segment where the maximum crack is located. i max The location corresponding to the maximum crack width l d The calculation method is as follows: i max =argmax(W i ) SVM model training: The data is trained and partitioned based on a Support Vector Machine (SVM) model, and the specific implementation is as follows: (1) Collect low-strain, ultrasonic, and crack detection data and establish datasets. Collect corresponding types of detection data based on engineering data and survey reports accumulated over the years, extract key information such as pile type, pile diameter, pile length, concrete number, wave velocity, and stratum conditions, as well as pile integrity detection results, and standardize and structure the raw data. According to provinces, cities, regions, and engineering projects, establish high-quality training datasets for low-strain, ultrasonic, and crack detection respectively.

[0124] (2) Extract feature vectors. The extraction methods and steps are the same as those used in low strain, ultrasonic, and crack detection.

[0125] (3) The high-dimensional nonlinear spatial data obtained from pile defect detection is mapped using the RBF (Radial Basis Function) dataset, with the regularization parameter C and the Gaussian kernel function parameter σ set. The RBF kernel function classification formula used is as follows: In the formula: x i , x j There are two sample points; σ is the standard deviation of the Gaussian distribution; (4) Train and optimize the SVM model. Use the dataset established in (1) as the training set and new engineering data as the test set to train the initial vector machine. The training method is the same as steps 1) to 5) of the low strain, ultrasonic, and crack detection steps. Optimize the RBF kernel hyperparameters using grid search and cross-validation methods to improve the accuracy of the SVM model.

[0126] The Lagrangian function used to solve optimization problems is: In the formula: α i μ i These are the Lagrange multipliers corresponding to the constraints; y i Represents training data; w It is the normal vector of the hyperplane; b It is a bias term; ξ i It is a slack variable; C It is the regularization parameter.

[0127] The SVM model is successfully trained when the cross-entropy loss between the training and test sets reaches a standard threshold. It can then be used for classifying new samples.

[0128] Multimodal data fusion unit: Pile crack detection is a visual inspection method suitable for inspecting exposed sections of the pile. Since stress wave transmission depends on reflection at the pile-soil interface, low-strain testing is limited to the embedded section of the pile. Ultrasonic testing actively transmits and receives signals via a transmitter and receiver, unaffected by the surrounding soil. It can detect defects in both exposed and embedded sections of the pile using acoustic logging tubes.

[0129] Multimodal data fusion, combining low-strain testing, ultrasonic testing, and crack detection, can significantly improve the accuracy and reliability of defect identification in both embedded and exposed sections of the pile. For exposed sections, crack detection identifies surface cracks, while ultrasonic testing identifies internal defects. For embedded sections, low-strain combined with ultrasonic testing can be used. Multimodal data analysis provides added assurance for pile defect detection. The multimodal data fusion unit is a crucial part of the defect discrimination module, performing fusion analysis based on the detected data. The specific implementation is as follows: A defect identification standard is established by adopting a priority classification principle.

[0130] The results of low-strain testing and ultrasonic testing are divided into four levels: no defects, minor defects, obvious defects, and serious defects.

[0131] The possible judgment results and judgment process are as follows: Step 1 (Prioritize critical defects and consistent obvious defects): In the case of combined low-strain testing and ultrasonic testing, both methods have already determined the integrity of each pile segment, and the judgment in step one is carried out based on the results.

[0132] The detection method may detect more than one defect; therefore, a loop algorithm is introduced in this unit to index segment by segment until all defect locations are output. See the attached flowchart for the loop algorithm. Figure 13 As shown in the flowchart, if either of the two detection methods shows a serious defect, it is marked as a serious defect, and the defective pile segment and its location are output. If both detection methods show that the pile segment has a significant defect, it is marked as a consistently significant defect, and the defective pile segment and its location are output. These two cases represent the highest defect risk, so they are prioritized and marked accordingly. For all other cases, a weighted calculation is performed; see Case 2 for details.

[0133] After executing the flowchart, if both the initial and final results indicate a serious defect, the defect location A is located using low-strain testing, and the defect location B is located using ultrasonic testing. If A and B coincide, it indicates a very high risk of a serious defect at that location; if A and B do not coincide, the coordinates of A and B are output separately, and the test results (including the pile segment number and defect location) are output in the parameter input module. Other testing methods (such as core sampling) are required to determine the pile defect.

[0134] Step Two (Perform Weighted Calculation): This approach applies to situations other than severe defects and consistently obvious defects. Cracks are surface defects, and minor cracks do not affect the internal integrity of the pile. Similarly, low-strain and ultrasonic testing of internal pile defects may not reveal them on the outer surface. Therefore, the results of both methods are weighted and included in the overall judgment.

[0135] Scores of W=0, W=1, and W=2 are assigned to no defects, minor defects, and significant defects, respectively, and a weighted score is calculated for each pile segment i (i=1, 2, ..., 30). The low-strain test result is W. l The weight is w l =0.4, the ultrasonic test result is W u The weight is w u =0.6; the overall score is calculated by weighting both methods. If both detection methods report defects in the same pile segment or adjacent pile segments, the weighted score is increased by 10%. The calculation formula is as follows: Total weight= β ˙( W l ˙ w l + W u ˙ w u ) In the formula: W l W u For low-strain, ultrasonic testing results, no defects W=0, minor defects W=1, and significant defects W=2. The weight for low strain is w. l =0.4, the ultrasonic weight is w u =0.6. β This is the compensation coefficient for adjacent pile segments. I The value is 1.2 if a defect exists in the adjacent ±1 segment, otherwise it is 1.

[0136] Determine the integrity of each segment of the pile: No defects: Total weight = 0.0 Minor defects: 0.4 ≤ Total weight ≤ 0.6, 1.0 ≤ Total weight ≤ 1.1 The most significant flaw in this debate is that 0.8 ≤ Total weight ≤ 0.9. Obvious defect: 1.2 ≤ Total weight ≤ 1.8 A disputed obvious defect refers to a situation where a significant defect is detected by low strain testing but no defect is detected by ultrasonic testing. In this case, a disputed obvious defect is indicated, and the equipment calibration process is initiated. Other methods (such as core sampling) can be used to verify the test results.

[0137] For the inspection of exposed sections of the pile body, a combination of crack detection and ultrasonic testing is used, where crack detection corresponds to W. c The weight is w c =0.4, the ultrasonic test result is W u The weight is w u =0.6; the overall score is calculated by weighting both. The specific implementation method is the same as the low strain testing + ultrasonic testing described above.

[0138] Based on the above four categories, the integrity of the entire pile is determined. All pile segments are then screened for defects: no defects, minor defects, disputed obvious defects, and obvious defects. If an obvious defective pile segment is found, it is classified as an obviously defective pile and labeled Y=4. If a disputed obvious defective pile segment is definitely present, a minor defective pile segment may be present, but no obvious defective pile segment is present, it is classified as a disputed obvious defective pile and labeled Y=3. If a minor defective pile segment is present, but no disputed obvious defects or obvious defects are present, it is classified as a minor defective pile and labeled Y=2. If all pile segments are complete and there are no other types of pile segments, it is classified as a complete pile.

[0139] Calculation of pile defect location: After determining the pile defect, output the defective pile segment number and the defect location.

[0140] In this embodiment, multimodal signal acquisition through low-strain detection, ultrasonic detection, and optical signal detection improves the accuracy of defect identification. Furthermore, a Support Vector Machine (SVM) classification model is used to intelligently analyze the detection data, enhancing the accuracy of pile integrity assessment and promoting the transformation of pile detection technology from "experience-dependent" to "data-driven." In summary, this invention proposes a multimodal signal acquisition and defect detection system for pile integrity assessment. This system comprises two parts: a signal acquisition device and a defect detection system. The device part collects low-strain, ultrasonic, and crack detection data, providing accurate data for the subsequent defect detection system.

[0141] This embodiment provides a signal acquisition device that integrates low-strain detection, ultrasonic testing, and crack detection for foundation piles. Two annular auxiliary detection devices are supported by a spiral frame. The auxiliary devices are connected by mortise and tenon joints, allowing for flexible disassembly of components. An electric lifting system enables the auxiliary devices to be raised and lowered synchronously within the height range of the spiral frame, achieving automatic lifting and positioning of the foundation pile at different heights. This solves the problem of inaccurate positioning that cannot be guaranteed by manually handheld optical sensors. Furthermore, using this detection device saves labor and avoids the safety hazards associated with working at heights.

[0142] Existing pile foundation testing technologies, including low-strain, ultrasonic, and crack detection technologies, often employ single methods and simple weighting during data fusion. Conflicting pile defect findings primarily rely on manual judgment. This invention's pile defect detection system proposes a two-tier arbitration mechanism, enabling single-item testing while prioritizing severe defects during data fusion, followed by weighted processing based on defect contribution. This invention addresses the shortcomings of existing pile foundation testing equipment and the limitations of single-method testing. It includes a self-developed sensor fixing device, solving the technical problem of unstable sensor coupling in traditional testing. Based on accurate data acquisition, it proposes a pile defect detection system employing an SVM-based data fusion algorithm, trained and optimized with extensive engineering data, significantly reducing the human error rate. This invention standardizes and automates the testing process, improving the convenience and reliability of pile defect detection.

[0143] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0144] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.

Claims

1. A multimodal signal acquisition and defect detection system for foundation pile integrity assessment, characterized in that, The system includes: a signal acquisition device and a defect detection system; The signal acquisition device is used to install accelerometers, ultrasonic transducers, or optical sensors. The signal acquisition device includes a support system, a lifting system, and an auxiliary system. The support system includes a spiral support frame and an angle steel support frame, which are connected and fixed by welding. The lifting system includes: a scissor-type folding frame, a pallet, hinge bolts, and an electric control box. The scissor-type folding frame is constructed by connecting steel plates with hinge bolts, and the scissor-type folding frame and the pallet are connected by hinge bolts. The electric control box includes: boom, motor system and pump station system. The auxiliary system includes optical sensor fixing rail and probe fixing ring. The optical sensor fixing rail and probe fixing ring are fixed by the magnetic attraction generated between the magnet and the adsorber. The defect detection system includes a defect discrimination module and a multimodal data fusion unit. The defect discrimination module is used to perform the following steps: S100: Based on the initial acceleration signal collected by the accelerometer, the pile is subjected to low-strain defect detection to obtain the low-strain defect detection results. S200: Based on the initial ultrasonic signal collected by the ultrasonic transducer, ultrasonic defect detection is performed on the foundation pile to obtain the ultrasonic defect detection result. S300 performs optical defect detection on the foundation pile based on the initial optical signal collected by the optical sensor, and obtains the optical defect detection result; The multimodal data fusion unit is used to perform the following steps: The S400 integrates low-strain defect detection results, ultrasonic defect detection results, and optical defect detection results, and uses a priority classification and weighted calculation mechanism to comprehensively identify pile defects.

2. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, Step S100 includes the following steps: S110 collects voltage signals corresponding to the unexcited background signal and the initial acceleration signal after excitation, and performs anomaly detection on voltage signal fluctuation, voltage threshold and noise level; S120, the pile body is divided into 30 equal segments, the signal is Fourier transformed and the RMS value of each segment is calculated, and several parameters of the pile are normalized with the RMS value. S130, the defect level is determined based on the minimum RMS value and the change rate of adjacent segments, and the location of the defect segment corresponding to the low strain defect detection is obtained.

3. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, Step S200 includes the following steps: S210 collects ultrasonic signals before and after excitation and converts them into electrical signals for sensor, voltage and signal anomaly detection. S220 divides the pile body into 30 equal segments, arranges the test profiles according to the specifications, calculates the ultrasonic signal energy and energy mean of each test profile segment after Fourier transform of the signal, as well as the minimum and maximum energy ratio, and normalizes several parameters and energy characteristics of the foundation pile. S230, the defect level is determined based on the ratio of the average energy value to the average energy value of the complete pile and the energy ratio, so as to obtain the location of the defect segment corresponding to the ultrasonic defect detection.

4. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, Step S300 includes the following steps: S310 acquires optical signals from the surface of the pile segment through an optical sensor and converts them into voltage signals. After Gaussian filtering to remove noise and histogram equalization to enhance crack features, it acquires background images and calculates the difference in light intensity through adaptive binarization. S320: The Canny operator is used to extract the crack edge contour, the Hough transform is used to calculate the maximum crack width of the current frame, the maximum crack width of each pile segment is screened and normalized. S330: Defect level is determined based on the relative size of the crack width and the maximum crack width in the database, so as to obtain the location of the pile segment with the largest crack.

5. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, Step S400 includes the following steps: S410, first mark the serious defects determined by low strain or ultrasonic testing, and the obvious defective pile segments determined by both. For S420, the weighted score is calculated with a low strain weight of 0.4 and an ultrasonic weight of 0.6 for the other cases. S430, determine the integrity level of the entire pile and locate the defect based on the weighted score.

6. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 5, characterized in that, In step S420, when calculating the weighted score, if there are defects in adjacent pile segments, the compensation coefficient is increased to 1.

2. The same rule is used for exposed pile segments: crack weight 0.4 and ultrasonic weight 0.

6. Finally, the defect location is locked and the integrity level of the entire pile is determined based on the weighted score.

7. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, The optical sensor mounting track and probe mounting ring both adopt a mortise and tenon joint design.

8. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, The spiral support frame is equipped with scale markings for real-time monitoring of the lifting range.

9. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, The probe fixing ring has pre-drilled holes on its side for clamping the ultrasonic probe; and four pre-drilled holes at the bottom for fixing the accelerometer.

10. The multimodal signal acquisition and defect detection system for foundation pile integrity assessment according to claim 1, characterized in that, The optical sensor fixing track and the probe fixing ring are pre-reserved notch structures. The notch curvature is the same as that of the spiral support, and the outer surface has the same features as the spiral support thread surface, so that they can interlock with each other through the protrusions and grooves of the thread.

Images