A reservoir earthquake chain risk intelligent identification system and method based on internet of things and block chain

Abstract

The application discloses a reservoir earthquake chain risk intelligent identification system and method based on the Internet of Things and a blockchain, and belongs to the field of disaster prevention and mitigation of water conservancy projects. The system comprises an Internet of Things sensing and trusted data chaining module, a topographic constraint mechanism modeling and feature extraction module, a chain risk intelligent identification and dynamic deduction module, and an intelligent contract driven early warning and response execution module. The method comprises collecting data and chaining storage for evidence, constructing a topographic coupling mechanism model to extract risk features, using a deep learning model for intelligent identification and chain deduction, and automatically triggering early warning response through an intelligent contract. The application solves problems such as poor monitoring data credibility, shallow topographic physical mechanism modeling, weak chain risk deduction capability, and delayed early warning response, and realizes full-chain trusted perception, fine modeling, intelligent forward deduction, and immediate automatic prevention and control of reservoir earthquake chain risks.

Description

Technical Field

[0001] This invention relates to the field of water conservancy engineering and disaster prevention and mitigation technology, specifically to a reservoir earthquake chain risk intelligent identification system and method based on the Internet of Things and blockchain. Background Technology

[0002] With the continuous advancement of national water network projects and major water conservancy infrastructure construction, reservoir-induced earthquakes and their potential chain-like geological disasters have become a major risk source threatening dam safety and watershed ecological stability. During the water storage and discharge cycle, reservoirs significantly disturb the regional geostress field through pore water pressure diffusion and reservoir water load, potentially activating hidden faults or weakening rock mass structures, thereby triggering microseismic activity and, under specific topographic conditions, inducing secondary disasters such as landslides, surges, and even dam failures, forming a chain-like risk evolution process with cascading, nonlinear, and spatial propagation characteristics. Existing risk identification systems generally rely on isolated monitoring indicators or static geological maps, making it difficult to dynamically depict the risk generation and transmission mechanisms under the coupled effects of multiple factors such as water, rock, faults, and topography, resulting in delayed early warning, ambiguous positioning, and disconnected response.

[0003] While IoT technology has enabled real-time sensing of multiple parameters such as water level, seepage pressure, microseismic activity, and displacement in reservoir safety monitoring, the lack of a unified and reliable data storage mechanism for each sensor node makes it susceptible to equipment drift, communication interruptions, or malicious tampering, resulting in a dilemma where the data is usable but not reliable. At the same time, current risk analysis models mostly use threshold over-limit alarms or simple statistical regression methods, failing to deeply integrate the key physical constraint factor of terrain undulation into seismic wave propagation modeling and energy attenuation calculation. This makes it impossible to accurately reflect the regulatory effect of landforms such as canyons and steep slopes on seismic wave focusing, scattering, and secondary disaster triggering paths, resulting in a serious lack of accuracy in risk identification in complex terrain reservoir scenarios.

[0004] Existing technologies, when addressing reservoir-induced seismic chain risks constrained by terrain, exhibit systemic flaws, including incomplete monitoring data, logical breaks in multi-source information fusion, superficial modeling of terrain physical mechanisms, weak ability to extrapolate risk transmission paths, and a loose early warning execution chain. Specifically, on the one hand, the lack of a cryptographically secured trust transfer mechanism between the perception and decision-making layers makes it difficult for high-frequency, high-dimensional monitoring data to support high-confidence intelligent identification. On the other hand, the risk assessment process severs the intrinsic connection between terrain geometry and seismic dynamics, failing to construct a physical-data dual-driven extrapolation framework from raw waveforms to the disaster chain. These problems make it difficult for existing systems to achieve the closed-loop prevention and control goals of early detection, accurate source tracing, rapid extrapolation, and automatic response when facing sudden, highly concealed, and spatially proliferating reservoir-induced seismic chain risks.

[0005] Therefore, there is an urgent need to establish a new paradigm for intelligent identification that deeply integrates real-time perception of the Internet of Things, trusted governance of blockchain, and the physical mechanism of terrain constraints. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a smart identification system and method for chain-like risks of reservoir earthquakes based on the Internet of Things and blockchain.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] This application provides a reservoir earthquake chain-based intelligent risk identification system based on the Internet of Things and blockchain, including:

[0009] The IoT sensing and trusted data on-chain module is used to collect multi-source monitoring data through a distributed sensor network deployed in the reservoir area, and to complete the trusted storage of data based on the blockchain consensus mechanism.

[0010] The terrain constraint mechanism modeling and feature extraction module is used to construct a coupled model of seismic wave field propagation and geostress field disturbance that integrates the physical constraints of terrain undulation based on the trusted evidence data and high-precision terrain data, so as to extract risk characteristics under terrain constraints.

[0011] The chain-based risk intelligent identification and dynamic simulation module is used to intelligently identify risk sources, assess risks, and dynamically simulate and simulate chain-based risk transmission paths based on the risk characteristics through a machine learning model.

[0012] The smart contract-driven early warning and response execution module is used to compare the simulation results with preset dynamic risk threshold rules related to terrain and geological conditions, and automatically trigger early warning and collaborative response actions through smart contracts deployed on the blockchain.

[0013] Optionally, in the IoT sensing and trusted data on-chain module, the distributed sensor network includes microseismometers, displacement monitoring stations, piezometers, and water level gauges; the trusted data storage process is completed based on an improved practical Byzantine fault-tolerant consensus mechanism.

[0014] Optionally, in the terrain constraint mechanism modeling and feature extraction module, the coupling model is a terrain-dependent variable coefficient wave equation, whose coefficients are determined by the elevation and lithological properties of the spatial location, and is used to accurately locate and correct the source parameters of monitored seismic events.

[0015] Optionally, in the chain-based intelligent risk identification and dynamic inference module, the machine learning model is a deep learning model that integrates spatiotemporal graph neural networks and sequence learning networks.

[0016] Optionally, in the smart contract-driven early warning and response execution module, the dynamic risk threshold rule divides the reservoir area into multiple sub-regions with different risk characteristics based on the digital elevation model, and sets an independent combination of threshold parameters for each sub-region.

[0017] Secondly, this application provides a method for intelligent identification of chain-like risks of reservoir earthquakes based on the Internet of Things and blockchain, including:

[0018] S1: Collect multi-source monitoring data of the reservoir area and complete the trusted storage of data based on blockchain technology;

[0019] S2: Based on the aforementioned reliable evidence data and high-precision terrain data, construct a coupled model of seismic wave field propagation and geostress field disturbance under terrain constraints, and extract risk features;

[0020] S3: Input the risk characteristics into the machine learning model to perform intelligent risk identification and dynamic deduction of the chain transmission path;

[0021] S4: Based on the aforementioned deduction results, a comparison is made using dynamic risk threshold rules associated with terrain and geological conditions, and an early warning and response are automatically triggered via a smart contract.

[0022] Optionally, in step S2, constructing the coupled model includes: calculating the dynamic perturbation tensor of the regional geostress field jointly induced by reservoir water load and pore pressure changes based on elasticity mechanics and porosity elasticity theory.

[0023] Optionally, in step S3, the dynamic simulation is implemented through a sequence generation algorithm, starting from the current risk state and constrained by the probability of disaster triggering and propagation determined by terrain and geological conditions, and performing multi-step forward simulation.

[0024] Optionally, in step S2, constructing the coupled model includes: solving the wave equation related to topographic spatial variables to obtain the seismic wave field propagation characteristics that reflect the influence of topography.

[0025] Optionally, in step S1, the data trustworthiness and evidence storage process is based on an improved Byzantine fault-tolerant consensus mechanism, in which multiple authorized verification nodes verify the consistency of the data before it is uploaded to the blockchain.

[0026] Compared with the prior art, this application has the following beneficial effects:

[0027] This invention proposes a reservoir earthquake chain-based intelligent risk identification system and method based on the Internet of Things (IoT) and blockchain. By deeply integrating real-time IoT sensing data with blockchain's distributed notarization and consensus verification technologies, a trusted data protection system is constructed across the entire chain from the data source to the application terminal. This fundamentally solves the problems of easily tampered, unclear sources, and difficulties in collaboration with monitoring data, providing a solid data foundation for high-confidence risk decision-making. It innovatively embeds high-precision topographic geometry and physical parameters into the seismic wave propagation model and geostress field calculation, achieving refined modeling of wave field distortion, energy anisotropy attenuation, and secondary disaster triggering conditions caused by topographic undulations. This significantly improves the accuracy of reservoir earthquake risk source location in complex terrain environments. The simulation of risk evolution is highly realistic. Employing a deep learning architecture that integrates spatiotemporal graph convolution and long short-term memory networks, it can simultaneously capture the spatial correlation of risks across multiple monitoring sites and their dynamic evolution over time. This enables intelligent identification, grading, and multi-step forward-looking extrapolation of reservoir-related seismic chain risks that are highly concealed and involve numerous coupling factors. By automatically connecting the dynamic risk identification results with smart contract rules based on terrain partitioning deployed on the blockchain, it achieves immediate, automatic, and non-repudiable triggering of risk warnings, directly driving physical response equipment. This constructs an integrated closed loop of perception, identification, warning, and execution, significantly shortening the lag time from risk identification to emergency response and enhancing the proactive prevention and control capabilities of major water conservancy projects. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention.

[0029] Figure 2 This is a schematic diagram illustrating the principle framework of terrain constraint mechanism modeling and feature extraction in this invention.

[0030] Figure 3 This is a flowchart illustrating the chain-based intelligent risk identification and dynamic simulation process in this invention.

[0031] Figure 4 This is a schematic diagram illustrating the interaction between early warning and response execution driven by smart contracts in this invention.

[0032] Figure 5 This is a flowchart illustrating the process framework for IoT sensing and trusted data uploading in this invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Furthermore, in this invention, an element referred to as fixed to or disposed on another element may be directly disposed on the other element, or there may be an intermediate element. When an element is considered to be connected to another element, it may be directly connected to the other element, or there may be an intermediate element present simultaneously. The terms vertical, horizontal, left, right, and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0035] Example 1

[0036] Based on the overall architecture of this invention, this embodiment elaborates in detail the technical principles, core formulas, and collaborative workflows of each module:

[0037] See Figure 1 The system consists of an IoT sensing and trusted data on-chain module, a terrain constraint mechanism modeling and feature extraction module, a chain-based risk intelligent identification and dynamic inference module, and a smart contract-driven early warning and response execution module connected in sequence. Each module interacts with trusted data and instructions through the blockchain network.

[0038] The implementation of the method includes the following key steps:

[0039] Trusted Data Sensing and On-Chain: A distributed sensor network deployed in the reservoir area collects multi-source monitoring data in real time. After hashing and timestamping, the data forms transaction proposals, which are cross-validated by a group of validator nodes based on the Practical Byzantine Fault Tolerance (PBFT) consensus mechanism. Upon successful cross-validation, a trusted data block is generated and uploaded to the blockchain. The system is designed to handle at least 500 sensor data transactions per second, with an average latency of less than 2 seconds from data acquisition to completion of blockchain consensus and notarization, meeting the demands of high-concurrency real-time monitoring. To ensure reliability under extreme network conditions, the system features local caching and degradation mechanisms: when the blockchain network experiences a partition or more than one-third of the validator nodes become disconnected, each edge gateway node automatically encrypts and temporarily stores the data in a local Trusted Execution Environment (TEE). Once the network recovers, the data is automatically synchronized and batch uploaded to the blockchain for consistency verification.

[0040] The core of this module is a distributed sensor network deployed throughout the reservoir monitoring area. This network employs a heterogeneous node deployment strategy to achieve comprehensive coverage of multi-physics information. Specifically, high-precision microseismometer arrays are deployed inside the concrete dam body and abutments to capture elastic wave signals generated by micro-fractures within the rock mass; global navigation satellite system displacement monitoring stations are installed at potential landslide sites, fault outcrops, and around the reservoir basin to continuously monitor horizontal and vertical displacements at the millimeter to centimeter level; a network of piezometers is embedded in the dam foundation, seepage channels around the dam, and within the reservoir bank rock mass to measure the dynamic changes in pore water pressure in real time; and water level gauges are installed at representative locations at different elevations within the reservoir to continuously record water level fluctuations. All sensor nodes have independent power supply and wireless communication capabilities. Each sensor node integrates lightweight blockchain client software. Upon completion of each data acquisition cycle, the client immediately performs the following operations: standardizes and formats the acquired raw monitoring data (such as a microseismic waveform, a set of displacement coordinates, and a pressure value); calculates the hash value of the formatted data packet using a hash algorithm; obtains the GPS timestamp from the built-in high-precision clock; reads the unique identification code of this node; and obtains the hash value of the previously confirmed data block. Subsequently, the client encapsulates this information into a data transaction proposal to be uploaded to the blockchain and broadcasts it via a wireless network to a pre-configured group of blockchain validator nodes. The data upload process employs an improved Practical Byzantine Fault-Tolerant Consensus mechanism. The validator node group consists of server nodes run by authoritative entities such as reservoir management agencies, geological monitoring centers, and dam safety monitoring centers. Upon receiving a data transaction proposal, each validator node executes cross-validation logic: verifies the legitimacy of the sensor's identification; compares the continuity and rationality of the data timestamp; and, through sampling and feedback instructions, requests the sensor node to sign and confirm the raw data to verify the authenticity of the data source. Only when more than two-thirds of the validator nodes reach a consensus on the same batch of data transaction proposals will the data be allowed to be packaged. Once consensus is reached, the master validator packages multiple verified data transaction proposals, the current timestamp, the hash value of this block, and the hash link of the previous block into a new trusted data block. This block is cryptographically linked to the end of the blockchain and synchronized to all nodes in the network for distributed storage. At this point, the original monitoring data has undergone a trusted transformation from the physical world to the digital world, possessing tamper-resistant and fully traceable characteristics, providing a solid data foundation for subsequent analysis. The logical flow of this process can be found in [link to relevant documentation]. Figure 5 .

[0041] Terrain Coupling Mechanism Modeling and Feature Extraction: The terrain constraint mechanism modeling and feature extraction module is launched. This module is deployed on a cloud or local server cluster with high-performance computing capabilities. Its primary task is to receive and parse trusted data blocks from the blockchain ledger, while simultaneously loading a high-precision digital elevation model of the target reservoir area. The accuracy of the digital elevation model is typically required to reach sub-meter level, capable of finely depicting the terrain undulations, slope, aspect, and valley morphology of the reservoir area. See also... Figure 2 The core of this module is constructing a terrain-constrained seismic wave field propagation model. Specifically, using the discrete elevation lattice provided by the digital elevation model as input, a three-dimensional non-uniform rational B-spline surface fitting algorithm is employed to generate a continuously differentiable terrain elevation function. This function defines the elevation value corresponding to any plane coordinate point on the reservoir surface. Based on this function, the system automatically calculates the terrain gradient vector, curvature tensor, and aspect angle for each calculation grid point. In the numerical simulation of seismic wave propagation, the traditional scalar wave equation in a homogeneous isotropic medium is modified into a terrain-dependent variable-coefficient wave equation. The form of this equation is:

[0042]

[0043] in For wave velocity fields related to topographic elevation and lithology, The damping coefficient tensor characterizes wavefront distortion and energy attenuation caused by topography. This equation is solved using the finite difference method to obtain a library of Green's functions under topographic constraints, which can be used for precise location and energy correction of subsequent microseismic events.

[0044] The wave velocity field is no longer a constant, but a function of spatial coordinates, its value determined by the elevation of the point and the lithology mapped from geological exploration data. For example, the surface overburden and deep bedrock have different wave velocities. The damping coefficient tensor is a function directly related to topographic gradient and curvature, used to quantitatively characterize wavefront scattering, diffraction, and the resulting anisotropic energy attenuation that occur when seismic waves encounter topographic undulations (such as ridges and canyons) during propagation. The system uses the staggered grid finite difference method to numerically solve this variable-coefficient wave equation. The solution process requires defining the source function, setting absorbing boundary conditions, and iteratively progressing throughout the entire three-dimensional computational domain containing complex terrain. By setting a series of virtual source points and solving them, a Green's function library under topographic constraints can be pre-calculated and established. This library essentially stores the wavefield propagation response, including topographic influences, from any potential source point within the library area to each microseismometer receiving point. After extracting real microseismic event waveform data from the trusted data block, the module calls the Green's function library and employs a double-difference localization algorithm based on waveform cross-correlation to accurately locate the microseismic events. Simultaneously, the module uses Green's function to correct the observed waveforms, retrieving a more accurate focal mechanism solution, including the fault strike, dip angle, slip angle, and seismic moment. On the other hand, the module processes water level and pore water pressure data in parallel to calculate the dynamic disturbance of the regional geostress field induced by reservoir water load changes. The specific implementation steps are as follows: Based on real-time reservoir water level data and the reservoir's capacity-area curve, the dynamic distribution of reservoir water mass is transformed into a normal surface load distribution function applied to the basement rock mass of the reservoir basin. Combined with the spatial diffusion data of pore water pressure monitored by the piezometer network, coupled calculations are performed based on Biot's consolidation theory. This theory considers porous media as a coupled body of an elastic solid skeleton and fluid. The calculation process requires input parameters such as the rock mass's elastic modulus, Poisson's ratio, permeability coefficient, and porosity. By solving the coupled equations, the regional geostress field increment tensor caused by the combined effects of reservoir water loading and pore pressure changes is obtained. This tensor reflects the "loading" or "unloading" effect of reservoir operation on the underground stress state. Finally, the module fuses the precisely located source parameters, the corrected focal mechanism solution, the calculated geostress field increment tensor, and the topographic feature parameters of the source points and potential hazard areas extracted from the terrain model, normalizing them to generate a unified multi-dimensional risk feature vector. This vector contains information on spatial location, intensity, stress state, and topographic environment.

[0045] Next, the chain-based risk intelligent identification and dynamic inference module begins operation. The core of this module is a pre-trained chain-based risk identification neural network model; see [link to relevant documentation]. Figure 3This model employs an encoder-decoder architecture, deeply integrating a spatiotemporal graph convolutional network and a long short-term memory network. The model's input is the multi-dimensional risk feature vector sequence generated by the previous module. First, the encoder utilizes a long short-term memory network layer to encode the input time-series features, capturing the long-term dependencies and evolutionary trends of risk features over time, and outputting a latent state vector containing temporal context information. Simultaneously, the system constructs a dynamic risk association graph using all monitoring stations within the reservoir area as nodes. The attributes of nodes in the graph represent their risk features, and the edge weights are determined by the spatial distance between nodes, geological connectivity, and terrain accessibility. The spatiotemporal graph convolutional network operates on this graph: the spatial convolutional layers aggregate the feature information of each node and its first- and second-order neighbor nodes, ensuring that the risk status assessment of each station considers the overall situation of its surrounding area; the temporal convolutional layers further capture the temporal dynamics of feature propagation on the graph. The encoder ultimately outputs a comprehensive risk representation integrating spatiotemporal information; then, the decoder is responsible for risk inference and result generation. For risk source identification and assessment, the decoder maps the comprehensive risk representation into specific outputs through a fully connected layer: including the most likely three-dimensional spatial coordinates of the risk source, the estimated magnitude, and a current risk level score calculated by integrating all current features. For dynamic simulation and extrapolation of risk transmission paths, the decoder enables a sequence generation mechanism; it uses the current risk state latent vector as the initial seed and the energy attenuation coefficient matrix and geological vulnerability index map provided by the terrain constraint model as physical constraints to simulate the chain-like development process of disaster events; the extrapolation uses a Monte Carlo tree search algorithm for multi-step forward exploration. Each simulation begins with a hypothetical mainshock event, and the algorithm assesses the probability of triggering secondary disasters along terrain channels and geologically weak zones at that source location and energy level. For example, the algorithm calculates the dynamic load of seismic waves on steep slopes and assesses their instability probability; or assesses whether an earthquake might trigger liquefaction of sediment at the reservoir bottom, thereby affecting dam stability; each simulation generates a possible event chain sequence. After numerous simulations, the system statistically analyzes all generated paths, outputting multiple potential risk transmission paths with different probability weights, and providing a probability distribution map of risk evolution over multiple future time steps. The training data for this deep learning model comes from long-term monitoring data blocks stored in historical blockchain ledgers that have undergone post-event verification, as well as expert-annotated information on historical reservoir earthquake events and their chain disasters corresponding to these data time periods. The loss function used during training is a weighted sum of risk location error, magnitude estimation error, and risk level classification cross-entropy loss, and the network parameters are optimized using a backpropagation algorithm.

[0046] The smart contract-driven alert and response execution module has entered standby mode. (See also...) Figure 4The core of this module is a smart contract for early warning deployed on the blockchain and a linked physical execution terminal. The smart contract embeds a set of dynamic risk threshold rules based on terrain zoning. The system first divides the reservoir area into multiple sub-regions with different risk characteristics based on a high-precision digital elevation model and geological data, such as the deep-water area of ​​the reservoir basin, the steep reservoir bank area, the known fault-affected area, the near-dam area, and the downstream residential area. For each sub-region, a set of independent, dynamically adjustable risk threshold parameters is set. These parameters include: the microseismic event frequency threshold per unit time, the cumulative seismic energy release threshold, the surface displacement acceleration rate threshold, and the daily water level change rate threshold. The smart contract is programmed to continuously monitor transaction events on the blockchain network related to new risk assessment results. When the chain-based risk intelligent assessment and dynamic deduction module generates a new transaction from its output and uploads it to the chain, the smart contract is automatically triggered. The contract logic reads the data from the transaction, including the risk level score, the sub-region where the risk source is located, and the risk evolution probability distribution. Subsequently, the contract compares the input data with the preset risk threshold rules for that sub-region one by one. The judgment logic is as follows: if the current risk level score exceeds a threshold, or if the risk evolution probability distribution shows that the probability of a chain disaster occurring within a specific period such as the next 24 hours or 48 hours exceeds a preset probability threshold, then the trigger condition is met. Once triggered, the smart contract automatically executes its internal immutable code logic. This logic first matches the most suitable response plan from a pre-set digital emergency plan library based on the trigger rule type and risk level. The plan library stores structured response measures for different risk scenarios. The contract then generates a standardized warning instruction message. This message typically includes the following fields: unique warning identifier, trigger time, precise coordinates of the risk source, risk type, risk level, list of affected sub-regions, suggested response level, and a list of specific response measures. After generating the warning instruction, the contract automatically initiates a new transaction, permanently recording this warning instruction and its generation context on the blockchain as the transaction content, completing the notarization of the warning action and ensuring its non-repudiation. Almost simultaneously, the blockchain network broadcasts this transaction containing the warning instruction to all emergency response terminals that have subscribed to the relevant topic.

[0047] The receiving and execution terminals for early warning commands include various emergency response devices. For example, after receiving a command, the centralized control system located in the dam's central control room can automatically send an opening command to the floodgate control system, either automatically or after confirmation by on-duty personnel, if the command suggests adjusting the operating conditions. Wireless broadcast sirens and loudspeaker arrays in the reservoir area and downstream banks, upon receiving the command, play corresponding early warning audio messages based on the warning level. Information release platforms of the reservoir management unit and local government emergency departments, upon receiving the command, automatically generate and release announcements, SMS messages, or application push notifications containing risk information. Through this process, the system achieves a fully automated and reliable closed loop from risk perception and intelligent identification to early warning issuance and physical execution, greatly improving response speed and reliability.

[0048] The impact of topography on seismic wave propagation is quantified using physical methods. Simultaneously, based on Biot's consolidation theory, reservoir water level load and pore water pressure data are coupled to calculate the dynamic disturbance of the regional geostress field induced by reservoir water activity. Finally, multi-source information is fused to generate a multi-dimensional risk feature vector. This physically coupled model reduces the microseismic event location error under complex terrain by an average of over 60% compared to traditional homogeneous medium models.

[0049] Chain-like Risk Intelligent Identification and Dynamic Inference: Risk feature vectors are input into a pre-trained deep learning model. This model employs an architecture that integrates a spatiotemporal graph convolutional network (ST-GCN) and a long short-term memory network (LSTM) to capture the correlation and evolution patterns of risks in the spatiotemporal dimensions. The model outputs risk source parameters and the current risk level. Furthermore, through a sequence generation mechanism and a Monte Carlo Tree Search (MCTS) algorithm, multi-step forward inference of chain disasters is performed, outputting potential risk transmission paths and risk evolution probability distribution maps. During the training phase, the deep learning model used a dataset covering tens of thousands of historical microseismic events and their corresponding chain disaster cases. Its binary classification accuracy for determining whether a chain disaster is likely to be triggered on the independent test set reached 92%, significantly higher than the approximately 70% accuracy of the traditional thresholding method.

[0050] Smart contract-driven automatic early warning and execution: The system divides the reservoir area into multiple sub-regions based on terrain and geological conditions, and sets differentiated dynamic risk threshold rules for each region. The early warning smart contract is deployed on the blockchain and continuously monitors the assessment results; when the risk exceeds the threshold, the contract automatically triggers, matches a solution from the contingency plan library, generates an early warning command, and drives emergency equipment to respond. To cope with complex risk scenarios, the smart contract embeds response priority and mutual exclusion logic: for example, flood discharge commands involving dam structural safety have the highest priority; broadcast alarms and information releases can be executed in parallel; however, large-scale flood discharge commands and scheduling commands to maintain high water levels for power generation are mutually exclusive and must be automatically arbitrated by the contract based on real-time risk levels.

[0051] Example 2

[0052] This embodiment uses a large hydropower station reservoir with steep terrain in southwestern my country as a scenario to illustrate the specific implementation and quantitative effects of the technical solution.

[0053] System Deployment and Trusted Data On-Chain: Over 200 monitoring nodes, including microseismometers, GNSS stations, piezometers, and water level gauges, were deployed across a 30-kilometer reservoir area. Each node integrates a blockchain light client. Data transaction proposals are verified by a group of seven nodes using an improved PBFT consensus mechanism. In actual deployment, the system ran continuously and stably for 180 days, processing over 200 million data transactions without any valid data tampering or loss, achieving a consensus success rate of 99.95%. The median end-to-end latency from data collection to on-chain confirmation was 1.5 seconds, fully meeting the requirements for real-time risk assessment.

[0054] Terrain-constrained mechanism modeling and feature extraction: The system loads a 0.5-meter resolution lidar DEM. A variable-coefficient wave equation is constructed and solved using the staggered-grid finite-difference method, with a pre-established Green's function library. This terrain-constrained model improves the location accuracy of microseismic events from an average error of approximately 300 meters using traditional methods to within 50 meters (an improvement of over 83%). The geostress calculation module is based on real-time data and rock mass parameters (elastic modulus E = 30 GPa, permeability coefficient k = 1e). -6 Solve the Biot coupling equation (m / s) to quantitatively output the stress disturbance.

[0055] Chain-based intelligent risk identification and dynamic extrapolation: A deep learning model is trained on a dataset containing over 30,000 historical events spanning 10 years. In a typical event, after inputting a feature sequence, the model outputs: the location of the mainshock (longitude X, latitude Y, depth 2.1 km) and an estimated magnitude. =2.5, current risk level orange. The model assesses the overall probability of this event triggering a chain reaction disaster as 35%. The simulation module (MCTS algorithm simulation 1000 times) shows that there are two main transmission paths in the next 24 hours: Path 1 (probability 32%) may trigger local instability of a steep slope upstream; Path 2 (probability 18%) may cause a small-scale surge near the dam area.

[0056] Smart contract-driven early warning and response execution: The reservoir is divided into 5 sub-regions (e.g., Zone I - reservoir basin area, Zone II - steep slope area, etc.); the risk threshold for each region is jointly calibrated based on its historical statistical characteristics (e.g., the 95th percentile of the number of microseismic events) and mechanical stability simulation. For example, the threshold for Zone II (steep slope area) is set as follows: daily frequency of microseismic events N ≥ 10, or cumulative energy E ≥ 1e8J.

[0057] When the above The event, value 2.5, was determined to be located in Zone II, and its computational energy, E=1.2e8J, exceeded the threshold, automatically triggering the early warning smart contract. The contract executed contingency plan "C-03," generating a structured early warning instruction and uploading it to the blockchain. The instruction contained clear priority responses: first, it activated video surveillance near the risk source; second, it sent a stop-work instruction to nearby work units; and third, it prepared for a scheduling meeting. The entire process, from risk assessment to instruction issuance, was completed within 3 minutes, achieving a closed-loop response within minutes.

[0058] 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 implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0059] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A smart reservoir earthquake chain risk identification system based on the Internet of Things and blockchain, characterized in that, include: The IoT sensing and trusted data on-chain module is used to collect multi-source monitoring data through a distributed sensor network deployed in the reservoir area, and to complete the trusted storage of data based on the blockchain consensus mechanism. The terrain constraint mechanism modeling and feature extraction module is used to construct a coupled model of seismic wave field propagation and geostress field disturbance that integrates the physical constraints of terrain undulation based on the trusted evidence data and high-precision terrain data, so as to extract risk characteristics under terrain constraints. The chain-based risk intelligent identification and dynamic simulation module is used to intelligently identify risk sources, assess risks, and dynamically simulate and simulate chain-based risk transmission paths based on the risk characteristics through a machine learning model. The smart contract-driven early warning and response execution module is used to compare the simulation results with preset dynamic risk threshold rules related to terrain and geological conditions, and automatically trigger early warning and collaborative response actions through smart contracts deployed on the blockchain.

2. The system according to claim 1, characterized in that, In the IoT sensing and trusted data on-chain module, the distributed sensor network includes microseismometers, displacement monitoring stations, piezometers, and water level gauges; the trusted data storage process is completed based on an improved practical Byzantine fault-tolerant consensus mechanism.

3. The system according to claim 1, characterized in that, In the terrain constraint mechanism modeling and feature extraction module, the coupling model is a terrain-dependent variable coefficient wave equation, whose coefficients are determined by the elevation and lithological properties of the spatial location, and is used to accurately locate and correct the source parameters of monitored seismic events.

4. The system according to claim 1, characterized in that, In the chain-based intelligent risk identification and dynamic deduction module, the machine learning model is a deep learning model that integrates spatiotemporal graph neural networks and sequence learning networks.

5. The system according to claim 1, characterized in that, In the smart contract-driven early warning and response execution module, the dynamic risk threshold rule divides the reservoir area into multiple sub-regions with different risk characteristics based on the digital elevation model, and sets an independent combination of threshold parameters for each sub-region.

6. A method for intelligent identification of chain-like risks of reservoir earthquakes based on the Internet of Things and blockchain, characterized in that, include: S1. Collect multi-source monitoring data of the reservoir area and complete the trusted storage of data based on blockchain technology; S2. Based on the aforementioned reliable evidence data and high-precision topographic data, construct a coupled model of seismic wave field propagation and geostress field disturbance under topographic constraints, and extract risk characteristics; S3. Input the risk characteristics into the machine learning model to perform intelligent risk identification and dynamic deduction of the chain transmission path; S4. Based on the aforementioned deduction results, a comparison is made using dynamic risk threshold rules associated with terrain and geological conditions, and an early warning and response are automatically triggered via a smart contract.

7. The method according to claim 6, characterized in that, In step S2, constructing the coupled model includes: calculating the dynamic perturbation tensor of the regional geostress field jointly induced by reservoir water load and pore pressure changes based on elasticity mechanics and pore elasticity theory.

8. The method according to claim 6, characterized in that, In step S3, the dynamic simulation is achieved through a sequence generation algorithm. Starting from the current risk state, and constrained by the probability of disaster triggering and propagation determined by terrain and geological conditions, a multi-step forward simulation is performed.

9. The method according to claim 6, characterized in that, In step S2, constructing the coupled model includes solving the wave equation related to topographic spatial variables to obtain the seismic wave field propagation characteristics that reflect the influence of topography.

10. The method according to claim 6, characterized in that, In step S1, the data trustworthiness and evidence storage process is based on an improved Byzantine fault-tolerant consensus mechanism, in which multiple authorized verification nodes verify the consistency of the data before it is uploaded to the blockchain.

Images