Underground cavity detection and filling quality safety assessment method based on elastic wave exploration

Abstract

The invention belongs to the technical field of traffic infrastructure underground disease detection, and particularly relates to an underground cavity detection and filling quality safety assessment method based on elastic wave exploration. Comprising the following steps: step 1, arranging elastic wave exploration survey lines along the longitudinal direction and the transverse direction of a target area, simultaneously extracting three feature values of reflection intensity, two-way travel time and instantaneous spectral bandwidth for each voxel to identify a cavity area, dividing injection and filling subareas through three-dimensional connected domain search, and determining a target filling volume of each injection and filling subarea; and 2, after filling media are injected into all the cavity areas, the filling effect is evaluated in combination with settlement monitoring, and a differential inspection management scheme is formulated. According to the invention, cavity refined partition identification based on elastic wave exploration, quantitative evaluation of filling quality and differentiated inspection management of multi-source data fusion are realized.

Description

Technical Field

[0001] This invention belongs to the field of underground defect detection technology for transportation infrastructure, and relates to intelligent sensing systems, specifically to a method for underground cavity detection and filling quality and safety assessment based on elastic wave exploration. Background Technology

[0002] With the continuous expansion of the construction and operation scale of transportation infrastructure, underground cavities beneath highway subgrades, urban roads, and bridge and culvert foundations have become a major hidden danger threatening traffic safety. The causes of underground cavities include soil loss due to groundwater erosion, loose areas formed by inadequate backfilling during construction, erosion of surrounding soil caused by leakage from existing pipelines, and dissolution in soluble rock areas. Under repeated traffic loads, these cavities may lead to road surface collapse, subgrade settlement, and even major traffic accidents. Therefore, accurate detection of underground cavities beneath transportation infrastructure and scientific evaluation of the backfilling quality after treatment are critical technical issues that urgently need to be addressed in the construction and maintenance management of transportation engineering. Currently, geophysical non-destructive testing methods such as elastic wave exploration and high-density electrical resistivity tomography are commonly used in transportation engineering practice to detect target areas beneath the subgrade and pavement. In elastic wave exploration, existing methods involve deploying seismic sources and detector arrays on the road surface to excite elastic wave pulses and receive reflected signals from the interface of the underground medium, forming elastic wave reflection profile data to identify abnormal areas beneath the subgrade.

[0003] However, existing elastic wave exploration methods still have the following shortcomings when applied to the identification of underground cavities in transportation infrastructure: On the one hand, the detection results are usually presented only in the form of two-dimensional profile diagrams. Transportation engineers rely on experience to manually interpret abnormal reflection areas in the reflected wave profiles to determine the approximate location and extent of the cavity. This method of relying on manual experience is highly subjective, and different interpreters may have significantly different interpretations of the same set of detection data. There is a lack of a systematic method for jointly analyzing multi-dimensional elastic wave properties such as reflection intensity, two-way travel time, and spectral characteristics, and it is difficult to accurately measure... On the one hand, existing technologies lack an automatic search and partitioning method for three-dimensional connected domains based on elastic wave voxel models. This makes it impossible to automatically group spatially connected candidate voxels of cavities into independent filling partitions. When formulating roadbed cavity backfilling plans, construction personnel can often only roughly estimate the total amount of filling material to be injected based on limited profile information. There is a lack of fine division of the three-dimensional spatial distribution of cavities and accurate calculation of their volume, resulting in a low degree of matching between the actual injection volume and the actual volume of the cavities. This leads to situations where underfilling results in cavity residue or overfilling causes material waste and road surface heave.

[0004] In terms of safety assessment of filling quality, existing elastic wave exploration technology also has significant shortcomings, restricting the accuracy and reliability of quality inspection and testing of roadbed cavity filling treatment in traffic engineering. Current technologies typically involve re-laying survey lines and re-measuring elastic waves after filling, but the analysis of the re-measuring data remains at the level of manually comparing the differences in profiles before and after filling, lacking a method for quantitatively calculating the similarity of elastic wave reflection waveforms at the same measuring point before and after filling. This qualitative comparison method is inherently subjective and imprecise. While changes in the reflection waveform can reflect the improvement in wave impedance matching between the filling medium and the surrounding roadbed soil to some extent, visual comparison alone cannot provide an accurate safety score for the filling effect, nor can it automatically identify the specific location and extent of insufficient filling sections, resulting in a lack of quantitative basis for the safety assessment conclusions of the filling quality. Traffic engineering technicians find it difficult to establish a quantitative correspondence between the elastic wave properties before and after filling based on retest data, and they are also unable to jointly analyze the filling effect evaluation results with the road settlement monitoring data to form a differentiated safety inspection and management strategy for different road sections. As a result, the quality and safety management of roads after filling still relies on a single means of extensive inspection, and cannot achieve refined quality and safety inspection and risk classification control based on multi-source data fusion. Summary of the Invention

[0005] The main objective of this invention is to provide a method for detecting underground cavities and assessing the quality and safety of filling based on elastic wave exploration. This method enables refined zoning identification of cavities based on elastic wave exploration, quantitative assessment of filling quality, and differentiated inspection management through multi-source data fusion.

[0006] To address the aforementioned technical problems, this invention provides a method for detecting and assessing the quality and safety of filling underground cavities based on elastic wave exploration, comprising the following steps: Step 1: Deploy elastic wave exploration lines longitudinally and laterally along the target area. Collect elastic wave reflection profile data below the target area through seismic source excitation and detector reception. Perform three-dimensional migration imaging processing on the elastic wave reflection profile data to obtain a three-dimensional voxel model of the reflection intensity of the underground space in the target area. Simultaneously extract three feature values ​​for each voxel in the three-dimensional voxel model of reflection intensity: reflection intensity, two-way travel time, and instantaneous spectral bandwidth. Voxels whose three feature values ​​all exceed their respective set anomaly thresholds are marked as cavity candidate voxels. Perform a three-dimensional connected domain search on all cavity candidate voxels. Spatially adjacent cavity candidate voxels are merged into the same connected region. Each connected region constitutes an independent filling zone, and the target filling volume of each filling zone is determined. Step 2: After the filling medium is injected into each cavity area identified in Step 1, a re-measurement is carried out along the same elastic wave exploration line as in Step 1. The reflection profile data after filling is obtained by excitation of the seismic source and reception of the detector. The reflection profile data after filling is cross-correlation calculation is performed on the reflection waveforms of the elastic wave reflection profile data obtained in Step 1 at each measurement point to obtain the structural similarity and generate a filling effect score distribution map. The sections with scores below the set qualified threshold in the filling effect score distribution map are marked as insufficient filling sections. Settlement monitoring points are set up longitudinally on the surface of the target area. The cumulative settlement of each settlement monitoring point is collected according to the set monitoring cycle. When the cumulative settlement of any settlement monitoring point exceeds the preset settlement alarm value, an early warning information is generated. According to the location of the insufficient filling section and the early warning information, supplementary filling or shortened inspection interval is arranged for the corresponding filling zone. For filling zones where the filling effect score reaches the qualified threshold and the cumulative settlement is within the settlement alarm value, the regular inspection interval is maintained.

[0007] Furthermore, in step 1, the reflection intensity characterizes the impedance difference at the medium interface, the two-way travel time characterizes the cavity depth, and the instantaneous spectral bandwidth characterizes the scattering characteristics of the cavity boundary; voxels whose only some characteristic values ​​exceed the abnormal threshold are marked as suspected voxels.

[0008] Furthermore, in step 1, for suspected voxels adjacent to the connected region space, it is determined whether to merge the suspected voxel into the adjacent filling partition based on whether the ratio of the adjacent area between the suspected voxel and the connected region to the surface area of ​​the suspected voxel exceeds the set merging ratio threshold. Suspected voxels whose ratio exceeds the merging ratio threshold are merged into the adjacent filling partition to participate in the filling volume calculation. The total volume of the void is estimated by summing the voxel volumes of all voxels contained in each filling partition. The total volume of the void is multiplied by the filling coefficient to obtain the target filling volume. The filling coefficient is greater than 1 and is used to compensate for the volume estimation error caused by 3D offset imaging processing and voxel segmentation. The filling priority order is determined according to the burial depth and the total volume of the void of each filling partition. Filling partitions with shallow burial depth and large volume are filled first.

[0009] Furthermore, the filling medium injection process includes: drilling one airbag placement hole at the boundary between the current filling zone and the adjacent filling zone according to the filling priority order; folding the reusable flexible airbag body and sending it through the airbag placement hole to the boundary section inside the cavity and unfolding it; the reusable flexible airbag body is made of alkali-resistant rubber material, and three contact pressure sensors are evenly arranged circumferentially on the outer surface of the airbag body; the airbag body is connected to the air pump control valve group on the ground through a high-pressure air pipe; starting the air pump control valve group to fill the airbag body with compressed air, so that the airbag body gradually expands inside the cavity until the outer surface of the airbag body is in contact with the cavity wall; during the inflation process, continuously reading the real-time pressure values ​​of the three contact pressure sensors; when the real-time pressure values ​​of the three contact pressure sensors all reach the preset contact pressure and the fluctuation amplitude is less than the preset stable deviation within 10 consecutive seconds, closing the air pump control valve group and locking the air pressure inside the airbag.

[0010] Furthermore, the verification of the sealing effectiveness is as follows: After locking the air pressure inside the bladder, the current air pressure value inside the bladder is recorded as the locking reference value. The preset pressure holding observation time is allowed to detect whether there is a leakage channel between the bladder and the cavity wall. After the pressure holding observation time ends, the air pressure value inside the bladder is read again, and the pressure drop of the air pressure value inside the bladder relative to the locking reference value is calculated. If the pressure drop is less than the preset allowable pressure drop threshold, the sealing is determined to be effective. If the pressure drop reaches or exceeds the allowable pressure drop threshold, the air pump control valve group is restarted to replenish the pressure. After the pressure replenishment is completed, the air pressure value inside the bladder is recorded again as the new locking reference value, and the pressure holding observation is repeated until the sealing is determined to be effective.

[0011] Furthermore, the filling medium injection process includes: drilling filling holes and venting holes at the top of the current filling zone. The venting holes are located at the end of the current filling zone away from the filling holes, used to expel air from the filling zone during the filling process and as an observation channel for slurry overflow when the filling is terminated; inserting the filling pipeline into the interior of the current filling zone through the filling holes, installing a pressure transmitter and an electromagnetic flowmeter sequentially at the orifice section of the filling pipeline, and installing a displacement sensor at the surface of the orifice of the filling holes; and installing a conductivity probe near the back of the recyclable flexible airbag through the airbag placement hole of the adjacent filling zone or a separately drilled monitoring hole to detect whether the foamed concrete slurry bypasses or penetrates the recyclable flexible airbag into the adjacent filling zone.

[0012] Furthermore, the grouting pump is a variable frequency screw pump, which serves as the grouting power source to inject foamed concrete into the current grouting zone through the grouting pipeline. The slipform controller operates cycle by cycle with a fixed control period. In each control cycle, it reads the instantaneous injection flow rate value collected by the electromagnetic flowmeter and adds it to the cumulative injection volume accumulated in previous control cycles to obtain the current cumulative injection volume. At the same time, it reads the actual rotational speed value of the variable frequency screw pump and converts the actual rotational speed value into an estimated pump discharge volume value based on the pump's single-rotation displacement. The estimated pump discharge volume value is compared with the current cumulative injection volume. If the deviation between the two exceeds the preset integral verification deviation threshold, the current cumulative injection volume is corrected using the estimated pump discharge volume value. If the deviation between the two is within the integral verification deviation threshold, the current cumulative injection volume is directly used to proceed to the next control calculation.

[0013] Furthermore, the sliding mode controller calculates the volume deviation between the corrected current cumulative injection volume and the target filling volume in each control cycle. At the same time, it calculates the change in volume deviation in the current control cycle relative to the volume deviation in the previous control cycle to obtain the volume deviation change rate. The weighted combination of volume deviation and volume deviation change rate is used as the sliding mode variable. When the sliding mode variable is greater than zero, it indicates that the injection process is still in the stage of approaching the target filling volume. When the sliding mode variable is equal to zero or less than zero, it indicates that the current cumulative injection volume has reached the target filling volume.

[0014] Furthermore, when the sliding mode variable is greater than zero, the sliding mode controller enters the approach phase: if the current injection pressure value collected by the pressure transmitter is lower than the pressure safety limit, the current road rebound displacement value collected by the displacement sensor is lower than the displacement safety limit, and the current conductivity value collected by the conductivity probe is lower than the cross-contamination conductivity threshold, then the sliding mode controller outputs a positive pump speed increment command according to the constant velocity approach method, driving the variable frequency screw pump to maintain or increase the injection flow rate; if the current injection pressure value reaches the pressure safety limit or the current road rebound displacement value reaches the displacement safety limit, then the pump speed increment command is switched to a negative value, waiting for the current injection pressure value and the current road rebound displacement value to both fall back to their respective safety limits. After falling below the upper limit, the positive trend resumes. If the current conductivity value reaches or exceeds the cross-conductivity threshold, the sliding mode controller outputs a pump stop command to pause filling. After the reclaimable flexible airbag is pressurized, reinforced, or replaced, sliding mode control resumes from the cumulative injection volume at the time of the pause. When the sliding mode variable is equal to or less than zero, the sliding mode controller outputs a pump stop command to shut down the variable frequency screw pump and confirms full filling by observing whether foamed concrete slurry overflows from the vent hole. After full filling, the air pump control valve group is opened to perform a depressurization operation on the reclaimable flexible airbag. After the airbag is fully contracted, the reclaimable flexible airbag is retrieved to the ground through the airbag placement hole, and the conductivity probe is removed.

[0015] Furthermore, in step 2, the cross-correlation calculation specifically involves calculating the normalized cross-correlation coefficient for the elastic wave reflection waveforms at the same measuring point before and after filling within a set time window. The higher the cross-correlation coefficient, the better the impedance matching between the filling body and the surrounding medium, and the higher the corresponding filling effect score. The section corresponding to the measuring point with a cross-correlation coefficient lower than the set qualified threshold is judged as an insufficiently filled section.

[0016] The method for detecting and assessing the quality and safety of filling underground cavities based on elastic wave exploration, as described in this invention, has the following beneficial effects: This invention, at the level of elastic wave exploration data processing, performs three-dimensional offset imaging processing on the reflection profile data acquired by elastic waves and constructs a three-dimensional voxel model of reflection intensity. Based on this, it simultaneously extracts three feature values ​​for each voxel: reflection intensity, two-way travel time, and instantaneous spectral bandwidth, and performs joint discrimination. Compared with the traditional single-index threshold judgment method, this effectively eliminates interference from non-cavity anomalies such as buried pipelines and aquifer changes in the roadbed, significantly reducing the false detection rate of cavity identification and providing a reliable basis for the accurate location of underground defects in transportation infrastructure. Through three-dimensional connected domain search and suspected voxel merging mechanism, this invention can automatically divide spatially continuous cavity areas into independent filling zones and accurately calculate the target filling volume of each zone. This provides a quantitative basis for subsequent roadbed cavity filling construction, overcoming the problem of inaccurate cavity range and volume estimation caused by relying on manual experience to interpret two-dimensional profile data in existing technologies, effectively ensuring the scientific and economical nature of filling and treatment construction in transportation engineering.

[0017] In terms of filling quality assessment, this invention quantitatively compares elastic wave data before and after filling by repeating measurements along the same elastic wave exploration line and using cross-correlation calculation of reflected waveforms at each measurement point. This establishes a filling effect scoring system based on normalized cross-correlation coefficients. Compared to the qualitative judgment method in existing technologies that relies on engineers visually comparing the differences in cross-sections before and after filling, the waveform cross-correlation method of this invention can provide accurate filling effect scores at each measurement point, automatically generate a filling effect score distribution map, and locate insufficient filling sections, eliminating the subjective differences of manual interpretation. Meanwhile, this invention combines the results of elastic wave retest assessment with pavement settlement monitoring data, and implements differentiated inspection and management strategies for different filling zones based on the filling effect score and cumulative settlement. Road sections with satisfactory filling effect scores and settlement within the alarm value maintain regular inspection intervals, while road sections with unsatisfactory scores or abnormal settlement are arranged for refilling or shortened inspection intervals. This achieves a systematic improvement in the quality assessment of traffic infrastructure filling from qualitative to quantitative and from single indicators to multi-source data fusion, providing strong technical support for safe maintenance and disease prevention during road operation. Attached Figure Description

[0018] Figure 1This is a schematic diagram of the elastic wave exploration reflection profile and the distribution of voxel eigenvalues ​​along the survey line provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the recyclable flexible airbag partitioning and sensor deployment principle provided in an embodiment of the present invention. Detailed Implementation

[0019] The method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0020] A method for detecting and assessing the quality and safety of filling underground cavities based on elastic wave exploration includes the following steps: Step 1: Deploy elastic wave exploration lines longitudinally and laterally along the target area. Collect elastic wave reflection profile data below the target area by excitation from the seismic source and reception by the detector. Perform three-dimensional migration imaging processing on the elastic wave reflection profile data to obtain a three-dimensional voxel model of the reflection intensity of the underground space in the target area. Extract three feature values ​​of reflection intensity, two-way travel time and instantaneous spectral bandwidth for each voxel to identify cavity areas. Divide the filling zones through three-dimensional connected domain search and determine the target filling volume of each filling zone. Step 2: After the filling medium is injected into each cavity area identified in Step 1, the same elastic wave exploration line in Step 1 is re-measured. The reflection profile data after filling is obtained by excitation by the seismic source and reception by the detector. The reflection profile data after filling is cross-correlated with the elastic wave reflection profile data obtained in Step 1 at each measurement point to obtain the structural similarity. A filling effect score distribution map is generated. The filling effect is evaluated in combination with settlement monitoring and a differentiated inspection and management plan is formulated. The filling medium injection process after step 1 and before step 2 includes: placing recyclable flexible airbags at the boundary of adjacent filling zones according to the filling priority order and inflating them to form a physical isolation boundary. After verifying the effectiveness of the sealing, foamed concrete is injected into the current filling zone through the filling pipeline. During the filling process, a sliding form controller is used to adjust the output flow of the grouting pump in real time according to the deviation between the cumulative injection volume and the target filling volume. The injection pressure and the surface rebound displacement are used as safety constraints. At the same time, the conductivity probes set in adjacent filling zones are used to monitor the grout cross-contamination status. After the filling is completed, the recyclable flexible airbags are recovered and the process is repeated for the next filling zone.

[0021] The following explanation uses the treatment of roadbed cavities as an example, but this method is also applicable to other scenarios such as tunnel lining, dam bodies, and mining subsidence areas for underground cavity detection and filling quality assessment. Before conducting preliminary elastic wave exploration of underground cavities in the target area, it is necessary to first determine the detection plan to obtain the spatial distribution information of cavities below the target area, providing accurate geometric basis for subsequent filling construction and filling quality assessment. This invention uses elastic wave exploration as the main detection method. Elastic wave exploration uses a seismic source to generate elastic wave pulses into the ground and a detector to receive the reflected signals to detect the distribution of the underground medium. When elastic waves encounter the interface of media with different wave impedances, reflection will occur. Because the cavity area is filled with air or loose materials, its wave impedance is significantly different from that of the surrounding dense soil, thus showing a strong reflection anomaly in the elastic wave reflection profile.

[0022] In practice, elastic wave detection lines are laid out longitudinally and laterally along the target area. Longitudinal lines are laid along the central axis and both sides of the target area, while lateral lines are laid out at certain intervals perpendicular to the direction of the target area. The selection of the line spacing affects the lateral resolution of the final 3D imaging. In one embodiment of the invention, the longitudinal line spacing is set to 1.0 meter, and the lateral line spacing is set to 0.5 meters. This density achieves a balance between detection efficiency and spatial resolution, meeting the engineering requirements for cavity boundary positioning accuracy in underground cavity detection scenarios. In another scenario with higher safety requirements, such as highways or heavy-haul railways, the line spacing can be further reduced to 0.5 meters longitudinally and 0.3 meters laterally to obtain more detailed underground structural information. The dominant frequency of the seismic source is selected based on the requirements of detection depth and resolution. A higher center frequency results in better longitudinal resolution but shallower penetration depth, while a lower center frequency results in greater penetration depth but decreased resolution. For underground cavity detection scenarios, the typical burial depth of the cavity ranges from 0.5 meters to 8 meters. Using a source-detector combination with a center frequency of 200 Hz can balance detection capability and resolution performance within this depth range. When the cavity depth exceeds 6 meters, a source-detector combination with a center frequency of 100 Hz can be used to ensure sufficient penetration depth.

[0023] Elastic wave reflection profile data is acquired point-by-point along each elastic wave exploration line by excitation and reception. Each line's elastic wave reflection profile data consists of a series of seismic traces arranged along the line direction. Each trace records the temporal amplitude sequence of the reflected wave after excitation by the source at that location. During acquisition, the trace spacing is typically set to 0.02 to 0.05 meters, and the time sampling interval is determined based on the source's dominant frequency. For a 200 Hz source-detector combination, the time sampling interval is generally set to 0.25 milliseconds. The number of sampling points per trace is set to 2048 or 4096, corresponding to a time window depth covering the target detection range. To ensure that the data from each line can be stitched together in a unified spatial coordinate system, the start and end coordinates of each line are recorded using a real-time dynamic differential positioning system, with positioning accuracy controlled to be better than 0.05 meters horizontally and better than 0.03 meters vertically.

[0024] After data acquisition, three-dimensional migration imaging was performed on all elastic wave reflection profile data obtained from all elastic wave survey lines. The purpose of three-dimensional migration imaging is to migrate the reflection events in the elastic wave reflection profile data from the time domain to the correct spatial location, eliminating the diffraction tails and positional offsets caused by elastic waves during underground propagation, thereby restoring the true spatial location and shape of the underground reflector. The basic idea of ​​migration imaging is to calculate the possible set of underground scattering points corresponding to each time sampling point in the elastic wave reflection profile data, based on the propagation speed of elastic waves in the underground medium. Then, the energy from different survey lines and different seismic traces is superimposed and focused at the correct spatial location. When multiple seismic traces generate energy superposition at the same spatial location, the reflection intensity at that location is enhanced, while diffraction energy deviating from the true location is suppressed due to the phase inconsistency during superposition, ultimately achieving high-resolution three-dimensional imaging of the underground structure.

[0025] refer to Figure 1 , Figure 1The upper part of the image shows the elastic wave exploration reflection profile acquired along the longitudinal survey line of the target area and processed by 3D migration imaging. The horizontal axis represents the survey line distance in meters, and the vertical axis represents the two-way travel time in milliseconds. Several approximately horizontal continuous reflection bands can be observed in the profile. These bands correspond to the medium interfaces between different strata within the target area. Elastic waves are reflected at interfaces where the wave impedance changes abruptly, forming reflection phase axes extending horizontally. Within a range of approximately 3.5 to 8.5 meters along the survey line and approximately 7 to 12 milliseconds in two-way travel time, the first anomalous area, marked by a dashed rectangle, appears. The reflection signal in this area exhibits a distinct hyperbolic shape, a typical characteristic of diffracted waves generated when elastic waves encounter cavity boundaries, remaining after migration imaging. This indicates the presence of a voxel accumulation of cavity candidate at this location. Within a range of approximately 11.5 to 16.5 meters from the survey line and approximately 12 to 17.5 milliseconds of two-way travel time, a second anomalous area, marked with a dashed rectangle, appeared. This area also exhibited localized enhanced reflection and hyperbolic diffraction characteristics, corresponding to another cavity area. The reflection signal intensity of both anomalous areas was significantly higher than the background reflection of the surrounding normal soil layers. Furthermore, the reflection signals within the anomalous areas displayed a discontinuous and disordered distribution, contrasting with the smooth and continuous pattern of normal layered reflections. Figure 1 The lower half of the figure shows the distribution curves of voxel eigenvalues ​​extracted along the same survey line direction from the three-dimensional voxel model of reflection intensity. The horizontal axis is also the survey line distance, and the vertical axis is the normalized eigenvalue. The figure contains two curves: the solid line represents the normalized reflection intensity value of each voxel, and the dashed line represents the normalized instantaneous spectral bandwidth value of each voxel. The horizontal dotted line marks the anomaly threshold, which is calculated based on the mean and standard deviation of all voxels. Within the survey line distance range corresponding to the two cavity regions, both the reflection intensity curve and the instantaneous spectral bandwidth curve show peaks significantly higher than the anomaly threshold, and the peak intervals of the two curves highly overlap in space. This indicates that not only is there a drastic change in dielectric impedance at the cavity boundary leading to an increase in reflection intensity, but the scattering effect of the irregular cavity wall on elastic waves also causes a synchronous increase in spectral broadening. The segments where both curves exceed the anomaly threshold are marked with filled markers. The voxels in these segments are the candidate voxel segments of the cavity that meet the anomaly discrimination criteria in both reflection intensity and instantaneous spectral bandwidth. Between the two void areas and outside the void areas, the values ​​of both curves are below the anomaly threshold, with small fluctuations, reflecting the stable distribution of voxel characteristic values ​​in the normal soil layer area.

[0026] In a preferred embodiment, the three-dimensional migration imaging process is implemented using the Kirchhoff migration algorithm. For any point to be imaged in three-dimensional space... ,in The coordinates are along the longitudinal direction of the target area. The coordinates are along the horizontal axis of the target area. For depth coordinates pointing vertically downwards, the Kirchhoff migration process involves: traversing all seismic traces along all survey lines, and for the first... The first test line The location of the source-detector for each seismic channel is as follows: Calculate the source-detector position To the point of imaging spatial distance And based on the elastic wave propagation speed of the underground medium Calculate the two-way travel time required for the elastic wave to travel from the seismic source to the imaging point and then be received by the detector. Then from the first Extracting two-way travel time from the reflected wave amplitude sequence of each seismic trace The corresponding amplitude values ​​are then used to finally locate all seismic traces at the points to be imaged. The amplitude values ​​extracted at each point are summed to obtain the offset imaging amplitude value at that point. Among these, the elastic wave propagation velocity... The elastic wave propagation velocity of the target area strata can be determined in advance through common center point gather analysis or borehole calibration. It is generally in the range of 150 m / s to 600 m / s. The longitudinal wave velocity is higher when the water saturation is high and lower when the velocity is dry and loose.

[0027] In another implementation, three-dimensional migration imaging processing can also be achieved using a frequency wavenumber domain migration algorithm. This algorithm converts the elastic wave reflection profile data to the frequency wavenumber domain by performing a two-dimensional or three-dimensional Fourier transform. In the frequency wavenumber domain, phase correction is performed using the analytical relationship of the wave equation, and then an inverse transform is performed to return to the spatial domain. The computational efficiency is higher than that of the Kirchhoff migration algorithm, and it is suitable for situations with a large amount of survey line data.

[0028] After performing the aforementioned offset imaging calculations on all points to be imaged in three-dimensional space, a three-dimensional voxel model of reflection intensity is obtained. This model discretizes the three-dimensional space below the target area into a regularly arranged voxel grid, with each voxel corresponding to a spatial cube unit. The size of the voxel determines the spatial resolution of the three-dimensional model. In one embodiment of the invention, the voxel's side length is set to 0.2 meters in both the longitudinal and transverse directions, and 0.1 meters in the vertical direction. This voxel size matches the theoretical resolution of elastic wave exploration at a frequency of 200 Hz. The value stored in each voxel is the reflection intensity value obtained after three-dimensional offset imaging processing at its spatial location. A higher reflection intensity value indicates a greater difference in wave impedance at the interface of the medium at that location.

[0029] To accurately identify cavity regions from a three-dimensional voxel model of reflection intensity, this invention extracts three feature values ​​for each voxel simultaneously: reflection intensity, two-way travel time, and instantaneous spectral bandwidth, rather than relying solely on a single reflection intensity index. The reason for using these three feature values ​​for joint discrimination is that strong reflections in underground media are not limited to cavities. Buried pipelines, aquifer variations, and differences in backfill materials can all cause anomalies in high reflection intensity. Using only reflection intensity as the discrimination criterion would introduce a large number of false positives. Reflection intensity characterizes the impedance difference at the medium interface. The wave impedance difference between a cavity and the surrounding dense soil is usually large, therefore the reflection intensity at the cavity boundary is significantly higher than that at a typical layered interface. Two-way travel time characterizes the cavity depth. Cavities are usually located within a specific depth range, and two-way travel time can eliminate interference reflections from shallow surfaces and deep bedrock reflections. Instantaneous spectral bandwidth characterizes the scattering characteristics of the cavity boundary. The irregular boundary of a cavity produces a scattering effect on elastic waves, causing the spectrum of the reflected signal to broaden, while the spectral bandwidth of the reflected signal at a smooth layered interface is relatively narrow. The three feature values ​​characterize the differences between void and non-void regions from different physical dimensions, and their combined use can significantly improve the reliability of void identification.

[0030] The method for extracting reflection intensity is to directly read the offset imaging amplitude value stored in the three-dimensional voxel model of reflection intensity for each voxel, and take its absolute value as the reflection intensity feature value of that voxel.

[0031] The two-way travel time extraction method is based on the vertical depth coordinates of each voxel. and elastic wave propagation speed ,according to Calculate the two-way travel time eigenvalue corresponding to this voxel. For two-way travel, The vertical depth of the voxel center point. Let be the elastic wave propagation velocity at this depth. When there is a longitudinal variation in the stratum velocity, the calculation of the two-way travel time requires integrating the velocity profile from the surface to the voxel depth.

[0032] The instantaneous spectral bandwidth is extracted by extracting a short time window segment centered on the time corresponding to the voxel from the seismic trace signal after migration imaging at the spatial location of each voxel. This short time window signal is then subjected to a Fourier transform to obtain its spectrum, and the equivalent bandwidth of the spectrum is calculated as the instantaneous spectral bandwidth characteristic value for that voxel. In one embodiment, the length of the short time window is set to the time length corresponding to 2 to 4 cycles of the source's dominant frequency. For a 200 Hz source-detector combination, one cycle is 5 milliseconds, and the short time window length is 10 to 20 milliseconds. The spectral equivalent bandwidth is calculated using the spectral moment method, that is, using the power of each frequency component of the spectrum as a weight, the weighted standard deviation of the frequencies is calculated, and this standard deviation is the instantaneous spectral bandwidth value. The scattering effect at the cavity boundary results in the reflected signal containing richer frequency components, and its instantaneous spectral bandwidth is usually more than 0.5 times the source's dominant frequency, while the spectral bandwidth of the reflected signal from layered interfaces is generally only 0.2 to 0.3 times the source's dominant frequency.

[0033] After extracting the three feature values, a cavity candidate detection is performed on each voxel. An anomaly threshold is set for each feature value, and voxels whose three feature values ​​all exceed their respective set anomaly thresholds are marked as cavity candidate voxels. The anomaly thresholds are set as follows: statistical analysis is performed on the same feature value of all voxels in the three-dimensional voxel model of reflection intensity, calculating its mean and standard deviation. The mean plus a certain multiple of the standard deviation is used as the anomaly threshold for that feature value. In one embodiment, the anomaly threshold for reflection intensity is set to the mean of reflection intensity of all voxels plus 2 times the standard deviation; the anomaly threshold for two-way travel time is set to the travel time interval corresponding to the target detection depth range. For example, for a scenario with a detection depth of 0.5 meters to 8 meters, under the condition that the elastic wave propagation speed is 300 meters per second, the anomaly threshold range for two-way travel time is 3.3 milliseconds to 53.3 milliseconds; and the anomaly threshold for instantaneous spectral bandwidth is set to the mean of instantaneous spectral bandwidth of all voxels plus 1.5 times the standard deviation. Voxels with only some eigenvalues ​​exceeding the anomaly threshold are marked as suspected voxels. Suspected voxels represent areas that may belong to the transition zone of cavities, but the evidence is not yet sufficient. These voxels will be further processed in the subsequent partitioning and merging stage based on their spatial relationship with candidate cavity voxels. Voxels with none of their three eigenvalues ​​exceeding their respective anomaly thresholds are considered normal soil areas and are excluded in subsequent processing.

[0034] After all voxels have been identified as candidates, a 3D connected component search is performed on all void candidate voxels. The purpose of the 3D connected component search is to group spatially adjacent void candidate voxels into the same connected region. Each connected region represents a spatially continuous void, corresponding to the area that needs to be filled as a whole in the engineering process. The 3D connected component search is implemented using breadth-first search or depth-first search. The specific process is as follows: Starting from any void candidate voxel that has not yet been assigned to a connected region, its neighboring voxels in the 3D voxel grid are checked. The adjacency relationship is defined using the 6-neighbor connectivity criterion, that is, each voxel has 2 direct neighbors in the vertical, horizontal, and vertical directions, for a total of 6 neighboring voxels. If a neighboring voxel is also a void candidate voxel and has not yet been assigned to a connected region, it is added to the current connected region, and the process continues to check its neighboring voxels from that voxel. This process is repeated until no new void candidate voxels are found in the 6-neighborhood of all void candidate voxels in the current connected region. Then, the next starting point is selected from the remaining unassigned candidate voxels for voids, and the search for the next connected region begins. This process is repeated until all candidate voxels for voids are assigned to their respective connected regions. In an alternative implementation, the connected region search can also employ the 26-neighborhood connectivity criterion, which includes 12 edge neighbors and 8 corner neighbors in addition to 6 face neighbors, for a total of 26 neighboring voxels. The 26-neighborhood criterion has better connectivity capture capability for irregularly shaped voids, but the computational cost increases accordingly. Each connected region constitutes an independent filling zone, which is then treated as an independent unit for airbag isolation and foam concrete filling in subsequent steps.

[0035] After completing the connected component search, suspected voxels spatially adjacent to the connected regions are merged. The principle of merging is to calculate the ratio of the adjacency area between each suspected voxel and its adjacent connected regions to the total surface area of ​​the suspected voxel. When this ratio exceeds a set merging ratio threshold, the suspected voxel is merged into the adjacent filling partition for filling volume calculation. The adjacency area refers to the sum of the areas of the voxel surfaces shared by the suspected voxel and the candidate voxels for voids in the adjacent connected regions. For example, when the voxel side length is 0.2m x 0.2m x 0.1m, the total surface area of ​​a single voxel is... If a suspected voxel has two faces adjacent to a candidate voxel with a cavity in the same connected region within its 6-neighborhood, and these two shared faces are both 0.2m x 0.1m side faces, then the adjacent area is [area missing]. square meters, the proportion of adjacent area to the suspected voxel surface area is That is, 25%. In one implementation, the merging ratio threshold is set to 30%, meaning that when the adjacent area of ​​a suspected voxel to a connected region reaches or exceeds 30%, the suspected voxel is considered to be highly likely to belong to the edge of the void region, and is merged into the filling partition to ensure sufficient coverage of the void edge during filling. When a suspected voxel is adjacent to multiple connected regions simultaneously, it is merged into the filling partition corresponding to the connected region with the largest adjacent area ratio. In another implementation with more stringent filling density requirements, the merging ratio threshold can be reduced to 20% to include more suspected edge regions.

[0036] The estimated total volume of voids in each filling zone is obtained by summing the voxel volumes of all voxels contained within the filling zone. The voxel volume of each voxel is the product of its side lengths in the longitudinal, transverse, and vertical directions. Under the above voxel size settings, the voxel volume of each voxel is... The volume of each cavity candidate voxel within the filling zone is summed, along with the volumes of any suspected voxels merged after merging, to obtain the estimated total volume of cavities in that filling zone. The estimated total volume is then multiplied by a filling factor to obtain the target filling volume for that filling zone. A filling factor greater than 1 is used to compensate for volume estimation errors caused by 3D migration imaging and voxel segmentation. Inaccuracies in the velocity model during 3D migration imaging can lead to deviations in the imaging position of cavity boundaries, and the binarization of threshold determination during voxel segmentation can cause underestimation or overestimation of the volume in the cavity edge region. Setting a filling factor greater than 1 can, to some extent, compensate for these systematic errors and prevent the actual filling volume from being insufficient to fill the cavity. In one implementation, the filling factor is set to 1.15, meaning the target filling volume is 15% greater than the estimated total volume of cavities. In cases of complex cavity shapes and irregular walls, the filling factor can be increased to 1.25 to 1.30 to allow for a larger margin. When the cavity shape is relatively regular and the imaging quality is high, the filling factor can be appropriately reduced to 1.10.

[0037] After determining the target filling volume for each filling zone, the filling priority order is determined based on the burial depth and estimated total volume of cavities in each filling zone. The filling priority is determined according to the principle of prioritizing filling zones with shallow burial depth and large volume. The engineering basis for this is that shallow cavities are closer to the surface of the target area, have thinner overlying soil layers, and are at higher risk and more urgent to address under traffic loads, thus requiring priority treatment. Under similar burial depth conditions, larger cavities have a more significant weakening effect on the bearing capacity of the geological structure and should also be prioritized for filling. In one implementation, the specific sorting method for filling priority is as follows: for each filling zone, calculate the average burial depth and estimated total volume of cavities at its center point. First, sort by average burial depth from shallowest to deepest. When the difference in average burial depth among multiple filling zones is within 0.5 meters, they are considered to be in the same burial depth level. Within the same burial depth level, sort by estimated total volume of cavities from largest to smallest. In another implementation, corresponding scores can be assigned to the burial depth and the estimated total volume of the cavity, and the two scores can be added together and sorted from high to low according to the total score to determine the priority order of filling. The score setting can be adjusted according to the needs of the geological structure safety assessment in actual engineering.

[0038] After completing the cavity detection and filling zone division in step 1, foam concrete filling construction is carried out on each filling zone in sequence according to the filling priority. The construction process of each filling zone includes three steps: placement and sealing effectiveness verification of reusable flexible airbags, layout of filling pipelines and slurry cross-contamination detection devices, and foam concrete filling control based on slipform controller. The process is repeated for the next filling zone only after the current filling zone is filled and the airbag is recovered.

[0039] First, the placement process of the reusable flexible airbag is explained. Before filling the current filling zone with foamed concrete, a physical isolation boundary needs to be established at the junction of the current filling zone and the adjacent filling zone to prevent the injected foamed concrete slurry from flowing into the adjacent unfilled filling zone, causing slurry cross-contamination and inaccurate volume control. Physical isolation is achieved through reusable flexible airbags. After inflation, the reusable flexible airbags can form a sealing barrier tightly against the cavity wall, blocking the slurry flow path between the current filling zone and the adjacent filling zone. One airbag placement hole is drilled at the junction of the current filling zone and each adjacent filling zone. The airbag placement hole is drilled vertically downward from the surface of the target area until it penetrates into the cavity. The hole diameter is determined according to the outer diameter of the reusable flexible airbag after folding. In one embodiment, the hole diameter is set to 150 mm, and the drilling depth is determined according to the burial depth of the filling zone junction section determined in step 1. After drilling is completed, the retrievable flexible airbag is folded and inserted into the boundary section inside the cavity through the airbag placement hole, where it unfolds. During insertion, a guide rod or flexible push tube is used to guide the folded airbag to the center area of ​​the boundary section, ensuring that the unfolded airbag covers most of the boundary section area.

[0040] refer to Figure 2 The spatial arrangement of the components of the filling construction system inside the cavity and on the road surface is shown in the form of a cross-section of the target area. Figure 2The top layer is the pavement structure layer, indicated by parallel diagonal lines. Below the pavement structure layer is the target area soil, containing two irregularly shaped voids, labeled Filling Zone I and Filling Zone II. Filling Zone I is located on the left side of the profile; its lower half is indicated by dotted filling of injected foamed concrete, with the foamed concrete level marked by a horizontal line. The space above the level is not yet filled with grout. Filling Zone II is located on the right side of the profile, marked as awaiting filling, and contains untreated voids filled with air. A retrievable flexible airbag is located at the junction of the two filling zones. The airbag body is vertically arranged along the junction section, with a slightly convex cross-section in the middle, indicating a sealed state formed by the airbag inflating and adhering to the void wall. Three pressure sensors are located on the outer surface of the airbag, distributed at 120-degree angular intervals along the circumference of the airbag's maximum cross-section, each pressure sensor marked with a small square. The recyclable flexible airbag is connected to the air pump control valve assembly above the road surface via an airbag placement hole. The airbag placement hole is drilled vertically downwards from the surface of the target area to the boundary section inside the cavity, indicated by double solid lines on the hole wall. A high-pressure air pipe extends along the airbag placement hole above the ground, indicated by dashed lines. The upper end of the high-pressure air pipe connects to a rectangle labeled the air pump control valve assembly. At the top of filling zone I, vent holes and filling holes are drilled from the surface of the target area. The vent hole is located at the end of filling zone I furthest from the recyclable flexible airbag, indicated by a thinner double line, its diameter being smaller than the filling hole. The filling hole is located above the centroid of filling zone I, indicated by a thicker double line. The filling pipeline extends downwards from the filling hole to near the bottom of the cavity, with arrows indicating the injection direction of the foamed concrete slurry. A pressure transmitter and an electromagnetic flowmeter are sequentially installed at the orifice section of the filling pipeline, indicated by a box labeled "P" and a circle labeled "Q," respectively. A displacement sensor, represented by a triangle, is installed at the surface of the target area of ​​the filling hole to collect the rebound displacement of the road surface in real time during the filling process. An electrical conductivity probe, represented by a solid circle, is installed inside filling zone II near the back of the recyclable flexible airbag to monitor whether foamed concrete slurry has leaked from filling zone I into filling zone II. Figure 2 The left side is marked with a depth scale, ranging from 0 meters to 8 meters, indicating the dimensional relationship of the target area in the depth direction. Gray dashed arrows indicate the flow direction of the grout inside the cavity, spreading downwards and laterally from the filling pipe outlet.

[0041] The recyclable flexible airbag is made of alkali-resistant rubber. Alkali-resistant material is chosen because the slurry of foamed concrete is alkaline. During cement hydration, a large amount of calcium hydroxide is produced, resulting in a pH of typically between 12 and 13. Ordinary rubber swells and degrades in mechanical properties after prolonged contact with alkaline slurry, leading to leakage or even rupture during filling. Alkali-resistant rubber maintains sufficient elasticity and strength in such alkaline environments. In one embodiment, the airbag wall thickness is set to 3 to 5 millimeters. In its uninflated state, the airbag can be folded to a diameter of less than 120 millimeters to pass through the airbag placement hole. After inflation, it can accommodate irregular void shapes with a cross-sectional diameter ranging from 300 to 1500 millimeters. Three contact pressure sensors are evenly distributed circumferentially on the outer surface of the airbag, spaced at 120-degree angular intervals along the circumference of the airbag's maximum cross-section. The pressure-sensing surface of each sensor faces outwards, used to detect the contact pressure between the outer surface of the airbag and the void wall. Three contact pressure sensors, instead of one, are used to determine whether the bladder is fully in contact with the cavity wall in different directions. The cavity cross-section is usually not a regular circle; there may be pits or protrusions on the wall in some directions. A single sensor can only reflect the local contact status, while three sensors evenly distributed circumferentially can cover the main directions of the bladder. Only when the readings of all three contact pressure sensors reach the set value can it be determined that the bladder has achieved full-range contact. In one embodiment, the range of the contact pressure sensors is set to 0 to 0.5 MPa, with an accuracy of 0.5% of the range, i.e., 0.0025 MPa. The bladder is connected to a pump control valve assembly on the ground via a high-pressure air pipe. The high-pressure air pipe extends from the bladder placement hole to the ground. The pump control valve assembly includes an air pump, a solenoid valve, a pressure gauge, and a safety relief valve, enabling precise control of the bladder's inflation, pressure holding, and depressurization processes.

[0042] The air pump control valve assembly is activated to inflate the bladder with compressed air, causing it to gradually expand inside the cavity until the outer surface of the bladder adheres to the cavity wall. During inflation, a staged, slow pressurization method is used to avoid excessive local stretching or rupture of the bladder due to sudden excessive pressure. In one embodiment, the inflation process is divided into three stages: the first stage increases the internal air pressure from 0 to 0.05 MPa at a rate of 0.005 MPa per second, causing the bladder to initially expand and form slight contact with the cavity wall; the second stage increases the internal air pressure from 0.05 MPa to 0.15 MPa at a rate of 0.003 MPa per second, causing the bladder to further expand and adhere tightly to the wall; the third stage applies fine pressure at a rate of 0.001 MPa per second until the adhesion criteria are met. During inflation, the real-time pressure values ​​of three bonding pressure sensors are continuously read. When the real-time pressure values ​​of all three sensors reach the preset bonding pressure and the fluctuation amplitude is less than the preset stability deviation within 10 consecutive seconds, the air pump control valve group is closed and the air pressure inside the bladder is locked. In one embodiment, the preset bonding pressure is set to 0.08 MPa. This value is selected considering both the sealing effect of the bladder on the wall and the disturbance effect on the soil of the cavity wall. Too low a bonding pressure will lead to poor sealing and grout leakage, while too high a bonding pressure may cause excessive lateral compression of the wall soil, leading to local collapse. In one embodiment, the preset stability deviation is set to 0.005 MPa, meaning that within a 10-second observation window, the fluctuation amplitude of the real-time pressure value of any of the three bonding pressure sensors relative to its average value within the observation window should be less than 0.005 MPa. This condition ensures that the bladder is in a stable bonding state and is not still adjusting for deformation. In another case where the cavity wall conditions are poor and the surface roughness is high, the preset bonding pressure can be increased to 0.12 MPa and the preset stability deviation can be relaxed to 0.008 MPa.

[0043] After shutting off the air pump control valve assembly and locking the air pressure inside the bladder, a sealing effectiveness verification is performed. The purpose of this verification is to confirm that a reliable seal has been formed between the bladder and the cavity wall, ensuring it can withstand the pressure of the foamed concrete slurry during subsequent filling operations without leakage. The specific procedure for sealing effectiveness verification is as follows: After locking the air pressure inside the bladder, the current air pressure value is recorded as the locking reference value. A preset pressure holding observation period is then maintained to detect whether there is a leakage channel between the bladder and the cavity wall. The basic logic of the pressure holding observation is that if a leakage channel exists between the bladder and the wall, the gas inside the bladder will slowly leak out through this channel, causing the air pressure inside the bladder to decrease over time. By monitoring the change in air pressure during the pressure holding period, the reliability of the seal can be determined. In one implementation, the pressure holding observation period is set to 60 seconds, which is sufficient for even a small leakage channel to cause a detectable pressure drop. After the pressure holding observation period ends, the air pressure value inside the bladder is read again, and the pressure drop relative to the locking reference value is calculated. If the pressure drop is less than the preset allowable pressure drop threshold, the sealing is considered effective. In one embodiment, the allowable pressure drop threshold is set to 0.005 MPa, meaning that if the pressure drop inside the bladder does not exceed 0.005 MPa within a 60-second pressure holding observation period, the seal is considered reliable. If the pressure drop reaches or exceeds the allowable pressure drop threshold, it indicates a significant leakage path between the bladder and the wall surface. This could be due to localized fragments protruding from the wall surface preventing the bladder from fully adhering, or loose soil on the wall surface causing gas to escape from the pores. In this case, the air pump control valve assembly is restarted to replenish the pressure. By further increasing the pressure inside the bladder, the bladder is compressed and deformed to better adapt to the irregular shape of the wall surface, thus sealing the leakage path. After replenishment, the pressure value inside the bladder is recorded again as the new locking reference value, and the pressure holding observation is repeated until the sealing is deemed effective. In one embodiment, the pressure increment during replenishment is 0.02 MPa each time. If three consecutive pressurization attempts still fail to pass the sealing effectiveness verification, consider removing the retrievable flexible airbag and replacing it with a larger or thicker one. In another implementation, the pressure observation time can be extended to 120 seconds, and the allowable pressure drop threshold can be tightened to 0.003 MPa to accommodate construction scenarios with higher sealing requirements.

[0044] After the sealing effectiveness verification is passed, the next step is to install the detection device for cross-contamination between the filling pipeline and the grout. Filling holes and vent holes are drilled at the top of the current filling zone. The filling hole is the only channel for the foamed concrete grout to inject into the voids, and its location is chosen near the centroid of the current filling zone on the horizontal projection plane to facilitate uniform diffusion of the grout from the center outwards. The vent hole is located at the end of the current filling zone furthest from the filling hole. It is used to expel air from the filling zone during the filling process and serves as an observation channel for grout overflow at the end of the filling process. The vent hole is located at the end furthest from the filling hole because the foamed concrete grout gradually flows outwards after being injected through the filling hole, squeezing the air in the voids towards the far end. The far-end vent hole provides the longest path for air to escape. When the grout finally flows to the far end and overflows from the vent hole, it can be determined that the entire filling zone is filled with grout, and there are no unfilled dead zones caused by trapped gas. In one embodiment, the diameter of the filling hole is set to 110 mm, and the diameter of the vent hole is set to 50 mm. The vent hole has a smaller diameter because it is only needed to vent air and observe a small amount of slurry overflow, and does not need to be used as a slurry injection channel.

[0045] The injection pipeline is inserted into the current injection zone through the injection hole, with the lower end of the pipeline extending near the bottom of the cavity. This allows the foamed concrete slurry to gradually fill the cavity from the bottom upwards. This bottom-up filling method helps reduce the mixing of air bubbles in the slurry and the residual air in the upper part of the cavity. A pressure transmitter and an electromagnetic flowmeter are installed sequentially at the orifice section of the injection pipeline, and a displacement sensor is installed at the ground surface at the orifice opening of the injection hole. The pressure transmitter is used to collect the injection pressure value in real time. In one embodiment, its range is set to 0 to 1.0 MPa, with an accuracy of 0.25% of the range, i.e., 0.0025 MPa. The electromagnetic flowmeter is used to collect the instantaneous injection flow value in real time. Foamed concrete slurry is a conductive fluid. The electromagnetic flowmeter uses Faraday's law of electromagnetic induction to measure the induced electromotive force generated by the flow of conductive liquid in a magnetic field to obtain the flow rate. In one embodiment, its range is set to 0 to 50 liters per minute, with an accuracy of 0.5% of the reading. A displacement sensor is installed at the surface of the grouting hole to collect real-time pavement rebound displacement values, which are the upward bulging deformation of the pavement under the pressure of the foamed concrete grout during the grouting process. Monitoring the pavement rebound displacement is crucial for ensuring grouting safety. Excessive injection pressure can cause the overlying soil layer in the cavities to bulge upwards under the pressure of the grout, potentially leading to pavement cracking or bulging in severe cases. In one embodiment, the displacement sensor is a linear variable differential transformer type sensor with a range of 0 to 30 mm and a resolution of 0.01 mm.

[0046] Simultaneously, a slurry cross-contamination detection device is deployed. In an unfilled slurry zone adjacent to the current slurry zone, a conductivity probe is installed near the back of the reusable flexible airbag, using either an existing airbag placement hole in that adjacent slurry zone or a separately drilled monitoring hole. The conductivity probe is installed immediately adjacent to the back of the reusable flexible airbag, i.e., facing the adjacent slurry zone, and the distance between the probe and the reusable flexible airbag is controlled within the range of 50 mm to 150 mm in one embodiment. The conductivity probe is used to detect whether the foamed concrete slurry bypasses or penetrates the reusable flexible airbag and enters the adjacent slurry zone. The basis for using conductivity testing as a means of detecting grout cross-contamination lies in the fact that the interior of cavities before filling is filled with air or dry, loose debris, which has extremely low conductivity, typically below 0.001 Siemens per meter. In contrast, foamed concrete grout contains a large number of ionic metal cations such as calcium, potassium, and sodium, as well as hydroxide anions, with a conductivity typically between 0.5 and 2.0 Siemens per meter—a difference of more than two orders of magnitude compared to air. When grout crosses from the current filling zone into an adjacent filling zone and reaches the location of the conductivity probe, the conductivity of the medium surrounding the probe will experience a sudden increase. By monitoring the change in conductivity value, the occurrence of grout cross-contamination can be detected in real time. In one embodiment, the conductivity probe's range is set to 0 to 5.0 Siemens per meter, with a response time of less than 1 second.

[0047] After the grouting pipeline and sensors are installed, the foamed concrete grouting construction begins. A variable frequency screw pump is used as the grouting power source, injecting foamed concrete into the current grouting zone through the grouting pipeline. The selection of the variable frequency screw pump is based on the fact that it minimizes the damage to air bubbles in the foamed concrete slurry, maintaining the design density of the foamed concrete without significant changes. Simultaneously, the variable frequency drive allows for continuous changes in pump speed and displacement by adjusting the motor frequency, providing precise execution capability for the subsequent slipform controller output commands. In one embodiment, the rated displacement of the variable frequency screw pump is set to 40 liters per minute, with a variable frequency speed range of 10% to 100% of the rated speed, corresponding to a displacement adjustment range of 4 liters per minute to 40 liters per minute. In one embodiment, the wet density of the foamed concrete is set to 500 kg / m³ to 700 kg / m³, and the flowability is controlled between 160 mm and 200 mm to ensure that the slurry has good self-leveling properties and can fully diffuse within the cavities.

[0048] During the injection process, a sliding mode controller is used to control the variable frequency screw pump in real time. The sliding mode controller operates cycle by cycle with a fixed control period. In one embodiment, the duration of the fixed control period is set to 1 second, that is, a complete data acquisition, status calculation, and command output cycle is performed every 1 second. In another embodiment, where higher control precision is required, the control period can be shortened to 0.5 seconds.

[0049] Within each control cycle, the sliding mode controller first performs the calculation and verification of the cumulative injection volume. It then reads the instantaneous injection flow rate value for the current control cycle from the electromagnetic flowmeter and records it as... subscript Indicates the sequence number of the current control cycle. The unit is liters per minute. Converted to the volume increment injected within the current control cycle ,in The duration of the fixed control cycle is in seconds. Dividing by 60 converts liters per minute to liters per second. The current cumulative injection volume is obtained by adding it to the cumulative injection volume accumulated in previous control cycles. ,in This represents the cumulative injection volume at the end of the previous control cycle. .

[0050] Because electromagnetic flowmeters may experience zero drift or transient measurement deviations due to air bubbles in the slurry during long-term operation, relying solely on flow integral calculations to determine the cumulative injection volume may introduce gradually accumulating errors. Therefore, the rotational speed information of a variable frequency screw pump is introduced as a cross-verification method. Simultaneously, the actual rotational speed of the variable frequency screw pump in the current control cycle is read and recorded as follows: The unit is revolutions per minute (rpm). This is based on the single-revolution displacement of the variable frequency screw pump. (This refers to the theoretical volume of slurry discharged per revolution of the pump rotor, expressed in liters per revolution. This value is determined by the pump's geometric parameters and is an inherent factory parameter.) The actual rotational speed is then converted into an estimated pump discharge volume. ,in This is an estimate of the total discharge volume calculated cumulatively by pump speed. The estimated pump discharge volume... Compared with the current cumulative injection volume Compare the two; if the absolute value of the deviation is... Exceeding the preset integral verification deviation threshold Then, the current cumulative injection volume is corrected using the estimated pump discharge volume, that is, let Integral verification deviation threshold In one implementation, the correction is set to the larger of 5% or 0.5 liters of the current cumulative injection volume. Choosing the larger value avoids frequent correction triggers when the cumulative injection volume is small in the initial stages of filling. If the deviation between the two values ​​is within the integral verification deviation threshold, the current cumulative injection volume is used directly for the next control calculation. In an optional implementation, the correction method can also use a weighted average instead of direct replacement, i.e. ,in To correct the weighting coefficient, the value ranges from 0.5 to 0.8. This method can correct the flow integral deviation using pump discharge information while avoiding the introduction of new errors due to fluctuations in pump volumetric efficiency.

[0051] After completing the calculation and verification of the cumulative injection volume, the sliding mode controller calculates the volume deviation between the corrected current cumulative injection volume and the target filling volume in each control cycle, denoted as . ,in The target filling volume determined in step 1, This represents the current cumulative injected volume and volume deviation. The physical meaning is the remaining volume that still needs to be injected in the current filling zone. Simultaneously, the change in volume deviation in the current control cycle relative to the volume deviation in the previous control cycle is calculated to obtain the volume deviation change rate, denoted as . ,in This is the volume deviation from the previous control cycle. To control cycle duration, the rate of change of volume deviation is used. This reflects the rate at which the remaining volume decreases during the filling process, and during normal filling... The value is negative, and the larger the absolute value, the faster the filling speed.

[0052] The weighted combination of volume deviation and the rate of change of volume deviation is used as the sliding mode variable, denoted as . ,in It is a positive constant used to adjust the contribution of the volume deviation rate to the sliding mode variable. The value of affects the dynamic response characteristics of the sliding mode controller. A larger value results in a more sensitive controller to changes in deviation, which helps to quickly suppress fluctuations in the injection rate. However, excessively large values ​​can lead to problems. This value may exacerbate chattering in control commands; The smaller the value, the more the controller relies on volume deviation for adjustment, resulting in a smoother response but potentially overshoot. In one implementation, A value of 5 seconds is set such that when the absolute value of the volume deviation change rate reaches 20% per second of the remaining deviation, the contribution of the change rate component and the deviation component to the sliding mode variable is roughly equal. In another case where the cavity volume is large and the filling time is long, The time can be increased to 8 seconds to enhance the smoothness of control.

[0053] Sliding mode variable The sign of the sliding form variable directly reflects the status of the filling process. When the sliding form variable is greater than zero, it indicates that the filling process is still in the stage of approaching the target filling volume, that is, the voids have not yet been filled and foamed concrete needs to be injected. When the sliding form variable is equal to zero or less than zero, it indicates that the current cumulative injection volume has reached the target filling volume and the filling task is completed.

[0054] When the sliding mode variable is greater than zero, the sliding mode controller enters the approach phase. During each control cycle of the approach phase, the sliding mode controller simultaneously reads the current injection pressure value acquired by the pressure transmitter. The current road surface rebound displacement value collected by the displacement sensor and the current conductivity value collected by the conductivity probe. The system compares these three real-time measurements with their respective preset safety limits and executes different control actions based on the comparison results.

[0055] If the current injection pressure value Below the pressure safety limit Current road surface rebound displacement value Below the upper limit of displacement safety And the current conductivity value Below the cross-conductivity threshold If all three safety indicators are within the normal range, the sliding mode controller outputs a positive pump speed increment command according to the constant velocity approach method, driving the variable frequency screw pump to maintain or increase the injection flow rate. The constant velocity approach method means that the amplitude of the pump speed increment command output by the sliding mode controller remains constant, denoted as . ,in A positive constant is called the approach velocity parameter, and its unit is revolutions per minute per control cycle. For sliding mode variables The sign function, in the approach phase Therefore When the pump speed increment command is positive, the speed of the variable frequency screw pump increases or remains constant. (Approaching speed parameter) In one implementation, the value is set to 5 revolutions per minute per control cycle, which corresponds to an increase in pump displacement of approximately [missing value] per control cycle. Increase the traffic increment per minute, in The flow rate increment is 0.25 liters per minute per revolution, a relatively gentle adjustment range that avoids drastic flow fluctuations. After receiving the pump speed increment command, the variable frequency screw pump updates the current speed setpoint to... The motor is then driven to the new speed setpoint via a frequency converter. When the speed setpoint reaches the pump's maximum allowable speed, the pump speed increment command is limited, and the speed is maintained at its maximum value. In another implementation, an exponential approach can be used instead of a constant-speed approach, where the amplitude of the pump speed increment command varies with the sliding mode variable. The rate of injection decreases synchronously with the decrease in volume, which automatically slows down the injection rate as the filling process approaches the target filling volume, reducing the risk of overshoot when the target volume is reached.

[0056] If the current injection pressure value Reaching the pressure safety limit Or the current road surface rebound displacement value Reaching the upper limit of displacement safety This means that continuing to inject grout at the current rate will pose a safety hazard to the geological structure. Excessive injection pressure may cause grout to leak from the gap between the injection pipe and the borehole wall to the surface, or it may cause the soil on the cavity wall to split and expand; excessive road surface rebound displacement indicates that the overlying soil layer of the cavity has begun to bulge upwards, and continued pressure may lead to damage to the road structure. At this point, the slipform controller switches the pump speed increment command to a negative value, i.e., outputs... The output flow rate of the variable frequency screw pump is forcibly reduced. As the pump speed continuously decreases, the injection pressure and road surface rebound displacement gradually decrease accordingly. In each subsequent control cycle, the sliding mode controller continuously monitors the current injection pressure and road surface rebound displacement values. Once both values ​​fall below their respective safety limits, it resumes a positive approach, i.e., it re-outputs a positive pump speed increment command to continue filling. This safety constraint mechanism allows the filling process to automatically decelerate when approaching the safety boundary and automatically resume when moving away from it, forming an adaptive safety protection. Pressure safety limit. In one implementation, the pressure safety limit is set to 0.3 MPa. This value is determined based on the thickness of the overlying soil layer and the soil strength; the thinner the overlying soil layer or the lower the soil strength, the lower the pressure safety limit should be set. Displacement safety limit. In one implementation, the safety limit is set at 5 mm; road surface heave exceeding this value may cause cracking of the asphalt surface layer. In another scenario where the overlying soil layer is thicker and the road structure has higher bearing capacity, the upper limit for pressure safety can be relaxed to 0.5 MPa, and the upper limit for displacement safety can be relaxed to 8 mm.

[0057] If the current conductivity value Reaching or exceeding the crosstalk conductivity threshold If this is the case, then it is determined that cross-contamination has occurred in the slurry. Cross-contamination conductivity threshold. In one implementation, the conductivity is set to 0.1 Siemens per meter. This value is much lower than the conductivity of the foamed concrete slurry itself but much higher than the background conductivity of air or dry soil, reliably distinguishing between normal conditions and slurry arrival events. When the conductivity probe detects slurry cross-contamination, the sliding form controller immediately outputs a pump stop command to pause filling, preventing more slurry from flowing into adjacent filling zones and causing uncontrollable slurry loss and inaccurate volume estimation of adjacent filling zones. After pausing filling, the construction personnel reinforce or replace the recyclable flexible airbags by adding air pressure to the airbags through the air pump control valve group to enhance the sealing effect, or, if pressurization still cannot solve the cross-contamination problem, remove the airbags and replace them with new airbags of a larger size. After the cross-contamination problem is resolved, filling is restarted and sliding form control continues from the cumulative injection volume at the time of the pause. That is, the sliding form controller does not reset the cumulative injection volume to zero but continues to operate based on the volume accumulated before the pause, and the injected slurry volume is still included in the filling volume of the filling zone.

[0058] When sliding mode variable When the volume is equal to or less than zero, it is determined that the current cumulative injection volume has reached the target filling volume, and the sliding mode controller outputs a pump stop command to shut down the variable frequency screw pump. After the pump stops, the filling is confirmed by observing whether foamed concrete grout overflows from the vent hole. If grout overflows from the vent hole, it indicates that the space inside the cavity has been completely filled with grout, and the overflowing grout indicates that there is no room for the excess volume. If no grout overflow is observed from the vent hole when the pump stops, it may be that there is a deviation between the cavity shape and the estimation of the 3D offset imaging, resulting in the actual cavity volume being larger than the target filling volume. In this case, a small amount of injection can be continued as appropriate until grout overflows from the vent hole.

[0059] After full filling, the air pump control valve group is activated to depressurize the recyclable flexible airbag. The depressurization process also employs a slow depressurization method to avoid the instantaneous contraction of the airbag causing adsorption and tension on the not-yet-fully-cured foamed concrete. In one embodiment, the depressurization rate is set to 0.005 MPa per second. After the airbag has fully contracted, the recyclable flexible airbag is retrieved to the ground through the airbag placement hole, and the conductivity probe is removed simultaneously. The retrieved recyclable flexible airbag, after cleaning and inspection, can be reused in the construction of the next filling section. This recyclability reduces construction material costs. At this point, the filling construction of the current filling section is complete. Following the filling priority order, the above operations—airbag placement, sealing effectiveness verification, installation of the filling pipeline and slurry cross-contamination detection device, and slipform controlled filling—are repeated for the next filling section until all filling sections are completed.

[0060] After all filling zones are filled, the next stage is the evaluation and inspection management of the filling effect. The purpose of the filling effect evaluation is to verify whether the foamed concrete has fully filled the voids in each filling zone, identify any areas that may be underfilled, and provide a monitoring and early warning basis for the subsequent safe operation and maintenance of the target area.

[0061] Re-measurement is conducted along the same elastic wave exploration line as in step 1 to obtain post-filling reflection profile data. The same source frequency, channel spacing, and sampling parameters as in step 1 are used for the re-measurement, and data is strictly collected along the survey line in step 1 to ensure a one-to-one spatial correspondence between the two sets of data. The re-measurement is best performed after the foamed concrete has been injected and reached initial setting but before final setting. In one embodiment, the re-measurement is conducted within 24 to 72 hours after filling. At this time, the foamed concrete has a certain strength but its moisture content is still relatively high. Its wave impedance differs from that of dry foamed concrete but can be clearly distinguished from the air-filled state.

[0062] The reflection profile data after filling and the elastic wave reflection profile data obtained in step 1 are cross-correlated at each measuring point to generate a filling effect score distribution map along the measuring line. The basic idea of ​​structural similarity calculation is to perform waveform cross-correlation calculation on two seismic trace signals at the same measuring point before and after filling to quantify the similarity between the two in waveform structure. Before filling, cavities exhibit strong reflection anomalies. After filling with foamed concrete, the cavity area is filled with dense material, the medium interface disappears or is significantly weakened, and the waveform of the seismic trace signal at the corresponding location will change significantly. In one implementation, structural similarity calculation is achieved using normalized cross-correlation coefficients. The seismic trace signal at the same measuring point before filling is cross-correlated. and the seismic trace signal after filling After extracting signal segments within the same time window, the normalized cross-correlation values ​​of the two signal segments are calculated. ,in and Signals and signal The mean within the time window and Signals and signal variance Let covariance be the variance of the two signals. and To prevent small positive numbers with a denominator of zero, normalize the cross-correlation values. The value range is from -1 to +1. When the two signals are exactly the same... The value equals 1, and the greater the difference, the better. The lower the value, the greater the signal difference before and after filling, in the context of filling effect evaluation (i.e., A lower value indicates a significant change in the state of the underground medium in the area, meaning the cavities have been successfully filled. The filling effect score at this measuring point is used as the rating, with a higher score indicating a better filling effect. In one embodiment, Set to the square of 1% of the root mean square amplitude of the signal. It is set to the square of 3% of the root mean square amplitude of the signal divided by 2.

[0063] After calculating the filling effect score for each measuring point along the entire measuring line, a filling effect score distribution map is generated. Sections with scores below a set acceptable threshold are marked as insufficiently filled sections. In one implementation, the acceptable threshold is set at 0.4; that is, when the filling effect score of a measuring point is below 0.4, the change in elastic wave signal before and after filling at that location is considered insufficient, potentially indicating incomplete filling of voids or insufficient filling density. In another implementation with higher filling quality requirements, the acceptable threshold can be increased to 0.5. Areas marked as insufficiently filled will be prioritized in subsequent inspections and management, with secondary filling arranged if necessary. In an optional implementation, in addition to structural similarity calculation, the filling effect evaluation can be supplemented by reflection energy attenuation rate analysis. This involves comparing the change in the total energy of the elastic wave reflection signal within the same section before and after filling. When the reflection energy attenuation rate exceeds a set proportion, the filling is confirmed as effective. Using both methods together improves the reliability of the evaluation results.

[0064] Simultaneously with the evaluation of the filling effect, a long-term settlement monitoring system is deployed. Settlement monitoring points are longitudinally arranged on the surface of the target area. The spacing of the settlement monitoring points is determined according to the distribution density of the filling zones. In one implementation, the spacing of the settlement monitoring points is set to 5 to 10 meters, and appropriately increased to 3 meters in concentrated areas of the filling zones. Each settlement monitoring point uses a stainless steel marker nail embedded in the pavement structure layer as an observation marker, and periodic observations are conducted in conjunction with a precision level. The cumulative settlement of each settlement monitoring point is collected according to the set monitoring cycle. The monitoring cycle is set to once a week for the first 3 months after the completion of filling, once a month for 3 months to 1 year, and once a quarter after 1 year. The cumulative settlement refers to the total settlement deformation of each settlement monitoring point since the first observation after the completion of filling. When the cumulative settlement of any settlement monitoring point exceeds the preset settlement alarm value, an early warning message is generated. In one implementation, the settlement alarm value is set at 10 mm. This value is determined based on the pavement structure type and traffic safety requirements. The permissible value for uneven settlement of asphalt pavement is typically between 10 mm and 20 mm, and the lower limit is taken as the alarm value to leave a safety margin. In another highway scenario, the settlement alarm value can be tightened to 8 mm.

[0065] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A method for detecting and assessing the quality and safety of filling underground cavities based on elastic wave exploration, characterized in that, Includes the following steps: Step 1: Deploy elastic wave exploration lines longitudinally and laterally along the target area. Collect elastic wave reflection profile data below the target area through seismic source excitation and detector reception. Perform three-dimensional migration imaging processing on the elastic wave reflection profile data to obtain a three-dimensional voxel model of the reflection intensity of the underground space in the target area. Simultaneously extract three feature values ​​for each voxel in the three-dimensional voxel model of reflection intensity: reflection intensity, two-way travel time, and instantaneous spectral bandwidth. Voxels whose three feature values ​​all exceed their respective set anomaly thresholds are marked as cavity candidate voxels. Perform a three-dimensional connected domain search on all cavity candidate voxels. Spatially adjacent cavity candidate voxels are merged into the same connected region. Each connected region constitutes an independent filling zone, and the target filling volume of each filling zone is determined. Step 2: After the filling medium is injected into each cavity area identified in Step 1, a re-measurement is carried out along the same elastic wave exploration line as in Step 1. The reflection profile data after filling is obtained by excitation of the seismic source and reception of the detector. The reflection profile data after filling is cross-correlation calculation is performed on the reflection waveforms of the elastic wave reflection profile data obtained in Step 1 at each measurement point to obtain the structural similarity and generate a filling effect score distribution map. The sections with scores below the set qualified threshold in the filling effect score distribution map are marked as insufficient filling sections. Settlement monitoring points are set up longitudinally on the surface of the target area. The cumulative settlement of each settlement monitoring point is collected according to the set monitoring cycle. When the cumulative settlement of any settlement monitoring point exceeds the preset settlement alarm value, an early warning information is generated. According to the location of the insufficient filling section and the early warning information, supplementary filling or shortened inspection interval is arranged for the corresponding filling zone. For filling zones where the filling effect score reaches the qualified threshold and the cumulative settlement is within the settlement alarm value, the regular inspection interval is maintained.

2. The method according to claim 1, characterized in that, In step 1, the reflection intensity characterizes the impedance difference at the medium interface, the two-way travel time characterizes the cavity depth, and the instantaneous spectral bandwidth characterizes the scattering characteristics of the cavity boundary; voxels with only some characteristic values ​​exceeding the abnormal threshold are marked as suspected voxels.

3. The method according to claim 2, characterized in that, In step 1, for suspected voxels adjacent to the connected region, it is determined whether to merge the suspected voxel into the adjacent filling partition based on whether the ratio of the adjacent area between the suspected voxel and the connected region to the surface area of ​​the suspected voxel exceeds a set merging ratio threshold. Suspected voxels whose ratio exceeds the merging ratio threshold are merged into the adjacent filling partition to participate in the filling volume calculation. The total volume of the void is estimated by summing the voxel volumes of all voxels contained in each filling partition. The total volume of the void is multiplied by the filling coefficient to obtain the target filling volume. The filling coefficient is greater than 1 and is used to compensate for the volume estimation error caused by 3D offset imaging processing and voxel segmentation. The filling priority order is determined according to the burial depth and the total volume of the void in each filling partition. Filling partitions with shallow burial depth and large volume are filled first.

4. The method according to claim 1, characterized in that, The filling medium injection process includes: drilling one airbag placement hole at the boundary between the current filling zone and the adjacent filling zone according to the filling priority order; folding the reusable flexible airbag body and sending it through the airbag placement hole to the boundary section inside the cavity and unfolding it; the reusable flexible airbag body is made of alkali-resistant rubber material, and three contact pressure sensors are evenly arranged circumferentially on the outer surface of the airbag body; the airbag body is connected to the air pump control valve group on the ground through a high-pressure air pipe; starting the air pump control valve group to inflate the airbag body into the airbag body, so that the airbag body gradually expands inside the cavity until the outer surface of the airbag body is in contact with the cavity wall; during the inflation process, the real-time pressure values ​​of the three contact pressure sensors are continuously read; when the real-time pressure values ​​of the three contact pressure sensors all reach the preset contact pressure and the fluctuation amplitude is less than the preset stable deviation within 10 consecutive seconds, the air pump control valve group is closed and the air pressure inside the airbag is locked.

5. The method according to claim 4, characterized in that, The verification of the sealing effectiveness is as follows: After locking the air pressure inside the bladder, the current air pressure value inside the bladder is recorded as the locking reference value. The preset pressure holding observation time is set to detect whether there is a leakage channel between the bladder and the cavity wall. After the pressure holding observation time ends, the air pressure value inside the bladder is read again, and the pressure drop of the air pressure value inside the bladder relative to the locking reference value is calculated. If the pressure drop is less than the preset allowable pressure drop threshold, the sealing is determined to be effective. If the pressure drop reaches or exceeds the allowable pressure drop threshold, the air pump control valve group is restarted to replenish the pressure. After the pressure replenishment is completed, the air pressure value inside the bladder is recorded again as the new locking reference value, and the pressure holding observation is repeated until the sealing is determined to be effective.

6. The method according to claim 1, characterized in that, The filling medium injection process includes: drilling filling holes and venting holes at the top of the current filling zone. The venting holes are located at the end of the current filling zone away from the filling holes, used to expel air from the filling zone during the filling process and to serve as an observation channel for slurry overflow when the filling is terminated; inserting the filling pipeline into the interior of the current filling zone through the filling holes, installing a pressure transmitter and an electromagnetic flowmeter sequentially at the orifice section of the filling pipeline, and installing a displacement sensor at the surface of the orifice of the filling holes; and installing a conductivity probe near the back of the recyclable flexible airbag through the airbag placement hole of the adjacent filling zone or a separately drilled monitoring hole to detect whether the foamed concrete slurry bypasses or penetrates the recyclable flexible airbag into the adjacent filling zone.

7. The method according to claim 6, characterized in that, The grouting pump is a variable frequency screw pump, which serves as the grouting power source to inject foamed concrete into the current grouting zone through the grouting pipeline. The slipform controller operates cycle by cycle with a fixed control period. In each control cycle, it reads the instantaneous injection flow rate value collected by the electromagnetic flowmeter and adds it to the cumulative injection volume accumulated in previous control cycles to obtain the current cumulative injection volume. At the same time, it reads the actual rotational speed value of the variable frequency screw pump and converts the actual rotational speed value into an estimated pump discharge volume value based on the pump's single-rotation displacement. The estimated pump discharge volume value is compared with the current cumulative injection volume. If the deviation between the two exceeds the preset integral verification deviation threshold, the current cumulative injection volume is corrected using the estimated pump discharge volume value. If the deviation is within the integral verification deviation threshold, the current cumulative injection volume is directly used to proceed to the next control calculation.

8. The method according to claim 7, characterized in that, In each control cycle, the sliding mode controller calculates the volume deviation between the corrected current cumulative injection volume and the target filling volume, and at the same time calculates the change in volume deviation in the current control cycle relative to the volume deviation in the previous control cycle to obtain the volume deviation change rate. The weighted combination of volume deviation and volume deviation change rate is used as the sliding mode variable. When the sliding mode variable is greater than zero, it indicates that the injection process is still in the stage of approaching the target filling volume. When the sliding mode variable is equal to zero or less than zero, it indicates that the current cumulative injection volume has reached the target filling volume.

9. The method according to claim 8, characterized in that, When the sliding mode variable is greater than zero, the sliding mode controller enters the approach phase: if the current injection pressure value collected by the pressure transmitter is lower than the pressure safety limit, the current road rebound displacement value collected by the displacement sensor is lower than the displacement safety limit, and the current conductivity value collected by the conductivity probe is lower than the cross-conduction conductivity threshold, then the sliding mode controller outputs a positive pump speed increment command according to the constant velocity approach method, driving the variable frequency screw pump to maintain or increase the injection flow rate; if the current injection pressure value reaches the pressure safety limit or the current road rebound displacement value reaches the displacement safety limit, then the pump speed increment command is switched to a negative value, waiting for the current injection pressure value and the current road rebound displacement value to both fall back to their respective safety limits. After the current conductivity value reaches or exceeds the cross-conductivity threshold, the sliding mode controller outputs a pump stop command to pause the filling process. After the reclaimable flexible airbag is pressurized, reinforced, or replaced, the sliding mode control will resume from the cumulative injection volume at the time of the pause. When the sliding mode variable is equal to or less than zero, the sliding mode controller outputs a pump stop command to shut down the variable frequency screw pump and confirms full filling by observing whether foamed concrete slurry overflows from the vent hole. After full filling, the air pump control valve group is opened to perform a depressurization operation on the reclaimable flexible airbag. After the airbag is fully contracted, the reclaimable flexible airbag is retrieved to the ground through the airbag placement hole, and the conductivity probe is removed.

10. The method according to claim 1, characterized in that, In step 2, the cross-correlation calculation is as follows: the normalized cross-correlation coefficient is calculated for the elastic wave reflection waveforms of the same measuring point before and after filling within a set time window. The higher the cross-correlation coefficient, the better the wave impedance matching between the filling body and the surrounding medium, and the higher the corresponding filling effect score. The section corresponding to the measuring point with a cross-correlation coefficient lower than the set qualified threshold is judged as an insufficient filling section.

Images