A method for detecting the quality of pile foundation concrete based on multi-source data

Abstract

This invention relates to the field of pile foundation engineering quality inspection technology, and more particularly to a method for inspecting the quality of pile foundation concrete based on multi-source data. This invention synchronously acquires fiber optic data, stress wave reflection data, acoustic signals, and environmental data using periodically calibrated multi-source sensors. Then, the data undergoes preprocessing, including fiber optic strain-temperature decoupling, stress wave energy entropy extraction, and acoustic density calculation. Subsequently, the preprocessed data is input into a periodically updated improved Bayesian network fusion model to calculate the comprehensive probability of each defect. Locations that do not meet preset probability thresholds are marked and adjusted. Finally, the detection characterization value is determined based on the number of adjustments meeting preset conditions within a preset time and the total number of adjustments. The quality inspection status of the pile foundation concrete is determined based on the detection characterization value, and the sensor calibration cycle and the model's preset probabilities are dynamically optimized. This invention improves the accuracy of pile foundation concrete quality inspection.

Description

Technical Field

[0001] This invention relates to the field of pile foundation engineering quality testing technology, and in particular to a pile foundation concrete quality testing method based on multi-source data. Background Technology

[0002] As the core load-bearing foundation of a building project, the quality of the concrete in pile foundations directly determines the safety and stability of the superstructure. Current methods for testing the quality of pile foundation concrete mainly include core drilling, low-strain reflection wave testing, acoustic wave transmission, and high-strain testing. Among these, non-destructive testing methods are the most widely used in engineering because they do not damage the pile structure.

[0003] However, existing detection technologies have significant limitations: core drilling, while highly accurate, is a destructive method with a limited sampling range, making it prone to missing local defects; low-strain reflected wave methods are greatly affected by excitation energy and sensor coupling quality, and are insufficient in identifying deep defects and micro-cracks in the pile body; acoustic transmission methods rely on only a single acoustic parameter, making it difficult to comprehensively reflect multi-dimensional quality indicators such as concrete density and strength, and are weak in resisting electromagnetic and temperature interference; existing multi-source data fusion detection technologies mostly adopt a fixed weight allocation mode, which cannot dynamically adjust the data confidence level according to the real-time detection environment, resulting in a high rate of false positives and false negatives, making it difficult to achieve accurate quantitative assessment of pile integrity, concrete strength, defect location, and type.

[0004] Chinese Patent Publication No. CN121257902A discloses a quality control and real-time monitoring system for pile foundation concrete, including a data storage unit, a human-computer interaction terminal, and also including: a supply and demand matching module, a detection allocation module, and a quality monitoring module, each of which is bound to several detection units with a unique detection number.

[0005] Therefore, the existing technology has the following problems: due to the single detection dimension and fixed weight allocation in the quality inspection of pile foundation concrete, the quality inspection accuracy of pile foundation concrete is low. Summary of the Invention

[0006] Therefore, this invention provides a pile foundation concrete quality inspection method based on multi-source data to overcome the problem of low quality inspection accuracy of pile foundation concrete in the prior art due to the existence of a single inspection dimension and fixed weight allocation in the quality inspection of pile foundation concrete.

[0007] To achieve the above objectives, the present invention provides a method for detecting the quality of pile foundation concrete based on multi-source data, comprising:

[0008] Multi-source data is acquired through periodically calibrated multi-source sensors, including fiber optic data of pile foundation concrete collected by distributed fiber optic sensors, broadband stress wave reflection data collected by receiving sensors, environmental parameters collected by environmental sensors, and acoustic wave signals penetrating concrete collected by receiving sensor array. The fiber optic sensors include reference fiber and monitoring fiber, and the fiber optic data includes fiber strain and temperature coupling data.

[0009] The fiber optic data is decoupled according to the strain-temperature decoupling model, and the stress wave reflection data is subjected to time-domain analysis to extract energy entropy. The concrete density is calculated based on the corrected acoustic wave data to preprocess the multi-source data. The strain-temperature decoupling model is constructed based on the fiber optic data.

[0010] The preprocessed multi-source data is input into an improved Bayesian network fusion model that is periodically updated using historical multi-source data. The model outputs the comprehensive probability of each defect and marks the defect location when the comprehensive probability of the defect does not meet the preset condition. The preset condition is that the comprehensive probability of the defect is less than or equal to the preset probability.

[0011] The pile foundation is adjusted based on the marked defect locations, and it is determined whether the comprehensive probability of defects in the adjusted pile foundation meets the preset conditions.

[0012] The detection characterization value is determined based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments. The quality detection status of the pile foundation concrete is determined based on the detection characterization value. The calibration cycle of the multi-source sensor is adjusted based on the quality detection status. The preset probability is adjusted based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensor.

[0013] Furthermore, the process of decoupling fiber optic data based on the strain-temperature decoupling model includes: according to To determine the wavelength shift caused by concrete strain, the formula is as follows: To monitor the total wavelength offset acquired by the optical fiber, This is the temperature sensitivity coefficient calibration factor. The wavelength offset is used as a reference fiber for acquisition; based on To determine the actual strain of the pile foundation concrete after fiber optic data parsing, the formula is as follows: This is the initial wavelength of the fiber Bragg grating. The photoelastic coefficient of the optical fiber. This is the coefficient of thermal expansion of the optical fiber.

[0014] Furthermore, the process of performing time-domain analysis on the stress wave reflection data to extract energy entropy includes: performing time-domain analysis on the stress wave reflection data using wavelet packet transform to determine the energy of each sub-band; based on... Determine the energy entropy, where, Let j be the energy of the j-th sub-band. This represents the total energy of all sub-bands.

[0015] Furthermore, the process of inputting the preprocessed multi-source data into an improved Bayesian network fusion model that is periodically updated using historical multi-source data, and outputting the comprehensive probability of each defect, includes: using the preprocessed multi-source data as the input source of the periodically updated improved Bayesian network fusion model, wherein the improved Bayesian network fusion model includes an input data dynamic weight adaptive allocation layer, a conditional probability inference layer, and a defect comprehensive probability fusion calculation layer; based on the environmental parameters and weight rules, the preprocessed multi-source data is weighted by confidence level to determine the contribution weight of each data to defect inference, and ensuring that the sum of the contribution weights is 1; based on the conditional probability rules and multiple weighted multi-source data, the preliminary occurrence probability of each defect type is determined; and the preliminary occurrence probability of each defect type is fused with the corresponding contribution weight to determine the comprehensive probability of each defect.

[0016] Furthermore, the process of determining the test characterization value based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments, and determining the quality test status of the pile foundation concrete based on the test characterization value, includes: calculating the ratio of the number of times the pile foundation meets the preset conditions after adjustment within a preset time to the total number of adjustments to determine the test characterization value; if the test characterization value is greater than or equal to the preset characterization value, the quality test status of the pile foundation concrete is determined to be qualified; if the test characterization value is less than the preset characterization value, the quality test status of the pile foundation concrete is determined to be unqualified.

[0017] Furthermore, the process of adjusting the calibration cycle of the multi-source sensor based on the quality inspection status includes: if the quality inspection status is unqualified, then obtain multiple detection characterization values ​​within a preset time period; calculate the variance and average value of the multiple detection characterization values ​​respectively; if the average value is less than the preset average value and the variance is less than the preset variance, then reduce the calibration cycle of the multi-source sensor based on the ratio of the variance to the preset variance, and the reduction of the calibration cycle is inversely proportional to the ratio.

[0018] Furthermore, the process of adjusting the preset probability based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensors includes: if the quality detection status after adjusting the calibration cycle of the multi-source sensors is unqualified, then the total number of adjustment operations issued within the preset cycle and the number of additional adjustments to the same spatial location are obtained, wherein the number of additional adjustments is the difference between the number of adjustments and the preset number of adjustments; the ratio of the number of additional adjustments to the total number is calculated, and if the ratio is greater than the preset ratio, then the preset probability is adjusted based on the difference between the ratio and the preset ratio.

[0019] Furthermore, the process of adjusting the preset probability based on the difference between the ratio and the preset ratio includes: increasing the preset probability based on the difference between the ratio and the preset ratio, and the increase in the preset probability is proportional to the difference.

[0020] Furthermore, the method further includes: if the quality detection status after adjusting the preset probability is unqualified, then the preset probability is repeatedly adjusted at least once until the number of adjustments is less than the preset number and the quality detection status is qualified, or the number of adjustments is equal to the preset number, at which point the adjustment is stopped; if the quality detection status after stopping the adjustment is unqualified, then the environmental change rate and update delay degree corresponding to each two adjacent updates of the improved Bayesian network fusion model are obtained, wherein the update delay degree... In the formula, To improve the time difference between two adjacent updates in the Bayesian network fusion model, The update cycle is preset; the first average value of each environmental change rate and the second average value of each update delay are calculated. If the second average value is greater than the preset delay and the first average value is greater than the preset change rate, the update cycle of the improved Bayesian network fusion model is adjusted based on the ratio of the second average value to the preset delay.

[0021] Furthermore, the process of adjusting the update cycle of the improved Bayesian network fusion model based on the ratio of the second average value to the preset delay includes: reducing the update cycle of the improved Bayesian network fusion model based on the ratio of the second average value to the preset delay, and the reduction in the update cycle is proportional to the ratio.

[0022] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention synchronously acquires fiber optic data, stress wave reflection data, acoustic signals, and environmental data through periodically calibrated multi-source sensors; then, the data undergoes preprocessing, including fiber optic strain-temperature decoupling, stress wave energy entropy extraction, and acoustic density calculation; subsequently, the preprocessed data is input into a periodically updated improved Bayesian network fusion model to calculate the comprehensive probability of each defect, and locations that do not meet preset probability thresholds are marked and adjusted; finally, based on the number of adjustments meeting preset conditions within a preset time and the total number of adjustments, a detection characterization value is determined, and the quality inspection status of the pile foundation concrete is judged according to the detection characterization value, with dynamic feedback to optimize the sensor calibration cycle and the model's preset probabilities. This invention improves the accuracy of pile foundation concrete quality inspection.

[0023] Furthermore, this invention constructs a strain-temperature decoupling model based on a reference optical fiber, which can eliminate the interference of temperature on monitoring data, thereby making the acquired multi-source data more effective and further improving the quality detection accuracy of pile foundation concrete.

[0024] Furthermore, this invention performs time-domain analysis on reflection data based on wavelet packet transform and determines the energy entropy, which can separate specific frequency components excited by defects of different properties, thereby more sensitively capturing local spectral distortions caused by defects, and further improving the quality detection accuracy of pile foundation concrete.

[0025] Furthermore, this invention determines the comprehensive probability of each defect based on an improved Bayesian network fusion model, which can more accurately determine the defect status of pile foundation concrete, thereby further improving the quality inspection accuracy of pile foundation concrete.

[0026] Furthermore, the present invention determines the quality inspection status of pile foundation concrete based on the comparison between the detected characterization value and the preset characterization value, which can more effectively determine the quality inspection status, thereby enabling more precise subsequent adjustments and further improving the quality inspection accuracy of pile foundation concrete.

[0027] Furthermore, this invention determines the reasons for unqualified quality inspection status based on the variance and average value of the detection characterization values ​​within multiple preset time periods. This allows for a more accurate determination of whether the unqualified quality inspection status is due to poor consistency in the acquisition of multi-source sensor data at different times and under different environments, resulting in unstable preprocessed feature values ​​and unreliable evidence input into the model. Consequently, adjustments can be made more effectively to address this cause, further improving the accuracy of quality inspection of pile foundation concrete.

[0028] Furthermore, this invention reduces the calibration cycle of multi-source sensors based on the ratio of variance to preset variance, which can more accurately determine the calibration cycle of multi-source sensors, making the acquired multi-source data more accurate, and thus further improving the quality detection accuracy of pile foundation concrete.

[0029] Furthermore, this invention determines the reasons for unqualified quality inspection status based on the ratio of additional adjustment times to total number of adjustments. This allows for a more accurate determination of whether the defect is due to repeated adjustments at many locations. In other words, the overall probability of defects may have decreased to a certain stable value after adjustment, but it is still higher than the preset probability, leading to the defect being judged as unqualified again. This results in unqualified quality characterization values, which in turn allows for more effective subsequent adjustments and further improves the quality inspection accuracy of pile foundation concrete.

[0030] Furthermore, the present invention increases the preset probability based on the difference between the ratio and the preset ratio. By relaxing the preset probability, some minor normal fluctuations or measurement noises in the pile foundation that do not affect structural safety are not judged as unqualified, thereby more accurately determining the location of defects and further improving the quality inspection accuracy of pile foundation concrete.

[0031] Furthermore, this invention determines the cause of unqualified quality inspection status based on the first average of multiple environmental change rates and the second average of multiple update delays. This can more effectively determine whether the output structure has errors due to the failure of the improved Bayesian network fusion model to be updated in time, thus causing the unqualified quality inspection status. This allows for more effective subsequent adjustments and further improves the quality inspection accuracy of pile foundation concrete.

[0032] Furthermore, the present invention reduces the update cycle of the improved Bayesian network fusion model based on the ratio of the first average value to the preset delay, which can more accurately determine the update cycle of the improved Bayesian network fusion model, thereby making the comprehensive probability of each defect output by the improved Bayesian network fusion model more accurate, and further improving the quality detection accuracy of pile foundation concrete. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the pile foundation concrete quality inspection system based on multi-source data according to an embodiment of the present invention;

[0034] Figure 2 This is a flowchart illustrating the steps of the pile foundation concrete quality inspection method based on multi-source data according to an embodiment of the present invention.

[0035] Figure 3 This is a flowchart illustrating the steps of determining the comparison result between the detected characterization value and the preset characterization value in an embodiment of the present invention.

[0036] Figure 4 This is a flowchart illustrating the steps of determining the quality detection status based on the adjusted preset probability in an embodiment of the present invention. Detailed Implementation

[0037] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0038] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0039] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0040] Please see Figure 1 The diagram shown is a structural schematic of a pile foundation concrete quality inspection system based on multi-source data according to an embodiment of the present invention. This embodiment of the present invention provides a pile foundation concrete quality inspection system based on multi-source data, which includes an acquisition unit, a preprocessing unit, a detection unit, a determination unit, and an analysis unit.

[0041] The acquisition unit uses periodically calibrated multi-source sensors to acquire multi-source data, including a distributed fiber optic sensor for acquiring fiber optic data of pile foundation concrete, a receiving sensor for acquiring broadband stress wave reflection data, an environmental sensor for acquiring environmental parameters, and a receiving sensor array for acquiring acoustic wave signals penetrating concrete. The fiber optic sensor includes a reference fiber and a monitoring fiber, and the fiber optic data includes fiber strain and temperature coupling data.

[0042] The preprocessing unit is connected to the acquisition unit and is used to decouple the optical fiber data according to the strain-temperature decoupling model, perform time-domain analysis on the stress wave reflection data to extract energy entropy, and calculate the concrete density based on the corrected acoustic wave data to preprocess the multi-source data, wherein the strain-temperature decoupling model is constructed based on the optical fiber data.

[0043] The detection unit is connected to the preprocessing unit. It is used to input the preprocessed multi-source data into the improved Bayesian network fusion model that is periodically updated using historical multi-source data, output the comprehensive probability of each defect, and mark the defect location when the comprehensive probability of the defect does not meet the preset condition. The preset condition is that the comprehensive probability of the defect is less than or equal to the preset probability.

[0044] The determining unit is connected to the detection unit and is used to adjust the pile foundation based on the marked defect location and determine whether the comprehensive probability of defects of the adjusted pile foundation meets the preset conditions.

[0045] The analysis unit is connected to the determination unit. It is used to determine the detection characterization value based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments, and to determine the quality detection status of the pile foundation concrete based on the detection characterization value. It also adjusts the calibration cycle of the multi-source sensor based on the quality detection status, and adjusts the preset probability based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensor.

[0046] Specifically, during the pile foundation construction stage, distributed optical fiber sensors are pre-embedded along the pile body. The sensors include a reference optical fiber that is only affected by temperature and a monitoring optical fiber that is affected by both temperature and strain. The reference optical fiber and the monitoring optical fiber are arranged in parallel, and the pre-embedding depth covers the entire pile length. The two ends of the optical fiber extend to the outside of the pile top and are connected to the signal acquisition module. Two to four sets of broadband stress wave transmitting devices and receiving sensors are symmetrically arranged at the pile top. A cross-hole acoustic CT system, including multiple sets of acoustic wave transmitting probes and receiving sensor arrays, is deployed in the pre-set acoustic logging tubes of adjacent pile foundations or the same pile foundation. All testing instruments are periodically calibrated to ensure that the sensor accuracy and the sampling frequency of the signal acquisition instrument meet the requirements.

[0047] Specifically, the wavelength shift of the pile foundation concrete is collected by distributed optical fiber sensors. The wavelength shift data of the optical fiber output containing only temperature effect is referenced, and the wavelength shift data of the optical fiber output containing temperature and strain coupling effect is monitored to obtain the axial data sequence of the entire pile length. Graded impact energy is applied by a broadband stress wave emitting device, and the stress wave reflection signal is collected by the receiving sensor. The time domain waveform data is recorded. Each energy level is repeatedly excited multiple times, and the signal with stable waveform is selected as valid data. The acoustic wave emitting probe is controlled to emit pulsed acoustic waves, and the receiving sensor array synchronously collects the acoustic wave signal penetrating the concrete.

[0048] Specifically, the travel time data in the acoustic signal is smoothed and denoised to remove outliers. The wave velocity field of the pile body is reconstructed using the SIRT algorithm and simultaneous iterative reconstruction technology, and the concrete density is calculated. In the formula, To reconstruct the wave velocity of the pile concrete, This is the standard wave velocity for dense concrete. This value is a known reference value, ranging from 3800 to 4200 m / s. The standard dense concrete density is 2400 kg / m³, which is a fixed reference value and serves as the benchmark parameter for calculating the actual density. In the formula, 1.2 is the empirical fitting index.

[0049] Specifically, the real-time strain mutation is calculated based on the average strain over a preset period after the initial setting of the concrete. A three-layer Bayesian network fusion model was constructed, with the input layer being the preprocessed fiber strain mutation. Energy entropy H, acoustic density The intermediate layer consists of defect type probability nodes (voids, cracks, loose areas); the output layer consists of the comprehensive probability and location coordinates of defects.

[0050] Specifically, this model uses a "batch inspection cycle" as the update unit. The cycle can be set according to the number of engineering inspection batches and the complexity of the environment. Typically, an update cycle consists of 10-20 pile foundation inspections completed, and the model parameters are iteratively updated before input data. First, after each cycle, the "preprocessed data - actual defect verification results" of all pile foundations within that cycle are collected to form a sample dataset. Second, based on the new sample dataset, the core parameters of the Bayesian network are iteratively optimized, namely, the conditional probability rules for intermediate layer defect types are corrected, and the baseline parameters for calibrating dynamic weight allocation are adjusted.

[0051] Specifically, the core operation process of the improved Bayesian network fusion model is as follows:

[0052] The input layer uses dynamic weight adaptive allocation. Based on real-time environmental parameters and updated weight rules, it performs confidence weighting on the three sets of input data to determine the contribution weight of each data source to defect inference and ensures that the sum of the contribution weights is 1. In this embodiment of the invention, the weight ω1 for optical fiber data is determined as follows: if the residual of the current decoupling model is greater than 0.05με, 80% of the base weight 0.35 is taken, i.e., 0.28; otherwise, the base weight 0.35 is maintained. The above data settings are based on the optimal solution of historical data and experimental data. In this embodiment of the invention, the weight ω2 for stress wave reflection data is set as follows: if the environmental vibration noise is less than 20dB, 70% of the base weight 0.35 is taken, i.e., 0.245; otherwise, the base weight 0.35 is maintained. The above data are all based on the stress wave data settings adapted to different noise environments in the experimental data. In this embodiment of the invention, the weight ω3 for acoustic signals is calculated according to the updated formula ω3=0.3+0.005×(50-h), where h is the current environmental humidity in %, and 50-h is the humidity deviation value in %, which is expressed as a 1% change in humidity corresponding to a 0.005 weight adjustment. The above data are all based on the experimental data that can optimally balance the influence of humidity on acoustic wave propagation.

[0053] The intermediate layer is conditional probability inference, which maps inputs to defect types. Based on the updated conditional probability rules, it infers the preliminary occurrence probability of each defect type, such as voids, cracks, and loose areas, using three sets of weighted input data. The inference process relies on the probabilistic relationships between intermediate nodes of the Bayesian network, combined with periodically updated prior probabilities and likelihood probabilities. This process is existing technology and will not be elaborated further.

[0054] The output layer uses a weighted summation formula to calculate the overall probability of defects by fusing the initial probability of defects from each data source with dynamic weights, thus obtaining the overall probability of occurrence for each defect type. In the formula, x represents three types of defects: voids, cracks, and loose areas. The combined probability of the three types of defects is calculated.

[0055] Specifically, when the overall probability of a defect does not meet the preset conditions, that is, when the overall probability of a defect is greater than the preset probability, it is determined that there is a corresponding defect, and the location and range of the defect are marked by acoustic CT three-dimensional imaging.

[0056] Please see Figure 2 The diagram shown is a flowchart of the pile foundation concrete quality detection method based on multi-source data according to an embodiment of the present invention.

[0057] The specific steps for pile foundation concrete quality detection based on multi-source data in this embodiment of the invention are as follows:

[0058] S1, by acquiring multi-source data through periodically calibrated multi-source sensors, including fiber optic data of pile foundation concrete acquired based on distributed fiber optic sensors, broadband stress wave reflection data acquired based on receiving sensors, environmental parameters acquired based on environmental sensors, and acoustic wave signals penetrating concrete acquired based on receiving sensor array. The fiber optic sensors include reference fiber and monitoring fiber, and the fiber optic data includes fiber strain and temperature coupling data.

[0059] S2, the preprocessing unit connected to the acquisition unit decouples the optical fiber data according to the strain-temperature decoupling model, performs time-domain analysis on the stress wave reflection data to extract energy entropy, and calculates the concrete density based on the corrected acoustic wave data to preprocess the multi-source data, wherein the strain-temperature decoupling model is constructed based on the optical fiber data;

[0060] S3, the preprocessed multi-source data is input into the improved Bayesian network fusion model that is periodically updated using historical multi-source data through the detection unit connected to the preprocessing unit, and the comprehensive probability of each defect is output. When the comprehensive probability of the defect does not meet the preset condition, the defect location is marked. The preset condition is that the comprehensive probability of the defect is less than or equal to the preset probability.

[0061] S4, the determination unit connected to the detection unit adjusts the pile foundation based on the marked defect location, and determines whether the comprehensive probability of defects in the adjusted pile foundation meets the preset conditions.

[0062] S5, the analysis unit connected to the determining unit determines the detection characterization value based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments, and determines the quality detection status of the pile foundation concrete based on the detection characterization value, and adjusts the calibration cycle of the multi-source sensor based on the quality detection status, and adjusts the preset probability based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensor.

[0063] Please see Figure 3As shown, it is a flowchart of the steps for determining the comparison result between the detected characterization value and the preset characterization value in an embodiment of the present invention.

[0064] Specifically, taking the detection of internal defects in pile foundation concrete as an example, and based on the physical performance limits of sensors such as distributed optical fibers, stress wave and acoustic wave arrays, as well as the fault tolerance requirements of complex engineering site environments such as temperature fluctuations, electromagnetic interference, and coupling differences, and combined with a large amount of historical detection data and adjustment feedback records accumulated in long-term engineering practice, the corresponding preset or critical parameter values ​​are set.

[0065] Specifically, in this embodiment of the invention, a preset characterization value L0 = 0.7 is set. The comparison process between the detected characterization value L and the preset characterization value L0 is as follows:

[0066] If the detected characterization value L is greater than or equal to the preset characterization value L0, then the quality test status of the pile foundation concrete is determined to be qualified.

[0067] If the detected value L is less than the preset value L0, then the quality of the pile foundation concrete is determined to be unqualified.

[0068] Specifically, when the quality inspection status is unqualified, multiple detection characterization values ​​within a preset time period are obtained; the variance and average value of the multiple detection characterization values ​​are calculated respectively. If the average value is less than the preset average value and the variance is less than the preset variance, it indicates that the data from the multi-source sensors are inconsistent in different times and environments, making the preprocessed feature values ​​such as energy entropy and density unstable, and making the evidence in the input model unreliable. Then, the calibration cycle of the multi-source sensors is adjusted based on the ratio of the variance to the preset variance.

[0069] Specifically, the preset average value is determined primarily based on the statistical results of historical qualified pile foundation testing data. By retrospectively analyzing a large number of verified qualified pile foundation samples, the distribution range of their "test characterization values" under stable testing conditions is calculated. Typically, the 15th percentile of this distribution is taken as a candidate for the preset average value, and then fine-tuned based on engineering tolerance requirements. Therefore, in this embodiment of the invention, the preset average value is set to 0.68.

[0070] Specifically, the preset variance is determined based on the evaluation of the repeatability of the sensor system under stable conditions. By analyzing the sequence of "detection characterization values" obtained from multiple repeated tests on the same qualified pile foundation within a short period, its dispersion is calculated, and the 95th percentile of the sequence variance is used as the basic threshold. Therefore, in this embodiment of the invention, the preset variance is set to 0.025.

[0071] Specifically, in this embodiment of the invention, a preset ratio P0 = 0.79 is set between the variance and the preset variance. The comparison process based on the ratio P0 between the variance and the preset variance is as follows:

[0072] If the ratio P of the variance to the preset variance is less than or equal to the preset ratio P0, the calibration period will be adjusted to 0.64 times the original calibration period, where the adjusted calibration period is rounded up.

[0073] If the ratio P of the variance to the preset variance is greater than the preset ratio P0, the calibration period will be adjusted to 0.83 times the original calibration period, where the adjusted calibration period is rounded up.

[0074] Specifically, the above multiples are determined based on a comprehensive analysis of historical experience and experimental data to identify the corresponding values ​​that yielded the best results.

[0075] Specifically, when the quality inspection status fails after adjusting the calibration cycle of the multi-source sensors, the total number of adjustment operations issued within the preset cycle and the number of additional adjustments to the same spatial location are obtained. The additional adjustment number is the difference between the number of adjustments to the same spatial location and the preset adjustment number. In this embodiment, the preset adjustment number is 1. The ratio of the additional adjustment number to the total number is calculated. If the ratio is greater than the preset ratio, it indicates that many locations have been repeatedly adjusted. This means that the overall probability of defects may have decreased to a stable value after adjustment, but it is still higher than the preset probability, leading to a renewed failure. In other words, some minor, normal fluctuations or measurement noise in the pile foundation that do not affect structural safety are also judged as "failure" in terms of overall defect probability. The preset probability is then adjusted based on the difference between the ratio and the preset ratio. The preset ratio is determined based on distribution analysis and engineering experience calibration of statistical data accumulated during historical removal processes. Through retrospective analysis of numerous qualified pile foundation inspection cases, it was found that when system parameters are set reasonably, the repeated adjustment rate usually exhibits a skewed distribution, with its 90th percentile not exceeding 35%, combined with the actual fault tolerance requirements of engineering projects. Therefore, in this embodiment of the invention, the preset ratio is set to 0.4.

[0076] Specifically, in this embodiment of the invention, a preset difference Q0 = 0.1 is set between the ratio and the preset ratio. The comparison process based on the difference Q between the ratio and the preset ratio and the preset difference Q0 is as follows:

[0077] If the difference Q between the ratio and the preset ratio is less than or equal to the preset difference Q0, the preset probability is adjusted to 1.3 times the original preset probability, and the adjusted preset probability is retained to two decimal places.

[0078] If the difference Q between the ratio and the preset ratio is greater than the preset difference Q0, the preset probability will be adjusted to 1.8 times the original preset probability, and the adjusted preset probability will be retained to two decimal places.

[0079] Specifically, the above multiples are determined based on a comprehensive analysis of historical experience and experimental data to identify the corresponding values ​​that yielded the best results.

[0080] Please see Figure 4 The diagram shown is a flowchart illustrating the steps of determining the quality detection status based on the adjusted preset probability in an embodiment of the present invention.

[0081] Specifically, if the quality detection status is unqualified after adjusting the preset probability, the preset probability is adjusted at least once until the number of adjustments is less than the preset number and the quality detection status is qualified, or the number of adjustments is equal to the preset number, at which point the adjustment stops. If the quality detection status is still unqualified after stopping the adjustment, the environmental change rate and update latency corresponding to each two adjacent updates of the improved Bayesian network fusion model are obtained, where the update latency... In the formula, To improve the time difference between two adjacent updates in the Bayesian network fusion model, The update cycle is preset. The first average value of each environmental change rate and the second average value of each update delay are calculated. If the second average value is greater than the preset delay and the first average value is greater than the preset change rate, it indicates that the output structure has an error due to the failure of the improved Bayesian network fusion model to be updated in time, resulting in the quality detection status being unqualified. The update cycle of the improved Bayesian network fusion model is then adjusted based on the ratio of the second average value to the preset delay. The first average value is the average value obtained based on the sum of multiple environmental change rates, and the second average value is the average value obtained based on the sum of multiple update delays.

[0082] Specifically, the preset latency is determined by analyzing the correlation between historical data updates and corresponding detection accuracy using a statistical model. Specifically, this involves analyzing the relationship between model update latency and detection error rates across a large number of historical cases to identify the critical latency point where accuracy significantly decreases. Therefore, in this embodiment of the invention, the preset latency is set to 0.5.

[0083] Specifically, the preset change rate is determined based on a sensitivity analysis of the impact of environmental parameters on defect probability. By calculating the correlation function between the variation amplitude of different environmental parameters and multi-source data characteristics such as energy entropy and density coefficient of variation using historical data, the critical inflection point of the characteristic coefficient of variation is determined. Therefore, in this embodiment of the invention, the preset change rate is set to 0.3.

[0084] Specifically, in this embodiment of the invention, a preset ratio R0 of the second average value to the preset delay is set to 1.46. The comparison process based on the ratio R of the second average value to the preset delay and the preset ratio R0 is as follows:

[0085] If the ratio R of the second average value to the preset delay is less than or equal to the preset ratio R0, the update cycle of the improved Bayesian network fusion model will be adjusted to 0.81 times the original update cycle, where the adjusted update cycle is rounded up.

[0086] If the ratio R of the second average value to the preset delay is greater than the preset ratio R0, the update cycle of the improved Bayesian network fusion model will be adjusted to 0.69 times the original update cycle, where the adjusted update cycle is rounded up.

[0087] Specifically, the above multiples are determined based on a comprehensive analysis of historical experience and experimental data to identify the corresponding values ​​that yielded the best results.

[0088] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0089] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A pile foundation concrete quality detection method based on multi-source data, characterized in that, include: Multi-source data is acquired through periodically calibrated multi-source sensors, including fiber optic data of pile foundation concrete collected by distributed fiber optic sensors, broadband stress wave reflection data collected by receiving sensors, environmental parameters collected by environmental sensors, and acoustic wave signals penetrating concrete collected by receiving sensor array. The fiber optic sensors include reference fiber and monitoring fiber, and the fiber optic data includes fiber strain and temperature coupling data. The fiber optic data is decoupled according to the strain-temperature decoupling model, and the stress wave reflection data is subjected to time-domain analysis to extract energy entropy. The concrete density is calculated based on the corrected acoustic wave data to preprocess the multi-source data. The strain-temperature decoupling model is constructed based on the fiber optic data. The preprocessed multi-source data is input into an improved Bayesian network fusion model that is periodically updated using historical multi-source data. The model outputs the comprehensive probability of each defect and marks the defect location when the comprehensive probability of the defect does not meet the preset condition. The preset condition is that the comprehensive probability of the defect is less than or equal to the preset probability. The pile foundation is adjusted based on the marked defect locations, and it is determined whether the comprehensive probability of defects in the adjusted pile foundation meets the preset conditions. The detection characterization value is determined based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments. The quality detection status of the pile foundation concrete is determined based on the detection characterization value. The calibration cycle of the multi-source sensor is adjusted based on the quality detection status. The preset probability is adjusted based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensor. The process of inputting preprocessed multi-source data into an improved Bayesian network fusion model that is periodically updated using historical multi-source data, and outputting the comprehensive probability of each defect, includes: The preprocessed multi-source data is used as the input source for the periodically updated improved Bayesian network fusion model, which includes a dynamic weight adaptive allocation layer for input data, a conditional probability inference layer, and a defect comprehensive probability fusion calculation layer. Based on the environmental parameters and weighting rules, the preprocessed multi-source data is weighted by confidence level to determine the contribution weight of each data to defect reasoning, and to ensure that the sum of the contribution weights is 1. Based on conditional probability rules and multiple weighted multi-source data, the preliminary occurrence probability of each defect type is determined. The initial occurrence probability of each defect type is combined with the corresponding contribution weight to determine the overall probability of each defect.

2. The pile concrete quality detection method based on multi-source data according to claim 1, characterized in that, The process of decoupling fiber optic data based on the strain-temperature decoupling model includes: According to , the wavelength shift caused by the concrete strain is determined, wherein, is the total wavelength shift collected by the monitoring optical fiber, is a temperature sensitivity coefficient calibration factor, is the wavelength shift collected by the reference optical fiber; According to , the actual strain of the pile concrete after the fiber data is analyzed, wherein, is the initial wavelength of the fiber grating, is the photoelastic coefficient of the fiber, is the thermal expansion coefficient of the fiber.

3. The pile foundation concrete quality inspection method based on multi-source data according to claim 2, characterized in that, The process of performing time-domain analysis on the stress wave reflection data to extract energy entropy includes: Wavelet packet transform was used to perform time-domain analysis on the stress wave reflection data to determine the energy of each sub-band. according to Determine the energy entropy, where, Let the energy of the j-th sub-band be... This represents the total energy of all sub-bands.

4. The pile foundation concrete quality inspection method based on multi-source data according to claim 3, characterized in that, The process of determining the test characterization value based on the number of times the pile foundation meets the preset conditions after adjustment within a preset time and the total number of adjustments, and determining the quality test status of the pile foundation concrete based on the test characterization value, includes: Calculate the ratio of the number of times the pile foundation meets the preset conditions after adjustment within a preset time to the total number of adjustments, in order to determine the test characteristic value; If the detected characterization value is greater than or equal to the preset characterization value, the quality test status of the pile foundation concrete is determined to be qualified. If the detected value is less than the preset value, the quality of the pile foundation concrete is determined to be unqualified.

5. The pile foundation concrete quality inspection method based on multi-source data according to claim 4, characterized in that, The process of adjusting the calibration cycle of multi-source sensors based on quality inspection status includes: If the quality inspection status is unqualified, multiple test characterization values ​​within a preset time period will be obtained; The variance and average of multiple detection characterization values ​​are calculated separately. If the average is less than the preset average and the variance is less than the preset variance, the calibration cycle of the multi-source sensor is reduced based on the ratio of the variance to the preset variance, and the reduction in calibration cycle is inversely proportional to the ratio.

6. The pile foundation concrete quality inspection method based on multi-source data according to claim 5, characterized in that, The process of adjusting the preset probability based on the quality monitoring status after adjusting the calibration cycle of the multi-source sensors includes: If the quality detection status is unqualified after adjusting the calibration period of the multi-source sensor, the total number of adjustment operations issued within the preset period and the number of additional adjustments to the same spatial location are obtained, where the number of additional adjustments is the difference between the number of adjustments and the preset number of adjustments. Calculate the ratio of the additional adjustment count to the total number of adjustments. If the ratio is greater than the preset ratio, adjust the preset probability based on the difference between the original ratio and the preset ratio.

7. The pile foundation concrete quality inspection method based on multi-source data according to claim 6, characterized in that, The process of adjusting the preset probability based on the difference between the ratio and the preset ratio includes: The preset probability is increased based on the difference between the ratio and the preset ratio, and the increase in the preset probability is proportional to the difference.

8. The pile foundation concrete quality inspection method based on multi-source data according to claim 7, characterized in that, The method further includes: If the quality inspection status is not qualified after adjusting the preset probability, the preset probability is adjusted at least once until the number of adjustments is less than the preset number and the quality inspection status is qualified, or the number of adjustments is equal to the preset number and the adjustment stops. If the quality inspection status after the adjustment is not satisfactory, then obtain the environmental change rate and update latency of the improved Bayesian network fusion model for each two adjacent updates, where the update latency is... In the formula, To improve the time difference between two adjacent updates in the Bayesian network fusion model, Preset update cycle; Calculate the first average value of each environmental change rate and the second average value of each update delay. If the second average value is greater than a preset delay and the first average value is greater than a preset change rate, then adjust the update cycle of the improved Bayesian network fusion model based on the ratio of the second average value to the preset delay.

9. The pile foundation concrete quality inspection method based on multi-source data according to claim 8, characterized in that, The process of adjusting and improving the update cycle of the Bayesian network fusion model based on the ratio of the second average value to the preset delay includes: The update cycle of the improved Bayesian network fusion model is reduced by lowering the ratio of the second average value to the preset delay, and the reduction in the update cycle is proportional to the ratio.

Images