A geological disaster monitoring system and method based on a deep perception neural unit

Abstract

The application discloses a geological disaster monitoring system and method based on a deep perception neural unit, and belongs to the field of geotechnical engineering safety monitoring. The system comprises: a macroscopic perception unit composed of a force-bearing component coupled with a sensing optical fiber, which is implanted into the soil body by applying prestress; a microscopic perception unit composed of a sensing rod coupled with a sensing optical fiber, which is implanted into the soil body in a non-tensioning manner; a demodulation device connected with the sensing optical fiber; and a data fusion early warning unit for spatiotemporal fusion of a macroscopic stiffness field and a local modulus field to realize hierarchical early warning. The monitoring method comprises the following steps: laying a macro-microscopic sensing network, synchronously collecting data, inverting a macroscopic stiffness field, estimating a local modulus, performing time series leading analysis, spatial cross verification and triggering hierarchical early warning. The application realizes a paradigm shift from "form monitoring" to "process monitoring" through macro-microscopic collaborative monitoring, can prolong the early warning window, and meets the demand of significant projects for advanced early warning.

Description

Technical Field

[0001] This invention belongs to the field of geotechnical engineering safety monitoring technology, and in particular relates to a geological disaster monitoring system and method based on deep sensing neural units, which is applicable to the early warning and intelligent diagnosis of slip zone, void, and crack defects in geotechnical structures such as slopes, roadbeds, and foundation pits. Background Technology

[0002] Slope instability is one of the major geological hazards in the construction and operation of critical infrastructure, and its monitoring and early warning are challenging issues in the industry. Current mainstream technologies, such as GNSS and inclinometers, are essentially based on the "morphological recognition" paradigm, which means that they provide post-event warnings by measuring changes in macroscopic morphology such as displacement and tilt. However, the warning window is extremely short, making it difficult to meet the safety requirements of engineering projects.

[0003] In recent years, distributed fiber optic sensing technologies (such as BOTDA and DAS) have been introduced into slope monitoring, providing a new approach by attaching sensing fibers to the structure for continuous measurement. However, existing solutions have the following drawbacks:

[0004] (1) Single monitoring scale: Most solutions only use a single sensing unit (such as only using smart anchor cables), which cannot simultaneously realize the perception of macroscopic system stiffness and the capture of microscopic local deformation, resulting in insufficient diagnostic dimensions;

[0005] (2) Lagging early warning logic: Most schemes still focus on monitoring the absolute value of "strain" or "displacement", failing to upgrade to capturing the "loss process" of "stiffness", the essential physical quantity of the disaster, and the early warning is still lagging.

[0006] (3) Limited positioning accuracy: The spatial resolution of a single sensing unit is limited by the spacing between the units, making it difficult to accurately locate and qualitatively diagnose diseases.

[0007] To address the aforementioned issues, the applicant previously proposed "A Targeted Vibration Characteristic Monitoring Method for Cavities and Cracks in Engineering Structures" (application number 202511804611.6), which solved the problem of target shift in monitoring. However, the above method still mainly focuses on monitoring at a single scale and fails to achieve coordinated perception of macroscopic system stiffness and microscopic local deformation. Summary of the Invention

[0008] The first objective of this invention is to provide a geological disaster monitoring system based on depth-sensing neural units.

[0009] The second objective of this invention is to provide a geological disaster monitoring method based on the aforementioned geological disaster monitoring system.

[0010] This invention enables the coordinated perception of macroscopic system stiffness and microscopic local deformation, which can extend the early warning window and meet the needs of major projects for advanced early warning.

[0011] The first objective of this invention is achieved through the following technical solution:

[0012] A geological disaster monitoring system based on depth-sensing neural units, characterized in that it includes:

[0013] The macroscopic sensing unit includes at least one load-bearing component embedded in the soil, with at least one first sensing optical fiber coupled inside the load-bearing component. The load-bearing component is prestressed and is used to obtain distributed static strain data through Brillouin scattering time-domain analysis.

[0014] The micro-sensing unit includes at least one sensing rod implanted in the soil in a non-tensioned manner, and at least one second sensing fiber is coupled inside the sensing rod for obtaining distributed static strain data through Brillouin scattering light time-domain analysis.

[0015] The demodulation device is connected to the first sensing fiber of the macroscopic sensing unit and the second sensing fiber of the microscopic sensing unit after being fused together in series.

[0016] The data fusion and early warning unit receives and fuses the distributed static strain data from the macroscopic sensing unit and the microscopic sensing unit through the demodulation device. It inverts the macroscopic comprehensive stiffness field of the soil through the distributed static strain data of the macroscopic sensing unit, and inverts the local equivalent modulus through the distributed static strain data of the microscopic sensing unit. It issues an early warning when the corresponding threshold is triggered.

[0017] Furthermore, the load-bearing component is one or more combinations of prestressed steel strands, steel bars, structural steel, or fiber materials.

[0018] Furthermore, the sensing rod is a carbon fiber rod with a diameter of 3-10 mm, a tensile strength of not less than 1500 MPa, and an elastic modulus of 100-200 GPa.

[0019] Furthermore, the sensing rods of the micro-sensing unit and the load-bearing components of the macro-sensing unit are arranged in the soil at an intersecting angle.

[0020] Furthermore, the sensing rods of the micro-sensing units are arranged vertically or obliquely through the potential sliding belt, or arranged parallel between the load-bearing components of two adjacent macro-sensing units, or arranged in a grid pattern in key areas.

[0021] Furthermore, it also includes a demodulation and excitation unit, which includes a DAS demodulation module and at least one active excitation source fixed to the end of the macroscopic sensing unit and / or a supplementary excitation source independently buried in the soil, for generating active excitation signals and acquiring dynamic vibration data through the DAS demodulation module.

[0022] The second objective of this invention is achieved through the following technical solution:

[0023] A geological hazard monitoring method based on the above-mentioned geological hazard monitoring system is characterized by comprising the following steps:

[0024] S1. Macro-sensing units and micro-sensing units are deployed in the soil to form a macro-micro collaborative sensing network.

[0025] S2. Simultaneously acquire the distributed static strain data of the macroscopic sensing unit and the distributed static strain data of the microscopic sensing unit through the demodulation device;

[0026] S3. Based on the distributed static strain data of the macroscopic sensing unit, invert the macroscopic comprehensive stiffness field K_s(x,t) of the soil and its spatiotemporal evolution process, and calculate the stiffness loss rate ΔK_s(x,t);

[0027] Based on the distributed static strain data of the micro-sensing unit, the local equivalent modulus E_local(x,t) of the surrounding soil is inverted, the rate of change of the local equivalent modulus ΔE_local(x,t) is calculated, and micro-deformation anomaly features are extracted.

[0028] S4. Perform spatial superposition analysis on the stiffness loss area determined by the macroscopic sensing unit and the micro-deformation concentration area sensed by the microscopic sensing unit, and calculate the spatial overlap.

[0029] S5. Compare the stiffness loss rate, local equivalent modulus change rate, and spatial overlap with the set thresholds, and conduct graded early warning based on the comparison of the three.

[0030] Furthermore, after step S3, a warning timing precedence analysis is performed, specifically: obtaining the time point T_p when the microscopic sensing unit first captures the micro-deformation anomaly feature, obtaining the time point T_s when the macroscopic sensing unit first triggers the stiffness warning, and calculating the difference. When ΔT_gain is positive, it is confirmed that the micro-sensing unit has early warning timing leadership.

[0031] Furthermore, in step S3, for a single load-bearing member, the relationship of the macroscopic comprehensive stiffness field of the soil is:

[0032] ; Where σ(x,t) is the stress distribution estimated based on the anchorage load, ε(x,t) is the measured strain distribution, and η(x) is the correction coefficient for the reinforcement-soil coupling effect;

[0033] Then the stiffness loss rate ΔK_s(x,t) is:

[0034] ; Where K_s0(x) is the initial state stiffness distribution.

[0035] Furthermore, in step S3, the local equivalent modulus E_local(x,t) of the surrounding soil is inverted using an elastic foundation beam model based on Euler-Bernoulli beam theory. The governing equation is:

[0036] ; Where E is the elastic modulus of the sensing rod, I is the moment of inertia of the section, w is the deflection, k(x) is the foundation reaction coefficient, q(x) is the distributed load caused by soil deformation, and k(x) is proportional to the local equivalent modulus E_local(x).

[0037] Then the rate of change of the local equivalent modulus ΔE_local(x,t) is:

[0038] ; Where E_local0(x) is the local equivalent modulus of the initial state.

[0039] Further, in step S3, the identification criteria for the micro-deformation anomaly features include: the strain of the sensing rod exceeds the baseline value ±50με and continues to increase for more than 3 days, and / or the daily change rate of the local equivalent modulus E_local(x,t) exceeds 5%; the above anomalies are confirmed within two consecutive monitoring cycles.

[0040] Furthermore, in step S4, the spatial overlap γ is:

[0041] ; Where A_stiffness is the area of ​​the stiffness loss region, A_micro is the area of ​​the micro-deformation concentration region, and A_overlap is the overlapping area of ​​the two.

[0042] Further, in step S5, the threshold for the graded early warning is set as follows:

[0043] First warning: Stiffness loss rate ≥5%, or local equivalent modulus change rate ≥10%, with no spatial overlap;

[0044] Second warning: Stiffness loss rate ≥10%, local equivalent modulus change rate ≥15%, spatial overlap ≥30%;

[0045] Third warning: Stiffness loss rate ≥15%, local equivalent modulus change rate ≥20%, spatial overlap ≥50%;

[0046] Fourth warning: Stiffness loss rate ≥20%, and local equivalent modulus change rate ≥30%, spatial overlap ≥70%.

[0047] Furthermore, it also includes a long-term performance evaluation step: establishing a stiffness loss development model based on historical monitoring data to predict the time window for the soil to reach a critical state. Specifically, by collecting historical monitoring data, the stiffness loss rate is fitted with a curve over time using exponential smoothing or a grey model GM(1,1) to predict the time window for reaching the preset critical stiffness loss rate.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) Early warning capability: The micro-sensing unit is implanted in a non-tensioned manner and flexibly coupled with the soil, which can detect the micro-deformation of the soil in the first time. Simulation and theoretical analysis show that the micro-sensing unit can capture the early signs of disease 15-30 days earlier than the macro-sensing unit, which significantly extends the early warning window.

[0050] (2) Diagnostic accuracy: The macro-micro collaborative architecture combined with spatial cross-validation can effectively distinguish different types of defects such as slip zones, cavities, and cracks, and achieve precise spatial positioning. The cross-validation of independent macro-sensing units and micro-sensing units at the same location greatly reduces the risk of false alarms and false negatives.

[0051] (3) Economic practicality: The spacing between macroscopic sensing units can be widened to 5 meters or more, serving as a sparse "skeleton" to achieve large-scale coverage; microscopic sensing units are low-cost and easy to deploy, and can be flexibly densified in key areas. The combination of the two significantly reduces the cost of full-area monitoring while ensuring effectiveness.

[0052] (4) High reliability: The macro-sensing unit and the micro-sensing unit perceive independently, and the data cross-validation makes the early warning conclusion more reliable. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the geological disaster monitoring system according to an embodiment of the present invention;

[0054] Figure 2 This is a flowchart of the geological disaster monitoring method according to an embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram comparing the spatiotemporal evolution of the macroscopic comprehensive stiffness field and the local equivalent modulus in the simulation analysis of an embodiment of the present invention. The horizontal axis of the figure is time, and the vertical axis is the stiffness loss rate (%) and the local equivalent modulus change rate (%), respectively. It shows the synchronous downward trend of the two curves, and the modulus inflection point of the local equivalent modulus change curve appears before the stiffness inflection point of the stiffness loss curve.

[0056] Figure 4 This is a schematic diagram comparing the leading time of early warning in the simulation analysis of an embodiment of the present invention. The horizontal axis of the diagram is time (days) and the vertical axis is the intensity of the anomaly. It shows that the micro-sensing unit (sensing rod) captures the abnormal signal about 20 days ahead of the macro-sensing unit (load-bearing component).

[0057] Figure 5 This is a schematic diagram of spatial cross-verification in the simulation analysis of an embodiment of the present invention, showing the spatial overlap between the stiffness loss area determined by the macroscopic sensing unit and the micro-deformation concentration area sensed by the microscopic sensing unit, as well as the trend of spatial overlap with the increase of load.

[0058] Figure 6 This is a strain contour plot of the slope model in the simulation analysis of an embodiment of the present invention;

[0059] Figure 7 This is a curve showing the strain of the macroscopic sensing unit (load-bearing component) versus displacement in the simulation analysis of an embodiment of the present invention.

[0060] Figure 8 This is a curve showing the strain of the micro-sensing unit (sensing rod) as a function of displacement in the simulation analysis of an embodiment of the present invention.

[0061] Meaning of the labels in the attached diagram:

[0062] 1-Bearing component; 2-Sensing rod; 3-Soil; 4-Potential slip zone; 5-Slope. Detailed Implementation

[0063] The present invention will be further described below through specific embodiments, but this is not a limitation of the present invention. Those skilled in the art can make various modifications or improvements based on the basic idea of ​​the present invention, but as long as they do not depart from the basic idea of ​​the present invention, they are all within the protection scope of the present invention.

[0064] The geological disaster monitoring system based on deep sensing neural units in this embodiment includes a macroscopic sensing unit, a microscopic sensing unit, a demodulation device, a data fusion and early warning unit, and a demodulation and excitation unit.

[0065] The macroscopic sensing unit is used to monitor the macroscopic comprehensive stiffness field of the soil, providing distributed static strain monitoring at the macroscopic scale. The macroscopic sensing unit includes at least one load-bearing component 1. In this embodiment, the load-bearing component 1 consists of prestressed steel strands embedded in the soil 3 and at least one first sensing optical fiber coupled inside the prestressed steel strands. The load-bearing component 1 is installed using standard geotechnical anchoring processes such as drilling, tensioning, and pressure grouting, forming a strongly coupled composite system with the soil 3. After prestressing is applied, the load-bearing component 1 serves as a sensing carrier for Brillouin scattering time-domain analysis, providing a distributed static strain field with wide coverage reflecting the overall mechanical state of the "reinforcement-soil" system.

[0066] The micro-sensing unit is used for high-sensitivity deformation sensing at the microscale. The micro-sensing unit includes at least one sensing rod 2. In this embodiment, the sensing rod 2 consists of a carbon fiber rod arranged in the soil 3 in a non-tensioned manner and at least one second sensing optical fiber coupled inside the carbon fiber rod. The carbon fiber rod preferably has a diameter of 5 mm, an ultimate strength of not less than 2000 MPa, and an elastic modulus of approximately 160 GPa.

[0067] The aforementioned "non-tensioning method" means that no prestressing is required during installation; it can be fixed in the soil 3 simply by direct burial, light pressure implantation, or simple grouting. Its deployment method is flexible and can be densely deployed in key disease target areas.

[0068] As a highly sensitive "mechanical ruler", the micro-sensing unit directly senses the strain caused by the minute deformation of the surrounding soil. It has a high signal-to-noise ratio and the signal directly reflects the initiation and expansion process of the micro-deformation inside the soil.

[0069] The spatial arrangement of the macroscopic sensing unit and the microscopic sensing unit in soil 3 in this embodiment is as follows: Figure 1 As shown: The load-bearing component 1 of the macroscopic sensing unit is embedded into the soil 3 at a preset angle along the slope height. The sensing rod 2 of the microscopic sensing unit is inclined and laid through the potential slip zone 4. The sensing rod 2 of the microscopic sensing unit is embedded into the soil 3 at an angle (e.g., 45°) to the load-bearing component 1 of the macroscopic sensing unit. The two form a spatial intersection within the soil 3, covering the same monitoring area. Approximately 1 meter of optical fiber lead-out section is reserved at the exposed ends of both units outside the soil for optical fiber splicing protection and connection.

[0070] In this embodiment, the demodulation device uses a BOTDA demodulator. The soil-exposed end of the first sensing fiber of the macroscopic sensing unit and the soil-exposed end of the second sensing fiber of the microscopic sensing unit are fused together in series and then connected to the same BOTDA demodulator to achieve synchronous data acquisition.

[0071] The data fusion and early warning unit is the "intelligent brain" of this system. It is connected to the BOTDA demodulator and synchronously acquires the distributed static strain data of the macroscopic sensing unit and the distributed static strain data of the microscopic sensing unit through the BOTDA demodulator.

[0072] The data fusion and early warning unit incorporates a stiffness field inversion sub-model, a local modulus estimation sub-model, a time-series leadership analysis sub-model, and a spatial cross-validation sub-model. Specifically, the stiffness field inversion sub-model, based on distributed static strain data from the macroscopic sensing unit and combined with a mechanical model, inverts the macroscopic comprehensive stiffness field of the soil and its spatiotemporal evolution, calculating the stiffness loss rate. The local modulus estimation sub-model, based on distributed static strain data from the microscopic sensing unit, inverts the local equivalent modulus of the surrounding soil using an elastic foundation beam model, calculates the local equivalent modulus change rate, and extracts micro-deformation anomaly features. The time-series leadership analysis sub-model calculates the difference ΔT_gain = T_p - T_s between the time point T_p when the microscopic sensing unit first captures micro-deformation anomaly features and the time point T_s when the macroscopic sensing unit first triggers a stiffness warning, confirming the warning's time-series leadership. The spatial cross-validation sub-model performs spatial overlay analysis of the stiffness loss zone determined by the macroscopic sensing unit and the micro-deformation concentration zone perceived by the microscopic sensing unit.

[0073] The aforementioned data fusion and early warning unit achieves a paradigm shift from "morphological monitoring" to "process monitoring" through multi-model collaboration.

[0074] The demodulation and excitation unit of this embodiment includes a DAS demodulation module and at least one active excitation source fixed to the end of the macroscopic sensing unit. The demodulation and excitation unit is used to generate an active excitation signal and acquire dynamic vibration data through the DAS demodulation module.

[0075] The flowchart of the geological disaster monitoring method based on the above-mentioned geological disaster monitoring system in this embodiment is as follows: Figure 2 As shown, it includes the following steps:

[0076] S1. Macroscopic and microscopic sensing units are deployed in soil 3 to form a macro-micro coordinated sensing network, and an initial state baseline database is established.

[0077] Specifically, load-bearing components 1 of the macroscopic sensing unit are drilled and implanted into the slope soil at the designed locations, prestressed, and grouted for consolidation; sensing rods 2 of the microscopic sensing unit are implanted in key areas (such as the potential slip zone 4 path) in a non-tensioned manner. The first sensing fiber of the macroscopic sensing unit and the second sensing fiber of the microscopic sensing unit are fused together in series at their exposed ends in the soil and then connected to the BOTDA demodulator.

[0078] S2. Distributed static strain data from both the macroscopic and microscopic sensing units are simultaneously acquired using a BOTDA demodulator. The BOTDA demodulator performs Brillouin scattering time-domain analysis on the first and second sensing fibers to obtain distributed strain and temperature data along the entire length of the fibers.

[0079] S3. Invert the macroscopic comprehensive stiffness field K_s(x,t) of the soil and calculate the stiffness loss rate ΔK_s(x,t), and invert the local equivalent modulus E_local(x,t) and calculate the local equivalent modulus change rate ΔE_local(x,t), as detailed below:

[0080] (1) Based on the distributed static strain data of the macroscopic sensing unit, the macroscopic comprehensive stiffness field K_s(x,t) of the soil and its spatiotemporal evolution process are inverted by the mechanical model, and the stiffness loss rate ΔK_s(x,t) is calculated.

[0081] For a single load-bearing member, the relationship of its macroscopic comprehensive stiffness field K_s(x,t) of the soil is:

[0082] ; Wherein, σ(x,t) is the stress distribution estimated based on the anchorage load, ε(x,t) is the measured strain distribution, and η(x) is the correction coefficient for the reinforcement-soil coupling effect, which is determined through numerical calibration.

[0083] Then the stiffness loss rate ΔK_s(x,t) is:

[0084] ; Where K_s0(x) is the initial state stiffness distribution.

[0085] (2) Based on the distributed static strain data of the micro-sensing unit, the local equivalent modulus E_local(x,t) of the surrounding soil is inverted by the elastic foundation beam model based on Euler-Bernoulli beam theory, the local equivalent modulus change rate ΔE_local(x,t) is calculated, and the micro-deformation anomaly features are extracted.

[0086] Carbon fiber rods can be considered as elastic foundation beams in soil. When the surrounding soil undergoes slight deformation, the carbon fiber rods generate flexural strain. Based on the Euler-Bernoulli beam theory, the governing equations are established as follows:

[0087] ; Where E is the elastic modulus of the carbon fiber rod, I is the moment of inertia of the section, w is the deflection, k(x) is the soil reaction coefficient, k(x) is proportional to the local equivalent modulus E_local(x) (k = β·E_local), ∂ is the geometric coefficient related to the diameter and burial depth of the carbon fiber rod, and q(x) is the distributed load caused by soil deformation.

[0088] During the calculation, w(x) is inverted by the measured strain distribution, and then k(x) and E_local(x) are solved.

[0089] Then the rate of change of the local equivalent modulus ΔE_local(x,t) is:

[0090] ; Where E_local0(x) is the local equivalent modulus of the initial state.

[0091] S4, Spatial Cross-Validation:

[0092] The stiffness loss zone determined by the macroscopic sensing unit and the micro-deformation concentration zone sensed by the microscopic sensing unit are spatially superimposed and analyzed to calculate the spatial overlap.

[0093] The spatial overlap γ is defined as:

[0094] ; Where A_stiffness is the area of ​​the stiffness loss region, A_micro is the area of ​​the micro-deformation concentration region, and A_overlap is the overlapping area of ​​the two.

[0095] When γ exceeds the set threshold, the warning level and credibility are increased.

[0096] S5, Tiered Early Warning:

[0097] When the stiffness loss rate ΔK_s(x,t) and the local equivalent modulus change rate ΔE_local(x,t) exceed the preset threshold, and the spatial overlap γ meets the requirements, a graded early warning is triggered.

[0098] The specific threshold settings are as follows:

[0099] Blue alert (first alert): Stiffness loss rate ≥5%, or local equivalent modulus change rate ≥10%, no spatial overlap (i.e., spatial overlap = 0).

[0100] Yellow alert (second alert): Stiffness loss rate ≥10%, local equivalent modulus change rate ≥15%, spatial overlap ≥30%;

[0101] Orange alert (third alert): Stiffness loss rate ≥15%, local equivalent modulus change rate ≥20%, spatial overlap ≥50%;

[0102] Red alert (fourth alert): Stiffness loss rate ≥20%, local equivalent modulus change rate ≥30%, spatial overlap ≥70%.

[0103] The geological disaster monitoring method in this embodiment further performs an early warning timing precedence analysis after step S3, specifically as follows:

[0104] The following time points were obtained:

[0105] T_p: The time point at which the micro-sensing unit first captures the micro-deformation anomaly feature;

[0106] T_s: The time point at which the macroscopic sensing unit first triggers the stiffness warning.

[0107] The criteria for identifying micro-deformation anomaly features are as follows:

[0108] (1) The strain of the carbon fiber rod exceeds the baseline value by ±50με and continues to increase for more than 3 days;

[0109] (2) The daily rate of change of the local equivalent modulus E_local(x,t) exceeds 5%;

[0110] The above-mentioned anomaly was confirmed in two consecutive monitoring cycles.

[0111] Calculate the time series leadership quantification metric ΔT_gain:

[0112] .

[0113] When ΔT_gain is positive, it is confirmed that the micro-sensing unit has a leading warning time sequence, and the number of leading days is ΔT_gain.

[0114] The physical essence of the timing-leading characteristic lies in the fact that the sensing rod 2 of the micro-sensing unit is implanted in a non-tensioned manner and is flexibly coupled with the soil 3, unaffected by the prestressing system. Micro-deformation of the soil 3 is directly transmitted to the sensing rod 2, generating a strain response immediately. In contrast, the load-bearing component 1 of the macro-sensing unit is prestressed, and the strain increment generated by the micro-deformation of the soil 3 is "submerged" by the prestress. Only when the stiffness loss accumulates to a certain extent can it be identified from changes in strain distribution. Therefore, the timing-leading characteristic of the micro-sensing unit is structural and inevitable.

[0115] To verify the technical effectiveness of this invention, a two-dimensional finite element model will be established using ABAQUS / Standard for simulation analysis:

[0116] Model parameters: A slope model with a height of 20m and a width of 40m and a slope angle of 45° was established. The slope consisted of topsoil, slip zone, and bedrock. A weak interlayer (potential slip zone) with a thickness of 0.5m and a length of 5m was set at a slope height of 10m. Three load-bearing members with macroscopic sensing units (15m in length and 5m apart) were arranged along the slope height, and five sensing rods with microscopic sensing units (3m in length and 1m apart) were arranged in the potential slip zone area. The Mohr-Coulomb model was used for the soil and rock mass; the strain softening model was used for the weak interlayer to simulate the asymptotic stiffness loss process. The load was gradually increased in 10 load steps, and the strain distribution and stiffness field evolution were recorded at each load step.

[0117] like Figures 3 to 8 As shown, the analysis results are as follows:

[0118] (1) Results of time series leadership verification:

[0119] Load step 1-2 (load factor 0.3-0.5): No abnormalities were detected in either the macroscopic sensing unit or the microscopic sensing unit;

[0120] Load step 3 (load factor 0.7): The micro-sensing unit detects micro-deformation anomaly (T_p) for the first time, while the macro-sensing unit does not issue a warning;

[0121] Load step 4 (load factor 0.8): The micro-sensing unit continues to malfunction, while the macro-sensing unit does not issue any warnings;

[0122] Load step 5 (load factor 0.9): The micro-sensing unit shows significant anomalies, and the macro-sensing unit triggers stiffness warning for the first time (T_s).

[0123] The timing lead gain ΔT_gain = 2 load steps, corresponding to an actual time of approximately 15-30 days.

[0124] (2) Results of correlation analysis between stiffness loss and microdeformation:

[0125] The stiffness loss rate of the sliding belt gradually increased from 3.2% in load step 3 to 32.1% in load step 10 (critical state); the peak strain of the carbon rod insert increased from 68 με to 538 με. The two showed a high positive correlation (R²=0.97), verifying the intrinsic relationship between microscopic deformation and macroscopic stiffness loss. The spatial overlap gradually increased from 25% in load step 3 to 91% in load step 10, indicating a high degree of spatial consistency between macroscopic and microscopic anomalies.

[0126] (3) Quantitative results of early warning window:

[0127] A blue alert corresponds to an actual warning window of approximately 60-90 days, a yellow alert approximately 30-60 days, an orange alert approximately 15-30 days, and a red alert approximately 3-15 days. This invention achieves full-cycle warning coverage from blue to red alerts through macro-micro coordinated monitoring, significantly extending the warning window compared to traditional displacement monitoring (which typically only provides warnings a few hours to a few days in advance).

[0128] The following are specific examples of applications of the present invention:

[0129] Example 1: This example is applied to on-site monitoring of newly constructed slopes. During the construction of a new embankment slope for a highway, sensing rods of micro-sensing units were installed in layers with a spacing of 2 meters. After the slope was formed, load-bearing components of macro-sensing units were drilled, installed, and tensioned with a spacing of 5 meters. Both were connected to a monitoring station at the slope toe via fiber optic cables and then to a BOTDA demodulator. After one year of system operation, the micro-sensing units were the first to detect a gradual decrease in the local equivalent modulus in a certain area, issuing a micro-risk warning 45 days earlier than traditional visual inspections. Later, the macro-sensing units' data on the overall macro-stiffness field of the soil also confirmed the slow loss of stiffness in this area.

[0130] Example 2: This example is applied to post-reinforcement monitoring of an existing landslide. In a landslide control project where creep had already occurred, anti-slide piles were laid along the landslide body, and load-bearing components of macroscopic sensing units were implanted into holes drilled at 5-meter intervals between the piles for reinforcement. Subsequently, in the identified main slip zone and its upper and lower edges, vertical holes were drilled and sensing rods of microscopic sensing units were implanted, allowing them to pass directly through the slip zone. After the system was running, the macroscopic sensing unit detected a continuous loss of stiffness in the slip zone area through stiffness field inversion; the microscopic sensing unit accurately captured the concentration of micro-deformations at the slip zone location, with a spatial overlap of over 85%, achieving precise delineation of the slip zone location and its range of motion.

[0131] This invention and the applicant's previous application for a method for targeted vibration monitoring of voids and cracks in engineering structures (application number 202511804611.6) belong to different levels of the same technical system. The previous method for targeted vibration monitoring of voids and cracks in engineering structures solved the problem of targeted vibration monitoring of a single armored reinforcing bar. Building upon this, this invention further introduces a microscopic sensing unit, realizing macro-micro coordinated stiffness-modulus joint diagnosis, elevating the monitoring paradigm from "single-scale vibration diagnosis" to "multi-scale process diagnosis," and extending the early warning window from several days to several weeks or even months. The two technical solutions are complementary and do not overlap, together forming a complete intelligent geological disaster monitoring technology chain.

[0132] The above embodiments of the present invention are not intended to limit the scope of protection of the present invention. The implementation of the present invention is not limited thereto. All other modifications, substitutions or alterations made to the above structure of the present invention based on the above content of the present invention, in accordance with ordinary technical knowledge and common practice in the field, without departing from the basic technical idea of ​​the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A geological disaster monitoring system based on a deep perception neural unit, characterized by ,include: The macroscopic sensing unit includes at least one load-bearing component embedded in the soil, with at least one first sensing optical fiber coupled inside the load-bearing component. The load-bearing component is prestressed and is used to obtain distributed static strain data through Brillouin scattering time-domain analysis. The micro-sensing unit includes at least one sensing rod implanted in the soil in a non-tensioned manner, and at least one second sensing fiber is coupled inside the sensing rod for obtaining distributed static strain data through Brillouin scattering light time-domain analysis. The demodulation device is connected to the first sensing fiber of the macroscopic sensing unit and the second sensing fiber of the microscopic sensing unit after being fused together in series. The data fusion and early warning unit receives and fuses the distributed static strain data from the macroscopic sensing unit and the microscopic sensing unit through the demodulation device. It inverts the macroscopic comprehensive stiffness field of the soil through the distributed static strain data of the macroscopic sensing unit, and inverts the local equivalent modulus through the distributed static strain data of the microscopic sensing unit. It issues an early warning when the corresponding threshold is triggered. 2.The geological disaster monitoring system based on deep perception neural unit according to claim 1, wherein, The load-bearing component is one or more combinations of prestressed steel strands, steel bars, structural steel, or fiber materials.

3. The geological disaster monitoring system based on depth-sensing neural units according to claim 1, characterized in that, The sensing rod is a carbon fiber rod with a diameter of 3-10 mm, a tensile strength of not less than 1500 MPa, and an elastic modulus of 100-200 GPa.

4. The geological disaster monitoring system based on depth-sensing neural units according to claim 1, characterized in that, The sensing rods of the micro-sensing unit and the load-bearing components of the macro-sensing unit are arranged at a non-parallel angle to form a spatial intersection.

5. The geological disaster monitoring system based on depth-sensing neural units according to claim 1, characterized in that, The sensing rods of the micro-sensing units are arranged vertically or obliquely through the potential sliding belt, or parallel between the load-bearing components of two adjacent macro-sensing units, or arranged in a grid pattern in key areas.

6. The geological disaster monitoring system based on depth-sensing neural units according to claim 1, characterized in that, It also includes a demodulation and excitation unit, which includes a DAS demodulation module and at least one active excitation source fixed to the end of the macroscopic sensing unit and / or a supplementary excitation source independently buried in the soil, for generating active excitation signals and acquiring dynamic vibration data through the DAS demodulation module.

7. A geological hazard monitoring method based on the geological hazard monitoring system according to any one of claims 1-6, characterized in that, Includes the following steps: S1. Macro-sensing units and micro-sensing units are deployed in the soil to form a macro-micro collaborative sensing network. S2. Simultaneously acquire the distributed static strain data of the macroscopic sensing unit and the distributed static strain data of the microscopic sensing unit through the demodulation device; S3. Based on the distributed static strain data of the macroscopic sensing unit, invert the macroscopic comprehensive stiffness field K_s(x,t) of the soil and its spatiotemporal evolution process, and calculate the stiffness loss rate ΔK_s(x,t); Based on the distributed static strain data of the micro-sensing unit, the local equivalent modulus E_local(x,t) of the surrounding soil is inverted, the rate of change of the local equivalent modulus ΔE_local(x,t) is calculated, and micro-deformation anomaly features are extracted. S4. Perform spatial superposition analysis on the stiffness loss area determined by the macroscopic sensing unit and the micro-deformation concentration area sensed by the microscopic sensing unit, and calculate the spatial overlap. S5. Compare the stiffness loss rate, local equivalent modulus change rate, and spatial overlap with the set thresholds, and conduct graded early warning based on the comparison of the three.

8. The geological disaster monitoring method according to claim 7, characterized in that, Following step S3, a warning timing precedence analysis is performed, specifically: obtaining the time point T_p when the microscopic sensing unit first captures the micro-deformation anomaly feature, obtaining the time point T_s when the macroscopic sensing unit first triggers the stiffness warning, and calculating the difference. When ΔT_gain is positive, it is confirmed that the micro-sensing unit has early warning timing leadership.

9. The geological disaster monitoring method according to claim 7, characterized in that, In step S3, for a single load-bearing member, the relationship of the macroscopic comprehensive stiffness field of the soil is: ； Where σ(x,t) is the stress distribution estimated based on the anchorage load, ε(x,t) is the measured strain distribution, and η(x) is the correction coefficient for the reinforcement-soil coupling effect; Then the stiffness loss rate ΔK_s(x,t) is: ； Where K_s0(x) is the initial state stiffness distribution.

10. The geological disaster monitoring method according to claim 7, characterized in that, In step S3, the local equivalent modulus E_local(x,t) of the surrounding soil is inverted using an elastic foundation beam model based on Euler-Bernoulli beam theory. The governing equation is: ； Where E is the elastic modulus of the sensing rod, I is the moment of inertia of the section, w is the deflection, k(x) is the foundation reaction coefficient, q(x) is the distributed load caused by soil deformation, and k(x) is proportional to the local equivalent modulus E_local(x). Then the rate of change of the local equivalent modulus ΔE_local(x,t) is: ； Where E_local0(x) is the local equivalent modulus of the initial state.

11. The geological disaster monitoring method according to claim 7, characterized in that, In step S3, the identification criteria for the micro-deformation anomaly features include: the strain of the sensing rod exceeds the baseline value ±50με and continues to increase for more than 3 days, and / or the daily change rate of the local equivalent modulus E_local(x,t) exceeds 5%; the above anomalies are confirmed within two consecutive monitoring cycles.

12. The geological disaster monitoring method according to claim 7, characterized in that, In step S4, the spatial overlap γ is: ； Where A_stiffness is the area of ​​the stiffness loss region, A_micro is the area of ​​the micro-deformation concentration region, and A_overlap is the overlapping area of ​​the two.

13. The geological disaster monitoring method according to claim 7, characterized in that, In step S5, the threshold for the tiered early warning is set as follows: First warning: Stiffness loss rate ≥5%, or local equivalent modulus change rate ≥10%, with no spatial overlap; Second warning: Stiffness loss rate ≥10%, local equivalent modulus change rate ≥15%, spatial overlap ≥30%; Third warning: Stiffness loss rate ≥15%, local equivalent modulus change rate ≥20%, spatial overlap ≥50%; Fourth warning: Stiffness loss rate ≥20%, and local equivalent modulus change rate ≥30%, spatial overlap ≥70%.

14. The geological disaster monitoring method according to claim 7, characterized in that, It also includes a long-term performance evaluation step: establishing a stiffness loss development model based on historical monitoring data to predict the time window when the soil reaches the critical state.

Images