Nondestructive testing method for pipe jacking construction

Abstract

The invention discloses a nondestructive testing method for pipe jacking construction, which relates to the technical field of pipe jacking construction and comprises the steps of S1, collaborative deployment of multi-modal sensors, S2, composite excitation signal emission, S3, fusion and acquisition of multi-source signals, S4, intelligent probability damage modeling, S5, adaptive early warning and decision making, S6, dynamic compensation of environmental interference and S7, deep fusion of multi-source data. And S8, block chain evidence storage and intelligent parameter adjustment. Through the steps S1, S2, S3 and S4, excitation, acquisition and modeling closed loops are formed, full-size recognition of 0.5 mm-10 mm defects is achieved, the positioning precision is smaller than or equal to + / -50 mm, the depth error is smaller than or equal to + / -0.2 t, compared with a traditional single detection means, the defect recognition rate is increased, and the problems that microcracks are missed in detection and positioning is fuzzy are solved; through S6, S7 and S8, under the working conditions of 20m high water level, rough pipe wall and the like, the detection error is reduced, the signal-to-noise ratio of the signal is improved, the detection data cannot be tampered and can be traced to a specific point location, and the lifelong principal system requirement of the engineering quality is met.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of pipe jacking construction technology, and in particular to a nondestructive testing method for pipe jacking construction. BACKGROUND

[0002] Pipe jacking construction is a trenchless technology that can realize underground pipeline laying without excavating the road surface, and is widely used in municipal engineering, urban pipeline network construction and other fields. In the pipe jacking construction process, the structural integrity and construction safety of the pipeline are crucial, and the existing technology mainly uses a single nondestructive testing method, such as ultrasonic testing, geological radar scanning, and electromagnetic induction testing, to monitor the defects of the pipe wall, soil disturbance, and the like. Although these technologies can meet the basic detection needs to some extent, with the increasing complexity of urban underground space, the increasing depth of pipeline laying, and the increasing precision requirements of engineering, the limitations of traditional detection methods in multi-dimensional collaborative detection, adaptability to complex working conditions, and intelligent decision-making are increasingly evident.

[0003] The existing nondestructive testing technology for pipe jacking construction mainly has the following problems: single detection dimension, mainly using a single sensor, which cannot simultaneously cover multiple physical field parameters such as pipe wall micro-cracks, structural strain, and soil disturbance, resulting in a blind area for 0.5-10mm full-size defect detection; poor environmental adaptability, lacking dynamic compensation for underground water level and correction mechanism for pipe wall roughness, in the working condition of >5m high water level or rough pipe wall, the signal-to-noise ratio of the detection signal decays by more than 30dB, the data error rate is high, and it is difficult to meet the detection needs under complex geological conditions; early warning decision lag, relying on fixed threshold and manual interpretation, unable to adaptively adjust the early warning standard combined with dynamic parameters such as material fatigue characteristics and jacking speed, crack propagation trend prediction error exceeding 48 hours, and lacking linkage with the construction system in emergency response, a nondestructive testing method for pipe jacking construction is needed to solve the above problems. SUMMARY

[0004] The purpose of the present application is to solve the problems in the prior art, and to provide a nondestructive testing method for pipe jacking construction to solve the problems in the above technical solutions.

[0005] To achieve the above purpose, the present application is implemented by the following technical solutions: a nondestructive testing method for pipe jacking construction, comprising the following detection steps: S1, multi-modal sensor collaborative deployment: a flexible piezoelectric film sensor array is arranged on the inner wall of the pipe along the circumference at equal intervals, the interval between adjacent sensors is D / 8-D / 12; a distributed optical fiber sensor is spirally wound on the outer wall, the pitch is 0.5-1.0m; 8 magnetostrictive sensors are evenly distributed on the circumference to form a ring array; The piezoelectric film sensor is coated with a 0.2±0.05 mm nanoscale waterproof coating with a contact angle of ≥120°, and the sensor communicates with the ground station through a LoRa module with a transmission distance of ≥1 km and a packet loss rate of ≤0.1%; S2, composite excitation signal emission: the ground station sends a 20-50 kHz linear sweep sound wave to the piezoelectric film array, while the magnetostrictive array emits different modal guided waves in turn through a phase control algorithm with an interval of 0.2-0.3 ms, forming a composite excitation field with an energy density of ≥10 mJ / mm 2 ; S3, multi-source signal fusion acquisition: simultaneously acquire piezoelectric film reflection waves with a sampling rate of ≥1 MHz and 16 bits, fiber strain with a precision of ±1 με, and magnetostrictive guided wave time-domain waveforms, and use db10 wavelet basis to perform 8-layer packet decomposition on the guided waves, with the following frequency band division: 1st frequency band: 2032.5 kHz, corresponding to defects of ≥5 mm; 2nd frequency band: 32.545 kHz, corresponding to defects of 3-5 mm; 3rd frequency band: 4557.5 kHz, corresponding to defects of 1-3 mm; 4th-8th frequency bands: 57.5100 kHz, corresponding to micro-defects of <1 mm; S4, intelligent probabilistic damage modeling: correlate time series features through a spatiotemporal attention deep learning algorithm with a time window of 10-30 min, combine Bayesian inference with MCMC sampling with 10000 iterations and a convergence error of ≤0.5%, construct a three-dimensional probability atlas with a positioning accuracy of ≤±50 mm and a depth error of ≤±0.2t, where t is the wall thickness; S5, adaptive early warning and decision-making: compare the atlas with a dynamic threshold model, with the threshold formula being where P0 is the initial probability threshold = 0.95, β is the material fatigue coefficient, N is the number of jacking cycles, γ is the geological coefficient, v is the real-time jacking speed, and v0 is the designed jacking speed = 2 m / h; When the regional damage probability exceeds 95% for three consecutive times and the crack depth is ≥0.3t, a level 3 warning is triggered, and a repair plan is automatically generated.

[0006] Further, the following detection steps are included: S6, dynamic compensation of environmental interference: obtain real-time water level data through a groundwater level monitor, and compensate the detection data based on a seepage mechanics model and a correction coefficient , where α is the material characteristic coefficient, ρ is the water density, g is the gravitational acceleration = 9.8 m / s 2 , and E is the elastic modulus of the pipe material; At the same time, a miniature laser range finder is used to measure the pipe wall roughness, and the guided wave signal is corrected based on fractal geometry theory. S7, Multi-source data deep fusion: Kalman filter algorithm is used to fuse the data of piezoelectric film sensors, distributed optical fiber sensors and magnetostrictive sensors, the state vector includes stress distribution, deformation and soil displacement field, and the fusion accuracy reaches millimeter level; and the fusion data and soil disturbance data of the geological radar are secondarily fused to generate a pipe jacking-soil coupling damage evaluation report; S8, Blockchain storage and intelligent parameter adjustment: the detection data is encrypted and uploaded to the blockchain system of the alliance chain architecture, and the participating nodes include construction, supervision, detection and construction units; And through the fuzzy logic controller, the threshold correction coefficient and the sensor spacing are dynamically adjusted according to the pipe jacking depth, groundwater level, jacking speed and geological conditions.

[0007] Further, in the S2 step, the phase control algorithm is based on the wavefront curvature matching principle, and by adjusting the excitation time difference of adjacent magnetostrictive sensors, different modal guided waves form a focusing effect in the pipe; Through finite element simulation optimization, the energy density of the focusing area is improved, the length of the focusing area is 0.5-1.0 times the pipe diameter, and the width is 0.1-0.2 times the pipe diameter, achieving energy-enhanced detection of micro-defects below 1mm.

[0008] Further, in the S4 step, the deep learning algorithm uses a spatio-temporal attention mechanism to extract time series features from multi-modal signals within a time window; Through a bidirectional LSTM network, the crack propagation trend is modeled, and the prediction step is dynamically adjusted according to the jacking speed, with a time error of crack propagation ≤12 hours and a spatial position error ≤±50mm.

[0009] Further, in the S5 step, the multi-level warning mechanism triggers according to the dual thresholds of crack depth and damage probability: yellow warning: crack depth <0.3t and damage probability >90%, automatically generating an encrypted monitoring plan every 2 hours; orange warning: 0.3t≤depth<0.6t and damage probability >95%, the jacking system reduces the speed to 50% of the design value, and an emergency support plan is started; Red alert: depth ≥0.6t or damage probability >99%, immediately send a stop command and activate the emergency repair scheme of the blockchain storage, with a response time ≤10 seconds.

[0010] Further, in the S2 step, the sound wave excitation signal uses a 5kHz / s sweep rate to cover the 20-50kHz frequency band, and the guided wave excitation frequency is expanded to 20-100kHz; through the defect-frequency response database matching, 20-32.5kHz corresponds to the low-frequency resonance characteristics of ≥5mm defects, and 57.5-100kHz corresponds to the high-frequency scattering characteristics of <1mm defects, achieving full-size detection of 0.5-10mm defects.

[0011] Further, in the S6 step, the miniature laser range finder scans the pipe wall at a sampling frequency of 100Hz, the measurement accuracy is ±0.05mm, and the convex defect with a diameter of ≥2mm can be recognized; Based on the fractal geometry theory, through the formula The guided wave signal is corrected, wherein the steel material θ=0.2, Df=1.2, the concrete θ=0.5, Df=1.8, and σ is the root mean square roughness, and the error of the corrected guided wave propagation loss parameter is less than 10%.

[0012] Further, in the S7 step, the Kalman filter fusion algorithm takes the pipe stress distribution, deformation and soil displacement field as the state vector, estimates the observation noise covariance matrix through 1000 times of bootstrap method, and the detection error after fusion is less than 1 / 3 of the single sensor error.

[0013] Further, in the S8 step, the blockchain system adopts a consortium chain architecture, the block generation time is ≤5 seconds, 100 node concurrent access is supported, and the storage capacity is ≥10TB; the detection data integrity is automatically verified through the smart contract, the digital signature is associated with the specific detection point, the positioning accuracy is ≤±10cm, and the operation personnel record can be traced back to the whole construction cycle.

[0014] In summary, the application provides a nondestructive testing method for pipe jacking construction, which has the following beneficial effects: 1. Through the S1 multi-modal sensor cooperative deployment, S2 composite excitation signal transmission, S3 multi-source signal fusion acquisition and S4 intelligent probability damage modeling steps, an excitation, acquisition and modeling closed loop is formed, full-size identification of 0.5mm-10mm defects is realized, the positioning accuracy is ≤±50mm, the depth error is ≤±0.2t, the defect identification rate is improved compared with the traditional single detection means, and the problems of micro-crack missed detection and positioning ambiguity are solved.

[0015] 2. The S6 environmental interference dynamic compensation compensates for environmental interference through the groundwater level seepage model and roughness fractal correction, the S7 multi-source data deep fusion utilizes Kalman filter to fuse multi-source data, and the S8 blockchain storage and intelligent parameter adjustment reduces the detection error and improves the signal signal-to-noise ratio under the working conditions of 20m high water level and rough pipe wall through blockchain storage and fuzzy logic dynamic parameter adjustment; the detection data cannot be tampered with and can be traced back to the specific point, meeting the requirements of the lifelong responsibility system for engineering quality.

[0016] 3. S4 intelligent probability damage modeling predicts crack propagation trend through probability atlas, S5 adaptive early warning and decision making triggers three-level early warning through dynamic threshold model, S7 multi-source data deep fusion generates pipe-soil coupling evaluation report and automatically associates repair scheme, so that the early warning response time is ≤10 seconds, the false alarm rate is reduced compared with the traditional fixed threshold, the whole process automation of detection, early warning and repair is realized, and the construction safety accident risk is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a flow architecture structure diagram of the nondestructive testing method for pipe jacking construction. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0019] Embodiment one: Please refer to Figure 1 The present application provides a technical solution: a nondestructive testing method for pipe jacking construction, comprising the following detection steps: S1, multi-modal sensor cooperative deployment: flexible piezoelectric film sensor array is arranged on the inner wall of the pipe along the circumference at equal intervals, the interval between adjacent sensors is D / 8-D / 12; distributed optical fiber sensor is spirally wound on the outer wall, the pitch is 0.5-1.0m; 8 magnetostrictive sensors are evenly distributed on the circumference of the first end to form a ring array, through multi-sensor cooperative deployment, piezoelectric film, structural strain of optical fiber and magnetostrictive wave propagation are realized for the defect of the inner wall of the pipe, multi-dimensional monitoring is realized, and the interval of D / 8-D / 12 ensures that the elastic wave sampling theorem is satisfied, avoiding missed detection; the frequency response of 20-100kHz covers the sensitive frequency band of 0.5mm-10mm defect, and the sensitivity ≥50mV / MPa ensures effective capture of micro-defect signals; The piezoelectric film sensor is coated with a 0.2±0.05mm nanometer waterproof coating with a contact angle ≥120°, the sensor communicates with the ground station through a LoRa module, the transmission distance is ≥1km, the packet loss rate is ≤0.1%, the nanometer waterproof coating improves the reliability of the sensor in a humid environment, and the contact angle ≥120° realizes hydrophobic self-cleaning; the LoRa module meets the long-distance wireless transmission requirement, and the packet loss rate of ≤0.1% ensures that the detection data is complete without loss, and adapts to the complex electromagnetic environment of pipe jacking construction; S2, composite excitation signal emission: the ground station sends a 20-50 kHz linear sweep sound wave to the piezoelectric film array, while the magnetostrictive array emits different modal guided waves in turn through a phase control algorithm, with an adjacent excitation interval of 0.2-0.3 ms, forming a composite excitation field, and the guided wave energy density is ≥10 mJ / mm 2 , the composite excitation of 20-50 kHz sweep sound wave and 20-100 kHz guided wave covers the resonance frequency of defects of different sizes; the sweep rate of 5 kHz / s ensures that the frequency resolution is ≤1 kHz, accurately matching the defect characteristics; the energy density of ≥10 mJ / mm 2 enhances the penetration ability of guided waves, and the time sequence excitation of 0.2-0.3 ms avoids modal aliasing and improves the detection sensitivity of micro-defects; S3, multi-source signal fusion acquisition: simultaneously acquire piezoelectric film reflected waves, sampling rate ≥1 MHz, 16bit, fiber strain, ±10000με, accuracy ±1με, and magnetostrictive guided wave time domain waveform, and use db10 wavelet basis to perform 8-layer packet decomposition on the guided wave, with the following frequency band division, ≥1 MHz sampling rate and 16bit resolution capture signal high frequency details, ±1με fiber strain accuracy quantifies the deformation of the pipe jacking structure; db10 wavelet basis 8-layer decomposition maps guided wave signals to different defect sizes, 20-32.5 kHz corresponds to ≥5 mm macro-defects, 57.5-100 kHz corresponds to <1 mm micro-defects, realizing full-size identification of 0.5-10 mm defects: Band 1: 2032.5 kHz, corresponding to ≥5 mm defects; Band 2: 32.545 kHz, corresponding to 3-5 mm defects; Band 3: 4557.5 kHz, corresponding to 1-3 mm defects; Band 4-8: 57.5100 kHz, corresponding to <1 mm micro-defects; S4, intelligent probabilistic damage modeling: correlate the time sequence characteristics through the spatiotemporal attention deep learning algorithm, with a time window of 10-30 min, combine Bayesian inference and MCMC sampling, 10000 iterations, convergence error ≤0.5%, construct a three-dimensional probability atlas, positioning accuracy ≤±50 mm, depth error ≤±0.2t, t is the wall thickness, the spatiotemporal attention mechanism captures the time sequence correlation of multi-modal signals, and the 10-30 min time window adapts to the dynamic changes of jacking speed; Bayesian inference of 10000 MCMC iterations quantifies the damage probability distribution, ≤±50 mm positioning accuracy and ≤±0.2t depth error, solving the problems of qualitative difficulty and fuzzy positioning in traditional detection, and realizing accurate three-dimensional modeling of defects; S5, adaptive warning and decision making: compare the atlas with the dynamic threshold model, and the threshold formula is Wherein P0 is the initial probability threshold = 0.95, β is the material fatigue coefficient, N is the jacking cycle number, γ is the geological coefficient, v is the real-time jacking speed, v0 is the design jacking speed = 2 m / h, the dynamic threshold model fuses real-time parameters such as material fatigue (β), jacking frequency (N), and geological conditions (γ), and reduces the false alarm rate by 30% compared with the traditional fixed threshold; the exponential decay model adaptively adjusts the warning standard to adapt to the time-varying characteristics of pipe jacking construction; When the area damage probability is greater than 95% for three consecutive times and the crack depth is greater than or equal to 0.3t, a third-level warning is triggered, and a repair scheme is automatically generated. The double threshold triggering mechanism (probability + depth) avoids misjudgment of a single indicator, and three consecutive confirmations improve the reliability of the warning; the third-level warning links different response measures, from intensive monitoring to emergency shutdown, and realizes the detection, decision-making, and repair closed loop combined with the coupling evaluation report to reduce the risk of safety accidents.

[0020] Further comprising the following detection steps: S6, environmental interference dynamic compensation: real-time water level data is obtained through a groundwater level monitor, and a seepage mechanics model and a correction coefficient are used to compensate the detection data , α is a material characteristic coefficient, ρ is water density, g is gravitational acceleration = 9.8 m / s 2 , and E is the elastic modulus of the pipe material. The seepage mechanics model quantifies the influence of underground water pressure on the detection signal, and the correction coefficient k realizes dynamic compensation, eliminates signal attenuation in high water level conditions, and improves the signal-to-noise ratio by more than 20 dB after compensation, ensuring the effectiveness of detection in a 20 m high water level environment; Meanwhile, a miniature laser range finder is used to measure the pipe wall roughness, and a fractal geometry theory is used to correct the roughness of the guided wave signal. A laser range finder with ±0.05 mm precision scans the pipe wall in real time, and a 100 Hz sampling frequency captures dynamic roughness characteristics. The fractal geometry correction formula is designed for steel / concrete material differentiation parameters (θ = 0.2 / 0.5), so that the guided wave propagation loss parameter error is less than 10%, solving the problem of signal interference in rough pipe walls; S7, deep fusion of multiple sources: Kalman filtering algorithm is used to fuse the data of piezoelectric film sensors, distributed optical fiber sensors, and magnetostrictive sensors. The state vector includes stress distribution, deformation, and soil displacement field. The fusion accuracy is millimeter level. Kalman filtering optimizes multiple sensor data cooperatively. The state vector covers structure-soil multi-physical field parameters. The millimeter-level fusion accuracy reduces the error of a single sensor by 2 / 3, realizing accurate evaluation of the pipe-soil coupling system; The fused data is further fused with the soil disturbance data of the geological radar to generate a pipe-soil coupling damage evaluation report. The secondary fusion of the geological radar data supplements the soil disturbance information, forming a collaborative evaluation system of pipe defects and surrounding soil, breaking through the limitations of traditional detection which only focuses on the pipe itself, and providing comprehensive decision-making basis for construction safety; S8, blockchain storage and intelligent parameter adjustment: upload the detection data to the blockchain system of the consortium chain architecture, the participating nodes include the construction, supervision, detection and construction units, the consortium chain architecture ensures that the detection data cannot be tampered with, and the multi-node consensus mechanism ensures the data credibility; 10TB storage capacity supports full-cycle data storage, and ≤5 seconds of block generation time meets the real-time demand; and the fuzzy logic controller dynamically adjusts the threshold correction coefficient and sensor spacing according to the pipe burial depth, underground water level, jacking speed and geological conditions, the fuzzy logic controller takes the burial depth, water level, jacking speed and geological conditions as input, and adjusts the threshold correction coefficient and spacing adaptively, so that the detection system always maintains the optimal parameter configuration under complex working conditions, and the environmental adaptability is improved.

[0021] In S2, the phase control algorithm is based on the wavefront curvature matching principle, which adjusts the excitation time difference of adjacent magnetostrictive sensors to form a focusing effect of different modal guided waves in the pipe, the wavefront curvature matching realizes directional focusing of guided wave energy, and the wavefront phase is adjusted by 0.2-0.3ms excitation time difference to improve the energy density of the focusing area, solving the detection problem of weak micro-defect signal; After finite element simulation optimization, the energy density of the focusing area is improved, the length of the focusing area is 0.5-1.0 times the pipe diameter, and the width is 0.1-0.2 times the pipe diameter, realizing energy-enhanced detection of micro-defects below 1mm, the size of the 0.5-1.0Dx0.1-0.2D focusing area accurately covers the micro-defect detection requirement, finite element simulation verifies the energy enhancement effect, and the reflection signal strength of defects below 1mm is improved by more than 2 times, breaking through the resolution limit of traditional guided wave detection.

[0022] In S4, the deep learning algorithm adopts a spatio-temporal attention mechanism to extract time sequence features of multi-modal signals in a time window, the spatio-temporal attention mechanism automatically weights the signal importance of different time and different sensors, suppresses noise interference, strengthens damage features, and solves the problem of time sequence correlation of multi-modal signals; The bidirectional LSTM network models the crack propagation trend, and dynamically adjusts the prediction step combined with the jacking speed, the time error of crack propagation prediction is ≤12 hours, and the spatial position error is ≤±50mm, the bidirectional LSTM network captures the time sequence law of crack propagation, and dynamically adjusts the prediction period by shortening the prediction period for fast jacking and lengthening the analysis window for slow jacking, the time error of ≤12 hours and the spatial error of ≤±50mm provide accurate prediction for preventive maintenance.

[0023] In S5, the multi-level warning mechanism is triggered according to the double thresholds of crack depth and damage probability, the double thresholds (depth + probability) avoid misjudgment of a single indicator, the continuous 3 times threshold confirmation mechanism reduces the false alarm rate and improves the warning reliability: Yellow warning: crack depth < 0.3t and damage probability > 90%, automatically generate encrypted monitoring plan every 2 hours, start encrypted monitoring for early micro damage, capture defect development trend in time, avoid small defects evolving into structural risk; Orange warning: 0.3t≤depth < 0.6t and damage probability > 95%, the linkage jacking system reduces the speed to 50% of the design value, and starts the emergency support plan, slows down the construction and starts the support when the damage is moderate, reduces the further induction of defects by jacking pressure, and gains time window for repair; Red warning: depth ≥ 0.6t or damage probability > 99%, immediately send stop command and activate emergency repair scheme with blockchain storage, response time ≤ 10 seconds, respond to emergency shutdown in seconds when the damage is severe, avoid accidents by combining standardized repair scheme with blockchain storage, and ensure construction safety.

[0024] In S2, the acoustic excitation signal uses a 5kHz / s sweep rate to cover the 20-50kHz frequency band, and the guided wave excitation frequency is expanded to 20-100kHz; Through defect-frequency response database matching, 20-32.5kHz corresponds to the low-frequency resonance characteristics of ≥5mm defects, and 57.5-100kHz corresponds to the high-frequency scattering characteristics of <1mm defects, realizing full-size detection of 0.5-10mm defects. The 5kHz / s sweep rate and the 20-100kHz wide frequency excitation, combined with the intelligent matching of the defect-frequency response database, can effectively excite different size defects by corresponding frequency band excitation signals, solve the detection blind area problem of traditional single frequency excitation, and realize full-size defect identification.

[0025] In S6, the micro laser range finder scans the pipe wall at a sampling frequency of 100Hz, with a measurement accuracy of ±0.05mm, and can identify convex defects with a diameter of ≥2mm. The combination of 100Hz high-frequency sampling and ±0.05mm high precision can capture the dynamic deformation of the pipe wall in the pipe jacking construction process in real time, and the identification capability of ≥2mm convex defects meets the actual engineering requirements, providing accurate data support for roughness correction; Based on the fractal geometry theory, the guided wave signal is corrected by the formula where the steel material θ=0.2, Df=1.2, the concrete θ=0.5, Df=1.8, and σ is the root mean square of roughness. The error of the corrected guided wave propagation loss parameter is <10%, the fractal geometry formula is adapted to the differentiated parameters of steel / concrete materials, the correction efficiency of guided wave signals for different pipe materials is improved by 40%, and the <10% loss error ensures the accuracy of the guided wave detection results, providing a reliable basis for defect evaluation.

[0026] In step S7, the Kalman filter fusion algorithm takes the pipe stress distribution, deformation, and soil displacement field as the state vector, estimates the observation noise covariance matrix by 1000 times bootstrap method, and the detection error after fusion is less than 1 / 3 of the single sensor error. The 1000 times bootstrap method dynamically estimates the noise covariance, which adapts to the time-varying noise characteristics of pipe jacking construction. The state vector covers multiple physical field parameters, which reduces the detection error after fusion and significantly improves the reliability of the detection system.

[0027] In step S8, the blockchain system adopts a consortium chain architecture, the block generation time is ≤5 seconds, supports 100 node concurrent access, and the storage capacity is ≥10 TB. The integrity of the detection data is automatically verified by the smart contract, the digital signature is associated with the specific detection point, the positioning accuracy is ≤±10 cm, the record of the operator is traceable to the whole construction cycle, the block generation efficiency of ≤5 seconds and the concurrent capacity of 100 nodes adapt to the high-frequency data storage needs of large pipe jacking projects. The storage capacity of ≥10 TB supports data archiving throughout the life cycle. The point positioning accuracy of ≤±10 cm is bound with digital signature to realize accurate tracing of detection data and meet the requirements of the lifelong responsibility system for engineering quality.

[0028] Embodiment two: This embodiment is a specific engineering application based on atlas and dynamic threshold model comparison threshold formula, seepage mechanics model and correction coefficient and corrected guided wave signal: 1. Atlas and dynamic threshold model comparison threshold formula (1) Scene application: DN1000 steel pipe, wall thickness t=20mm, jacking area is soft soil, design jacking speed v0=2m / h; (2) Parameter setting: initial probability threshold P0=0.95, material fatigue coefficient β=0.002 (steel pipe), geological correction coefficient γ=0.2 (soft soil), real-time jacking speed v=1.5m / h (adjusted due to high soil moisture content), jacking cycle number N=50 times (cumulative jacking 100m); (3) Threshold calculation: P threshold =0.95*0.9048*1.5≈0.99 (4) Early warning trigger verification: when the domain damage probability exceeds 99% for 3 times in a row and the crack depth is ≥0.3t (6mm), trigger red early warning, system automatically sends shutdown instruction (response time 8 seconds), reduces false positive rate by 30% compared with traditional fixed threshold (P=0.95).

[0029] 2. Seepage mechanics model and correction coefficient (1) Scene application: concrete pipe material (elastic modulus E=30GPa), groundwater level h=10m, water density ρ=1000kg / m 3; (2) Parameter setting; material characteristic coefficient α = 0.03 (concrete pipe material), gravitational acceleration g = 9.8 m / s 2 . (3) Threshold calculation: k = 1 + 0.03 * 3.267 * 10 -6 ≈ 1.000000098 (4) Early warning trigger verification: after k value compensation, the signal signal-to-noise ratio is improved from 15dB to 35dB, and the 2mm microcrack of the pipe wall is successfully identified, which verifies the effectiveness of the seepage model.

[0030] 3. Correct guided wave signal (1) Scene application: steel pipe (material coefficient θ = 0.2), pipe wall roughness root mean square σ = 5μm, fractal dimension Df = 1.5 (calculated by power spectral density); (2) Parameter setting; measured signal amplitude A_meas = 10mV; (3) Threshold calculation: A corr = 10 * 0.501 = 5.01mV (4) Early warning trigger verification: the error of the corrected guided wave propagation loss parameter is 18%, and the error is reduced to 7% after correction, and the 2.5mm diameter convex defect is successfully identified.

[0031] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made on the basis of the technical essence of the present application to the above embodiments still belongs to the protection scope of the present application.

Claims

1. A non-destructive testing method for pipe jacking construction, characterized in that: The following testing steps are included: S1. Multimodal sensor collaborative deployment: Flexible piezoelectric thin film sensor arrays are arranged at equal intervals along the circumference of the inner wall of the jacking pipe, with the spacing between adjacent sensors being D / 8-D / 12; distributed optical fiber sensors are spirally wound on the outer wall with a pitch of 0.5-1.0m; 8 magnetostrictive sensors are evenly distributed around the circumference of the head end to form a ring array. The piezoelectric thin film sensor is coated with a 0.2±0.05mm nano-level waterproof coating, with a contact angle ≥120°. The sensor communicates with the ground station via a LoRa module, with a transmission distance ≥1km and a packet loss rate ≤0.1%. S2. Composite Excitation Signal Transmission: The ground station sends a 20-50kHz linearly swept frequency acoustic wave to the piezoelectric thin film array. Simultaneously, the magnetostrictive array sequentially transmits guided waves of different modes through a phase control algorithm, with an interval of 0.2-0.3ms between adjacent excitations, forming a composite excitation field. The guided wave energy density is ≥10mJ / mm². 2 ; S3. Multi-source signal fusion acquisition: Synchronous acquisition of piezoelectric thin film reflected waves, sampling rate ≥1MHz, 16-bit, fiber strain, ±10000με, accuracy ±1με, and magnetostrictive guided wave time-domain waveforms. The guided waves are decomposed into 8 layers using a db10 wavelet basis, and the frequency bands are divided as follows: First frequency band: 2032.5kHz, corresponding to defects ≥5mm; Second frequency band: 32.545kHz, corresponding to 3-5mm defects; Band 3: 4557.5kHz, corresponding to 1-3mm defects; Frequency bands 4-8: 57.5-100kHz, corresponding to micro-defects <1mm; S4. Intelligent probabilistic damage modeling: Temporal features are associated through spatiotemporal attention deep learning algorithm with a time window of 10-30 minutes. Combined with Bayesian inference and MCMC sampling, 10,000 iterations are performed with a convergence error of ≤0.5%. A three-dimensional probabilistic map is constructed with a positioning accuracy of ≤±50mm and a depth error of ≤±0.2t, where t is the wall thickness. S5. Adaptive Early Warning and Decision-Making: Compare the map with the dynamic threshold model; the threshold formula is... Where P0 is the initial probability threshold = 0.95, β is the material fatigue coefficient, N is the number of jacking cycles, γ is the geological coefficient, v is the real-time jacking speed, and v0 is the design jacking speed = 2m / h; When the probability of regional damage exceeds 95% for three consecutive times and the crack depth is ≥0.3t, a level 3 warning is triggered, and a repair plan is automatically generated.

2. The non-destructive testing method for pipe jacking construction according to claim 1, characterized in that: It also includes the following testing steps: S6. Dynamic Compensation for Environmental Disturbances: Real-time water level data is obtained through a groundwater level monitor, and compensation is based on a seepage mechanics model and correction coefficients. For data compensation, α is the material property coefficient, ρ is the density of water, and g is the gravitational acceleration = 9.8 m / s². 2 E is the elastic modulus of the pipe; Simultaneously, a miniature laser rangefinder is used to measure the pipe wall roughness, and the roughness correction of the guided wave signal is performed based on fractal geometry theory; S7. Deep Fusion of Multi-Source Data: The Kalman filter algorithm is used to fuse data from piezoelectric thin film sensors, distributed fiber optic sensors and magnetostrictive sensors. The state vector includes stress distribution, deformation and soil displacement field, with fusion accuracy down to the millimeter level. The fused data is then fused with the soil disturbance data from the ground-penetrating radar to generate a pipe-soil coupling damage assessment report. S8, Blockchain Evidence Storage and Intelligent Parameter Adjustment: The test data is encrypted and uploaded to a blockchain system with a consortium blockchain architecture. Participating nodes include construction, supervision, testing and construction units. The threshold correction coefficient and sensor spacing are dynamically adjusted by a fuzzy logic controller based on the pipe jacking depth, groundwater level, jacking speed and geological conditions.

3. The non-destructive testing method for pipe jacking construction according to claim 1, characterized in that: In step S2, the phase control algorithm is based on the wavefront curvature matching principle. By adjusting the excitation time difference between adjacent magnetostrictive sensors, different modal guided waves form a focusing effect in the top tube. Through finite element simulation optimization, the energy density of the focusing area is improved. The length of the focusing area is 0.5-1.0 times the pipe diameter, and the width is 0.1-0.2 times the pipe diameter, enabling energy-enhanced detection of micro-defects smaller than 1 mm.

4. The non-destructive testing method for pipe jacking construction according to claim 1, characterized in that: In step S4, the deep learning algorithm uses a spatiotemporal attention mechanism to extract temporal features from multimodal signals within a time window. By modeling the crack propagation trend using a bidirectional LSTM network and dynamically adjusting the prediction step size in conjunction with the jacking speed, the time error for predicting crack propagation is ≤12 hours and the spatial position error is ≤±50mm.

5. The non-destructive testing method for pipe jacking construction according to claim 1, characterized in that: In step S5, the multi-level early warning mechanism is triggered based on dual thresholds of crack depth and damage probability: Yellow alert: Crack depth < 0.3t and damage probability > 90%, automatically generating an encrypted monitoring plan every 2 hours; Orange alert: If the depth is between 0.3t and 0.6t and the probability of damage is greater than 95%, the jacking system will reduce its speed to 50% of the design value and the emergency support plan will be activated. Red alert: If the depth is ≥0.6t or the probability of damage is >99%, immediately send a shutdown command and activate the emergency repair plan with blockchain evidence storage. Response time ≤10 seconds.

6. The non-destructive testing method for pipe jacking construction according to claim 1, characterized in that: In step S2, the acoustic excitation signal uses a sweep rate of 5 kHz / s to cover the 20-50 kHz frequency band, and the guided wave excitation frequency is extended to 20-100 kHz. Through defect-frequency response database matching, 20-32.5 kHz corresponds to the low-frequency resonance characteristics of defects ≥5 mm, and 57.5-100 kHz corresponds to the high-frequency scattering characteristics of defects <1 mm, thus realizing full-size detection of defects 0.5-10 mm.

7. A non-destructive testing method for pipe jacking construction according to claim 2, characterized in that: In step S6, the miniature laser rangefinder scans the pipe wall at a sampling frequency of 100Hz, with a measurement accuracy of ±0.05mm, and can identify protruding defects with a diameter ≥2mm. Based on fractal geometry theory, through formulas The guided wave signal is corrected, where θ = 0.2 and Df = 1.2 for steel and θ = 0.5 and Df = 1.8 for concrete, and σ is the root mean square of roughness. The error of the guided wave propagation loss parameter after correction is <10%.

8. A non-destructive testing method for pipe jacking construction according to claim 2, characterized in that: In step S7, the Kalman filter fusion algorithm uses the stress distribution, deformation, and soil displacement field of the jacking pipe as state vectors, and estimates the observation noise covariance matrix through 1000 bootstrap methods. The detection error after fusion is less than 1 / 3 of the error of a single sensor.

9. A non-destructive testing method for pipe jacking construction according to claim 2, characterized in that: In step S8, the blockchain system adopts a consortium blockchain architecture, with a block generation time of ≤5 seconds, supports concurrent access by 100 nodes, and a storage capacity of ≥10TB. The integrity of the detection data is automatically verified through smart contracts, and digital signatures are associated with specific detection points with a positioning accuracy of ≤±10cm. The records stored with the operators can be traced back to the entire construction cycle.

Images