Disaster early warning system for tunnel crossing active fault zone

Abstract

The invention discloses a catastrophe early warning system for a tunnel passing through an active fault zone, and relates to the technical field of tunnel early warning, the system comprises a multi-mode sensing layer used for obtaining micro-seismic wave, vibration, inclination and strain data along the tunnel fault zone, the multi-mode sensing layer comprises an optical fiber distributed sound wave sensing unit DAS, a data processing unit DAS and a data processing unit DAS, the optical fibers are arranged along the inner side of the tunnel lining at intervals of 10m and are used for monitoring tiny sound wave signals of a fracture surface in real time; the MEMS accelerometer arrays are arranged at four corners of each section and are used for detecting high-frequency vibration; gNSS and INS combined inclination strain sensors which are installed on the top plate and the side wall of the tunnel and are used for obtaining millimeter-level inclination and strain changes; and the acoustic cavity sensor is arranged in a near area of the sliding surface of the fault zone and is used for capturing an air sound wave precursor signal. According to the invention, through multi-mode fusion sensing, the function of capturing fracture micro-shock, vibration and deformation in all directions is realized, the problem of missed alarm and false alarm of single sensing is solved, the sensitivity order of magnitude is improved, and the false alarm rate is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel early warning, and in particular to a disaster early warning system for tunnels crossing active fault zones. Background Art

[0002] Existing tunnel fault zone disaster warning systems mostly rely on single sensing or centralized processing, and have the following major technical problems: the data source is single and can only capture single signals in vibration or surface deformation, which is prone to missed reports and false alarms; communication and storage are centralized, remote processing has high latency and is difficult to prevent tampering, resulting in poor security; the prediction model is simple, mostly based on static thresholds or linear regression, and cannot identify complex nonlinear fracture precursors; emergency linkage is delayed, and there is a lack of automated multi-level decision-making and rapid response mechanisms.

[0003] Patent CN118167430B discloses a dynamic early warning system and method for TBM tunnel rockbursts based on multi-source information fusion. The above patent realizes advanced classification, dynamic correction and real-time early warning of TBM tunnel rockburst disasters, thereby avoiding rockburst disasters and improving the efficiency and safety of TBM excavation.

[0004] The above patent conducts advanced classification, dynamic correction and real-time warning of tunnel rock burst disasters to avoid rock burst disasters and improve the efficiency and safety of TBM excavation. However, the collection of disaster data in the tunnel relies on a single sensor, which has the problem of single information.

[0005] To this end, this application proposes a tunnel crossing active fault zone disaster warning system that can fully capture fault microseismicity, vibration and deformation functions. Summary of the Invention

[0006] The purpose of the present invention is to provide a disaster warning system for tunnels crossing active fault zones to solve the technical problems raised in the above-mentioned background technology of relying too much on single sensing or centralized processing.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a disaster early warning system for tunnels crossing active fault zones, the system comprising a multimodal sensing layer for acquiring microseismic wave, vibration, tilt, and strain data along the tunnel fault zone, the multimodal sensing layer comprising:

[0008] Fiber-optic distributed acoustic sensing units (DAS) are installed every 10 meters along the inner side of the tunnel lining to monitor tiny acoustic signals from the fracture surface in real time.

[0009] MEMS accelerometer arrays are placed at the four corners of each section to detect high-frequency vibrations;

[0010] GNSS and INS combined tilt and strain sensors are installed on the tunnel roof and side walls to obtain millimeter-level tilt and strain changes;

[0011] The acoustic cavity sensor is placed in the area near the sliding surface of the fault zone to capture the air acoustic wave precursor signal.

[0012] Preferably, the multimodal sensing layer is directly connected to an edge intelligent gateway via industrial Ethernet. The edge intelligent gateway includes a combined architecture of FPGA and ARM Cortex-A series processors, and is used to execute:

[0013] Noise filtering processing uses an adaptive wavelet threshold algorithm to remove environmental and construction noise;

[0014] Online extraction of multiple features in time domain, frequency domain, time-frequency domain and entropy value;

[0015] The online transfer learning module adaptively updates the feature extraction and classification models based on the initial fracture simulation data.

[0016] Preferably, the system also includes a distributed blockchain storage module, which is built based on Hyperledger Fabric. The edge intelligent gateway is the blockchain network node. The distributed blockchain storage module is used to perform lightweight on-chain storage of feature vectors and preliminary warning results to ensure that the data cannot be tampered with and is traceable throughout the process.

[0017] Preferably, the system further includes a digital twin simulation module, which is deployed on the central control server and includes:

[0018] A three-dimensional tunnel-stratum-fault zone coupling model was established based on the finite element method;

[0019] The stress-displacement coupling simulation engine supports GPU acceleration and has a simulation latency of ≤1s, which is used to meet the needs of fast updates at the hourly level. The stress-displacement coupling simulation engine receives real-time feature input from the edge intelligent gateway to simulate the fracture evolution process.

[0020] Preferably, the system also includes a prediction and decision-making module, which integrates a graph neural network GNN and a long-term and short-term network and a LSTM combined architecture to fuse simulation data with historical monitoring sequences to generate multi-scale disaster warnings of levels I to IV.

[0021] Preferably, the system further includes an emergency linkage module and a remote diagnosis and maintenance module. The emergency linkage module is used to push the warning results to the mobile terminal supervision APP and the central control room visualization screen via 5G wireless and industrial Ethernet respectively, and automatically trigger the tunnel ventilation, lighting and escape indication system;

[0022] The emergency linkage module uses RESTful and MQTT dual-protocol communication, supports dual platforms on the mobile terminal, and generates automatic reports in standard PDF format, including warning levels, simulation snapshots, and emergency response recommendations;

[0023] The remote diagnosis and maintenance module triggers the gateway health status to be uploaded to the chain through the blockchain smart contract, and combines with the off-chain data analysis platform to provide active maintenance warnings for the gateway and sensor status.

[0024] Preferably, the fiber-optic distributed acoustic wave sensing unit adopts φ9 / 125μm single-mode optical fiber, adopts phase-sensitive OTDR technology, has a spatial resolution of ≤0.5m, and a sampling frequency of ≥10kHz;

[0025] The MEMS accelerometer array uses a three-axis ±4g range sensor, each with a bandwidth of 0.5Hz-2kHz and dual software and hardware temperature compensation.

[0026] The GNSS and INS combined tilt strain sensor achieves tunnel section micro-deformation detection accuracy of ≤0.2mm through a four-point differential layout.

[0027] Preferably, the online transfer learning module deployed in the edge intelligent gateway is pre-trained based on large-scale source domain fracture simulation data, and uses self-supervised comparative learning to fine-tune the newly collected data; when the model performance decreases by more than 5%, remote update is automatically triggered.

[0028] Preferably, the blockchain storage module supports dynamic node management, the document block size is ≤10KB, the chain delay is ≤100ms, and the hash storage of large-scale feature vectors is realized through the off-chain Merkle tree structure.

[0029] Preferably, the graph neural network submodule in the prediction decision module is used to construct the spatial topological relationship between sensor points, and the long short-term memory network submodule is used to capture temporal features; the two are fused through the attention mechanism, and the single prediction accuracy of the overall model is ≥92%.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. This invention uses multimodal fusion sensing to achieve all-round capture of fault microseismicity, vibration, and deformation, solving the problem of missed or false alarms caused by single sensing, increasing sensitivity by orders of magnitude and reducing the false alarm rate.

[0032] 2. This invention uses edge intelligent gateways and blockchain to achieve local real-time classification decisions and data traceability and tamper-proofing, solving the problems of high centralized latency and security risks, and improving data real-time performance, data integrity, and trustworthiness.

[0033] 3. This invention uses stress-displacement simulation and multi-scale models to achieve multi-scale risk prediction, solving the problem that linear models are difficult to capture complex nonlinear precursors, extending the early warning time window and improving prediction accuracy;

[0034] 4. The present invention realizes multi-level early warning automatic triggering of emergency control and proactive maintenance work order generation through automatic linkage and maintenance closed loop, solves the problems of slow manual response and passive maintenance, reduces response delay and maintenance work order processing time, and improves system availability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a schematic diagram of the disaster warning logic flow of the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0037] Example 1

[0038] See also Figure 1 A disaster early warning system for tunnels crossing active fault zones, comprising a multimodal sensing layer for acquiring microseismic wave, vibration, tilt, and strain data along the tunnel fault zone, the multimodal sensing layer comprising:

[0039] Fiber-optic distributed acoustic sensing units (DAS) are installed every 10 meters along the inner side of the tunnel lining to monitor tiny acoustic signals from the fracture surface in real time.

[0040] MEMS accelerometer arrays are placed at the four corners of each section to detect high-frequency vibrations;

[0041] GNSS and INS combined tilt and strain sensors are installed on the tunnel roof and side walls to obtain millimeter-level tilt and strain changes;

[0042] Acoustic cavity sensors are placed near the sliding surface of the fault zone to capture airborne acoustic precursor signals;

[0043] The fiber-optic distributed acoustic wave sensing unit adopts φ9 / 125μm single-mode optical fiber and phase-sensitive OTDR technology, with a spatial resolution of ≤0.5m and a sampling frequency of ≥10kHz;

[0044] The MEMS accelerometer array uses a three-axis ±4g range sensor, each with a bandwidth of 0.5Hz-2kHz and dual software and hardware temperature compensation.

[0045] The GNSS and INS combined tilt strain sensor achieves tunnel section micro-deformation detection accuracy of ≤0.2mm through a four-point differential layout;

[0046] Furthermore, the early microseismic wave fusion monitoring of the fault zone:

[0047] In the fault zone section of a mountain highway tunnel, a φ9 / 125μm single-mode optical fiber was laid every 10 meters along the inner side of the lining and fixed to the lining sidewalls and roof. A triaxial ±4g MEMS accelerometer was installed at each corner of the section, along with a temperature sensor for dual temperature compensation. A combined GNSS and INS tilt strain sensor was deployed on each of the tunnel roof and sidewalls, using a four-point differential arrangement. After calibration, a micro-deformation resolution of ≤0.2mm was achieved. Acoustic cavity sensors with a frequency response range of 20Hz–20kHz were suspended in areas with potential slip on the fault surface.

[0048] An edge intelligent gateway is installed every 100 meters next to the lining and directly connected to the aforementioned sensor units via industrial Ethernet. The DAS unit uses phase-sensitive OTDR technology, with a sampling frequency of 10kHz and a spatial resolution of 0.5m, to collect echo signals in real time and transmit them to the FPGA. The FPGA runs an adaptive wavelet threshold algorithm to filter out man-made construction noise above 80Hz and extract features such as the time domain envelope, spectrum peaks, time-frequency diagrams, and time series entropy. MEMS accelerometer array data is transmitted to the gateway via the SPI bus, which also extracts multiple features such as entropy and frequency band energy ratio.

[0049] The edge gateway is pre-installed with a comparative learning model pre-trained based on the fracture simulation library to perform real-time classification of newly collected features. When the microseismic event feature matching degree is ≥0.85 and the acceleration peak is >0.2g within 5 consecutive seconds, it is judged as a Level I warning (attention), and the feature vector and warning mark are uploaded to the chain through the Fabric lightweight block. After the central control server obtains the data uploaded by the edge node, it drives the three-dimensional digital twin model to run stress-displacement coupling simulation to simulate the possible opening amount of the fracture surface in the next hour. The simulation results and historical data are input into the GNN+LSTM prediction module. If it is predicted that there will be a risk of Level II or above in the next 1 hour, it will be pushed to the central control screen and pushed to the supervision APP via 5G.

[0050] Example 2

[0051] See also Figure 1 A disaster warning system for tunnels crossing active fault zones, which also includes a distributed blockchain storage module built on Hyperledger Fabric. The edge intelligent gateway is a blockchain network node. The distributed blockchain storage module is used to perform lightweight on-chain storage of feature vectors and preliminary warning results to ensure that the data cannot be tampered with and is fully traceable.

[0052] Furthermore, in the implementation of multi-scale coordinated emergency response:

[0053] Based on Example 1, each edge smart gateway is registered as a Fabric network node and a dynamic node management contract is configured; the central control system is connected to the local earthquake bureau emergency platform through a RESTful interface, and emergency linkage rules are pre-configured;

[0054] The GNN submodule constructs a fracture point cloud map based on the spatial topology of the fracture zone optical fiber and the acceleration array. The LSTM submodule inputs the entropy value sequence of the past 24 hours and outputs the sliding trend of the next 24 hours. The attention mechanism is used to fuse the outputs of the two to generate the risk probability distribution of levels I-IV, which is defined as:

[0055] Level Ⅰ (safe): risk <10%;

[0056] Level II (Caution): 10% ≤ risk < 30%;

[0057] Level III (Warning): 30% ≤ risk < 60%;

[0058] Grade IV (urgent): risk ≥ 60%;

[0059] When the warning level reaches Level III, the system automatically triggers the lighting and ventilation in the tunnel to switch to emergency mode through MQTT, and simultaneously generates a PDF report containing simulation snapshots; when the warning level reaches Level IV, the escape indication system is further activated, and SMS and email alerts are sent to the earthquake bureau platform; the blockchain smart contract monitors the gateway health data on the chain. If the node is offline for more than 30 minutes, it automatically alarms for maintenance and generates a diagnostic work order on the off-chain platform.

[0060] Example 3

[0061] See also Figure 1 : A disaster warning system for tunnel crossing an active fault zone, wherein the multimodal sensing layer is directly connected to an edge intelligent gateway via industrial Ethernet. The edge intelligent gateway includes a combined architecture of FPGA and ARM Cortex-A series processors, and is used to execute:

[0062] Noise filtering processing uses an adaptive wavelet threshold algorithm to remove environmental and construction noise;

[0063] Online extraction of multiple features in time domain, frequency domain, time-frequency domain and entropy value;

[0064] Online transfer learning module, which adaptively updates the feature extraction and classification model based on the initial fracture simulation data;

[0065] The online transfer learning module deployed in the edge intelligent gateway is pre-trained based on large-scale source domain fracture simulation data and fine-tuned on newly collected data using self-supervised contrastive learning. When the model performance degrades by more than 5%, remote updates are automatically triggered.

[0066] Furthermore, each sensor data channel is encapsulated into a UDP packet via industrial Ethernet and sent to the FPGA DMA buffer. The FPGA first performs phase demodulation and OTDR differential calculation on the DAS data, outputting a time-domain amplitude sequence. For accelerometer and acoustic cavity data, the FPGA samples and directly forwards it to the ARM side, where it is packaged and stored in parallel in a ring buffer queue.

[0067] The FPGA side applies a 4th-order FIR bandpass filter to the DAS and acceleration data to remove out-of-band signals. The ARM side runs an adaptive wavelet threshold algorithm module to further remove noise from each channel.

[0068] Divide the data into 1s chunks:

[0069] 1) Wavelet decomposition to level 5;

[0070] 2) Adaptively calculate the threshold λ=σ√(2lnN) based on the background noise estimate;

[0071] 3) Soft threshold shrinkage of high-frequency coefficients;

[0072] 4) Reconstruct the denoised signal.

[0073] After multiple features are extracted online, the ARM side outputs the feature vectors to the circular feature queue every 0.5 seconds for the prediction model to call;

[0074] Using large-scale fracture simulation experimental data (N≈10 6 samples) pre-trained contrastive learning encoder, and a lightweight fully connected network for binary classification based on feature vectors;

[0075] Self-supervised fine-tuning:

[0076] 1. Randomly extract 512 new samples from the local feature queue and construct positive and negative pairs together with the pre-training samples;

[0077] 2. Calculate contrast loss (NT-Xent);

[0078] 3. Perform 5 gradient updates on ARM;

[0079] The edge node evaluates the accuracy of the local model on the validation set once an hour. If the accuracy drops by more than 5% compared to the previous evaluation, it communicates with the cloud-based model management service via gRPC, automatically pulls the latest pre-trained weights, and seamlessly updates them.

[0080] Example 4

[0081] See also Figure 1 A disaster warning system for tunnels crossing active fault zones, the system also includes a digital twin simulation module deployed on a central control server, including:

[0082] A three-dimensional tunnel-stratum-fault zone coupling model was established based on the finite element method;

[0083] The stress-displacement coupling simulation engine supports GPU acceleration and has a simulation latency of ≤1s, meeting the need for hourly rapid updates. The stress-displacement coupling simulation engine receives real-time feature input from the edge intelligent gateway to simulate the fracture evolution process.

[0084] The system also includes a prediction and decision-making module that integrates a graph neural network (GNN) with a long-term and short-term network and a network LSTM combined architecture to fuse simulation data with historical monitoring sequences to generate multi-scale disaster warnings of levels I to IV;

[0085] Furthermore, the central control end digital twin simulation and multi-scale early warning decision-making process:

[0086] 3D tunnel-stratum-fault zone coupled model construction: Based on the tunnel design blueprint and geological survey report, a 3D model was established in Ansys Mechanical:

[0087] Geometry: Tunnel lining, circumferential perforation, and fault zone discontinuities are constructed according to measured strike and dip;

[0088] Meshing: The fault zone area uses a local refined tetrahedron mesh with a minimum cell size of 0.1m, the lining and surrounding bottom layer mesh size is 0.5m, and the total number of cells is about 5×10 6 ;

[0089] Material properties: Lining - high-strength fiber reinforced concrete (elastic modulus 40 GPa, Poisson's ratio 0.2), surrounding rock - layered sandstone-shale (E=10 GPa-15 GPa, v=0.25), fault zone filling - clay-fine sand (E=0.5 GPa, v=0.3);

[0090] Contact definition: The fracture surface adopts surface-to-surface contact element, the friction coefficient μ=0.6, and opening and closing are allowed;

[0091] Stress-displacement coupled simulation engine:

[0092] Solution strategy: implicit time-domain integral Newmark-β is used, β=0.25, γ=0.5;

[0093] Parallel computing: The model domain is divided into four subdomains, each mapped to two GPUs, and two MPIs are run on each GPU to accelerate sparse matrix solutions.

[0094] Real-time feature input: The edge intelligent gateway uploads feature vectors (including 30 dimensions such as entropy, main frequency, and strain increment) every 30 seconds.

[0095] The Python service layer adjusts the load boundary conditions based on the latest features: applying equivalent normal stress Δσ and shear stress Δτ on the fracture surface;

[0096] Simulation delay: Single simulation step length Δt = 60s, total calculation time is about 1s, which can meet the requirements of hour-level rapid feedback;

[0097] Forecast decision module structure and process:

[0098] Real-time acquisition:

[0099] Latest simulation results: Maximum fracture surface opening distance δ max , stress concentration ratio C=σ local / σ far ;

[0100] Historical monitoring sequence: multimodal features every 30 seconds over the past 24 hours;

[0101] The two parts of data are spliced ​​into a feature sequence with a time window length of 48 (24 hours) × 1 (one per simulation hour) = 48 + 48 = 96;

[0102] GNN input graph construction:

[0103] Node: represents each sensor substation (N≈5);

[0104] Edge weight: a weighted matrix calculated based on spatial distance and historical correlation;

[0105] Number of layers and parameters: 2-layer GraphSAGE, 64 hidden dimensions, and mean aggregation function.

[0106] Output: Spatial context vector Si∈R for each node 64 ;

[0107] LSTM input: concatenate GNN output S=concat i（Si） ∈R （N×64） With the time series feature matrix F∈R （96×30） ;

[0108] Structure: Frost-layer LSTM, 128 hidden units per layer, dropout=0.2;

[0109] Output: time series context vector T∈R 128 ;

[0110] Concatenate Sflat (which is fully connected and reduced to 128 dimensions) with T to obtain H∈R 256 ;

[0111] Through a two-layer fully connected network, softmax obtains four levels of risk probability P;

[0112] The initial model is built on the cloud using labeled data (N≈10 4 Articles, covering four levels of risk samples) training;

[0113] After local deployment, fine-tuning is performed using newly labeled data during daily off-peak hours;

[0114] The maximum probability level L* in P is used as the current warning level;

[0115] If L*≥Level III, then:

[0116] The warning level, P vector, simulation snapshot, and time series trend chart are packaged into JSON and pushed to the supervision app and central control screen via 5G and industrial Ethernet dual channels;

[0117] At the same time, the emergency linkage module is activated to execute the corresponding ventilation, lighting and escape indication switching.

[0118] Example 5

[0119] See also Figure 1 A disaster early warning system for tunnels crossing active fault zones, the system also includes an emergency linkage module and a remote diagnosis and maintenance module. The emergency linkage module is used to push the warning results to the mobile monitoring app and the central control room visualization screen via 5G wireless and industrial Ethernet, and automatically trigger the tunnel ventilation, lighting and escape indication systems;

[0120] The emergency linkage module uses RESTful and MQTT dual-protocol communication, supports dual platforms on the mobile terminal, and generates automatic reports in standard PDF format, including warning levels, simulation snapshots, and emergency response recommendations;

[0121] The remote diagnosis and maintenance module triggers the gateway health status to be uploaded to the chain through the blockchain smart contract, and combines with the off-chain data analysis platform to provide proactive maintenance warnings for the gateway and sensor status;

[0122] Furthermore, the edge gateway collects metrics such as its own CPU / GPU utilization, memory usage, on-chain transaction latency, and sensor communication packet loss rate every 30 minutes; calculates a status code (0 = normal, 1 = mild abnormality, 2 = major abnormality); uploads the status code and log digest (SHA-256) to the chain to trigger a HealthReport event; subscribes to the Fabric event stream and stores the HealthReport in a time series database (InfluxDB); the Grafana dashboard displays the health curve of each node in real time, and configures alarm rules: if any node has a status code ≥ 1 three times in a row, maintenance personnel will be notified via email and SMS.

[0123] When a single-node memory leak is detected (memory usage > 85% and lasts > 15 minutes) or on-chain transaction delay > 200ms:

[0124] Automatically generate and send work orders via email (including fault snapshots and log download links);

[0125] Highlight abnormal nodes on the "System Status" page of the central control screen;

[0126] Conduct emergency drills on the above processes once a month, simulating the switching of level Ⅰ→Ⅳ warnings and maintenance alarms to ensure the availability of the entire link from linkage to maintenance; calculate the average response time of the drills (target ≤30s) and the average processing time of maintenance work orders.

[0127] Working Principle: Fiber-optic DAS, MEMS accelerometers, GNN and INS tilt strain and acoustic cavity sensors are deployed along the fault zone. The data is processed through an FPGA+ARM gateway, which performs bandpass filtering, adaptive wavelet denoising, and multi-domain feature extraction to obtain highly sensitive, low-noise real-time monitoring information.

[0128] The edge intelligent gateway uses an online transfer learning module based on pre-training and self-supervised comparative learning to make local classification decisions based on extracted features. It also stores important feature vectors and warning results on the Hyperledger Fabric blockchain, ensuring real-time data and traceability.

[0129] The central control server builds a three-dimensional finite element coupling model, and uses the GPU-accelerated stress-displacement simulation engine to simulate the fracture evolution at the hourly level. It then combines the GNN+LSTM multi-scale deep neural network, integrates simulation and historical sequences, and generates graded warnings of levels I-IV.

[0130] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A disaster warning system for tunnels crossing active fault zones, characterized by: The system includes a multimodal sensing layer for acquiring microseismic wave, vibration, tilt and strain data along the tunnel fault zone, the multimodal sensing layer including: Fiber-optic distributed acoustic sensing units (DAS) are installed every 10 meters along the inner side of the tunnel lining to monitor tiny acoustic signals from the fracture surface in real time. MEMS accelerometer arrays are placed at the four corners of each section to detect high-frequency vibrations; GNSS and INS combined tilt and strain sensors are installed on the tunnel roof and side walls to obtain millimeter-level tilt and strain changes; The acoustic cavity sensor is placed in the area near the sliding surface of the fault zone to capture the air acoustic wave precursor signal.

2. The disaster warning system for tunnel crossing an active fault zone according to claim 1, characterized in that: The multimodal sensing layer is directly connected to the edge intelligent gateway via industrial Ethernet. The edge intelligent gateway includes a combination of FPGA and ARM Cortex-A series processor architecture, which is used to perform: Noise filtering processing uses an adaptive wavelet threshold algorithm to remove environmental and construction noise; Online extraction of multiple features in time domain, frequency domain, time-frequency domain and entropy value; The online transfer learning module adaptively updates the feature extraction and classification models based on the initial fracture simulation data.

3. The disaster warning system for tunnel crossing an active fault zone according to claim 1, characterized in that: The system also includes a distributed blockchain storage module, which is built based on Hyperledger Fabric. The edge intelligent gateway is the blockchain network node. The distributed blockchain storage module is used to perform lightweight on-chain storage of feature vectors and preliminary warning results to ensure that the data cannot be tampered with and is traceable throughout the process.

4. The disaster warning system for tunnel crossing an active fault zone according to claim 1, characterized in that: The system also includes a digital twin simulation module, which is deployed on the central control server and includes: A three-dimensional tunnel-stratum-fault zone coupling model was established based on the finite element method; The stress-displacement coupling simulation engine supports GPU acceleration and has a simulation latency of ≤1s, which is used to meet the needs of fast updates at the hourly level. The stress-displacement coupling simulation engine receives real-time feature input from the edge intelligent gateway to simulate the fracture evolution process.

5. The disaster warning system for tunnel crossing active fault zone according to claim 1 is characterized by: The system also includes a prediction and decision-making module that integrates a graph neural network (GNN) with a long-term and short-term network and a LSTM combined architecture to fuse simulation data with historical monitoring sequences to generate multi-scale disaster warnings of levels I to IV.

6. The disaster warning system for tunnel crossing an active fault zone according to claim 1, characterized in that: The system also includes an emergency linkage module and a remote diagnosis and maintenance module. The emergency linkage module is used to push the warning results to the mobile monitoring app and the central control room visualization screen via 5G wireless and industrial Ethernet, and automatically trigger the tunnel ventilation, lighting and escape indication systems. The emergency linkage module uses RESTful and MQTT dual-protocol communication, supports dual platforms on the mobile terminal, and generates automatic reports in standard PDF format, including warning levels, simulation snapshots, and emergency response recommendations; The remote diagnosis and maintenance module triggers the gateway health status to be uploaded to the chain through the blockchain smart contract, and combines with the off-chain data analysis platform to provide active maintenance warnings for the gateway and sensor status.

7. The disaster warning system for tunnel crossing an active fault zone according to claim 1, characterized in that: The fiber-optic distributed acoustic wave sensing unit adopts φ9 / 125μm single-mode optical fiber and phase-sensitive OTDR technology, with a spatial resolution of ≤0.5m and a sampling frequency of ≥10kHz; The MEMS accelerometer array uses a three-axis ±4g range sensor, each with a bandwidth of 0.5Hz-2kHz and dual software and hardware temperature compensation. The GNSS and INS combined tilt strain sensor achieves tunnel section micro-deformation detection accuracy of ≤0.2mm through a four-point differential layout.

8. The disaster warning system for tunnel crossing an active fault zone according to claim 2, characterized in that: The online transfer learning module deployed in the edge intelligent gateway is pre-trained based on large-scale source domain fracture simulation data, and uses self-supervised contrastive learning to fine-tune newly collected data; when the model performance degrades by more than 5%, remote updates are automatically triggered.

9. The disaster warning system for tunnel crossing an active fault zone according to claim 3, characterized in that: The blockchain storage module supports dynamic node management, document block size ≤10KB, chain latency ≤100ms, and hash storage of large data feature vectors through the off-chain Merkle tree structure.

10. The disaster warning system for tunnel crossing active fault zone according to claim 5, characterized in that: The graph neural network submodule in the prediction and decision module is used to construct the spatial topological relationship between sensor points, and the long short-term memory network submodule is used to capture temporal features; the two are integrated through the attention mechanism, and the overall model has a single prediction accuracy of ≥92%.

Images