Dam multi-physical field dynamic simulation method and device based on data and mechanism fusion

Abstract

The invention provides a dam multi-physical field dynamic simulation method and device based on data and mechanism fusion. According to the dam multi-physics field dynamic simulation method based on data and mechanism fusion, the dynamic simulation precision of the dam multi-physics field coupling effect and the long-term service performance can be improved, and the hidden disease recognition capability and the early warning reliability under the extreme working condition are enhanced.

Description

Technical Field

[0001] This application relates to the field of dam safety monitoring and dynamic simulation technology, and in particular to a method and device for dynamic simulation of dam multiphysics fields based on data and mechanism fusion. Background Technology

[0002] Dams, as key infrastructure in water-rock-structure coupled systems, are widely used in flood control, power generation, and water resource allocation.

[0003] With the development of smart hydropower technology, dam performance analysis is gradually evolving from experience-driven to data-mechanism fusion-driven approaches. Related technologies typically construct a dam safety assessment system through collaborative operations of physical modeling, monitoring data acquisition and processing, and model inversion analysis. Specifically, this system covers the entire process from material performance degradation and seepage-stress interaction to basic geological evolution, including key aspects such as finite element simulation, sensor network deployment, data fusion, and anomaly identification. However, existing dam performance analysis methods directly employ traditional physical models and point-based sensor networks, failing to fully consider the strong nonlinearity and time-varying characteristics of multi-physics coupling effects, as well as the limitations of monitoring data in spatial coverage and temporal resolution. Specifically, physical models struggle to accurately characterize long-term service performance degradation patterns, while sensor networks suffer from blind spots and noise interference, resulting in low model-data matching and high uncertainty in inversion analysis. Under extreme operating conditions or in cases of hidden defects, this low matching can easily lead to misjudgments, thus affecting the reliability of dam safety situational awareness and early warning. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the first objective of this application is to propose a multi-physics dynamic simulation method for dams based on the fusion of data and mechanisms.

[0006] The second objective of this application is to propose a multiphysics dynamic simulation device for dams based on the fusion of data and mechanisms. The third objective of this application is to propose an electronic device.

[0007] The fourth objective of this application is to provide a computer-readable storage medium.

[0008] The fifth objective of this application is to provide a computer program product.

[0009] To achieve the above objectives, the first aspect of this application proposes a multiphysics dynamic simulation method for dams based on data and mechanism fusion, comprising:

[0010] S1. Construct a distributed spiking neural network sensing system to collect dam monitoring data in an event-driven manner and convert continuous signals into sparse pulse sequences to improve spatiotemporal resolution.

[0011] S2, establish a multi-scale ontology evolution model, and use a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process;

[0012] S3, using a gauge field neural network to constrain and optimize the evolution model, introduces SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data;

[0013] S4, based on topological data analysis technology, constructs a simple complex representation of monitoring data, calculates the time-series variation curve of Betti number and generates a sustained homology significance index, which is used to identify hidden defects and abnormal patterns in dam structures.

[0014] In one embodiment of this application, the construction of a distributed spiking neural network sensing system, which collects dam monitoring data in an event-driven manner and converts continuous signals into sparse pulse sequences to improve spatiotemporal resolution, further includes:

[0015] S11, Deploy a dynamic threshold comparator to trigger pulse event recording when the sensor signal changes by more than 0.05%;

[0016] S12 employs the pulse timing-dependent plasticity STDP learning rule to adaptively train the spiking neural network, thereby enhancing its response sensitivity to abnormal signals.

[0017] In one embodiment of this application, the step of establishing a multi-scale ontology evolution model, which combines cellular automata and neural differential equations to simulate multi-physics coupling processes, further includes:

[0018] S21 decomposes the dam into 10^6 level micro-cells and defines a rule set for each micro-cell containing 128 local evolution rules including material degradation and seepage diffusion;

[0019] S22 employs a neural network constant differential equation solver for iterative calculations to output the evolution trajectory of the multiphysics field and simulate its time-varying behavior.

[0020] In one embodiment of this application, the method of using a gauge field neural network to perform constrained optimization of the evolutionary model, introducing SU(2) symmetric group representation learning and a differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data, further includes:

[0021] S31, construct a neural network layer with a SU(2) symmetric group structure for feature learning under symmetry constraints;

[0022] S32 introduces a symplectic geometric integrator to maintain the differential structure and conservation properties during the field evolution process.

[0023] In one embodiment of this application, the step of constructing a simplex complex representation of monitoring data based on topological data analysis technology, calculating the time-series variation curve of Betti number, and generating a persistent homology significance index for identifying hidden defects and abnormal patterns in the dam structure further includes:

[0024] S41, construct a simple complex representation of the monitoring data using continuous cohomology theory, and calculate the time-series variation curve of its Betti number;

[0025] S42, based on the variation of the Betti number, constructs a persistent cohomology significance index to quantify the topological significance of structural anomaly patterns.

[0026] In one embodiment of this application, it further includes:

[0027] S5 employs a quantum annealing optimizer to map the parameter calibration problem to the Ising model, and combines it with the classical quasi-Newton method for local fine-tuning, thereby improving the global search capability and convergence accuracy of the model parameter optimization.

[0028] To achieve the above objectives, a second aspect of this application proposes a multiphysics dynamic simulation device for dams based on data and mechanism fusion, comprising:

[0029] The distributed spiking neural network sensing module is used to build a distributed spiking neural network sensing system. It collects dam monitoring data in an event-driven manner and converts continuous signals into sparse pulse sequences to improve spatiotemporal resolution.

[0030] The multi-scale ontology evolution model building module is used to build multi-scale ontology evolution models. It adopts a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process.

[0031] The Gaussian field neural network constraint optimization module is used to perform constraint optimization on the evolution model using the Gaussian field neural network. It introduces SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data.

[0032] The topology data analysis and anomaly identification module is used to construct a simple complex representation of monitoring data based on topology data analysis technology, calculate the time-series variation curve of Betti number and generate a continuous homology significance index, which is used to identify hidden defects and anomaly patterns in dam structures.

[0033] To achieve the above objectives, a third aspect of this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0034] The memory stores computer-executed instructions;

[0035] The processor executes computer execution instructions stored in the memory to implement the method as described in any one of the first aspects.

[0036] To achieve the above objectives, a fourth aspect of this application provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, are used to implement the method as described in any one of the first aspects.

[0037] To achieve the above objectives, a fifth aspect of this application provides a computer program product that, when executed by a processor, implements the method described in any one of the first aspects.

[0038] The methods, apparatus, electronic devices, and computer-readable storage media of this application improve the dynamic simulation accuracy of the multi-physics coupling effect and long-term service performance of dams, enhance the ability to identify hidden defects, and improve the reliability of early warning under extreme conditions.

[0039] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0040] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0041] Figure 1 This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0042] Figure 2 This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0043] Figure 3 This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0044] Figure 4 This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0045] Figure 5 This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0046] Figure 6This is a flowchart of a multi-physics dynamic simulation method for dams based on data and mechanism fusion, according to an embodiment of the present invention.

[0047] Figure 7 This is a schematic diagram of the structure of a dam multiphysics dynamic simulation device based on data and mechanism fusion according to an embodiment of the present invention. Detailed Implementation

[0048] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0049] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.

[0051] Example 1

[0052] Figure 1 This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0053] like Figure 1 As shown, the multiphysics dynamic simulation method for dams based on data and mechanism fusion includes the following steps:

[0054] S1. Construct a distributed spiking neural network sensing system to collect dam monitoring data in an event-driven manner and convert continuous signals into sparse pulse sequences to improve spatiotemporal resolution.

[0055] Specifically, this step involves constructing a distributed spiking neural network (SNN) sensing system to collect dam monitoring data in an event-driven manner and convert continuous signals into sparse pulse sequences to improve the system's spatiotemporal resolution and sensing efficiency. In some implementations, the system employs an asynchronous event acquisition mechanism to replace the traditional periodic sampling method, thereby significantly reducing data redundancy and enhancing the response capability to transient changes.

[0056] From a technical implementation perspective, the system deploys a dynamic threshold comparator, which works as follows: when the change in a physical quantity (such as strain, displacement, temperature, etc.) acquired by the sensor exceeds a preset threshold (e.g., 0.05%), event recording is triggered and a pulse signal is generated. This threshold can be adaptively adjusted according to the environmental noise level and the dynamic characteristics of the signal to ensure effective capture of key changes under different operating conditions. Furthermore, the system maps continuous signals into sparse pulse sequences using Inter-Spike Interval Coding (ISIC), where the pulse frequency is proportional to the signal's rate of change, and the pulse interval reflects the signal's local stability. This encoding method significantly reduces the computational load on data transmission and processing while ensuring information integrity.

[0057] At the parameter level, the sampling latency of the event-driven acquisition module is controlled within 10ms, and the sparsity of the pulse code is typically maintained between 1% and 5%, ensuring that redundancy is reduced while maintaining high resolution. The communication protocol adopts MQTT-over-UDP, with a transmission latency of less than 50ms, which is suitable for real-time monitoring scenarios of dam structures.

[0058] In application scenarios, this technology can be widely used in dam crack monitoring, seepage change detection, and seismic response analysis. Through distributed deployment, the system can cover key parts of the dam, achieving synchronous perception of local abrupt changes and global dynamics.

[0059] The technical effect of this step is that, through event-driven mechanisms and pulse coding strategies, it improves the spatiotemporal resolution and energy efficiency of the system, providing high-quality, low-latency perceptual input for subsequent mechanism modeling and anomaly identification. It is a key technical support for realizing digital twins and intelligent early warning of dams.

[0060] S2 establishes a multi-scale ontological evolution model, using a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process.

[0061] Specifically, the step of "establishing a multi-scale ontological evolution model and simulating the multi-physics coupling process of dam material performance degradation and seepage-stress interaction by combining cellular automata and neural differential equations" is the core link in this application to realize the dynamic simulation of dam performance and the prediction of dam failure evolution. This model, by integrating physical mechanisms and data-driven methods, constructs a hybrid computational framework with spatiotemporal evolution capabilities, thereby effectively improving the simulation accuracy and adaptability of complex dam behavior.

[0062] At the technical implementation level, the model employs an architecture combining continuous-time cellular automata (CTCA) and neural differential equations (NeuralODE). The dam structure is discretized into a multi-scale cellular space, where each cell represents a microstructural unit with specific material properties and mechanical behaviors. The number of cells can reach over 10^6 to achieve simultaneous modeling of local damage and global response of the dam. Cellular state variables include stress, strain, pore water pressure, and damage degree, and the evolution rules are defined by local interaction functions in the form of differential equations, such as: Where F is a nonlinear dynamic function modeled by neural differential equations, which can adaptively learn the evolution law of material degradation and seepage-stress coupling.

[0063] At the parameter level, the cell update frequency in the model is set to be adjustable from 0.1 seconds to 10 seconds to match physical processes at different scales. The hidden layer dimension of the neural differential equations is 128–256, and the activation function uses Softplus or Swish to enhance nonlinear expressiveness. During model training, the L-BFGS optimization algorithm is introduced, and the loss function combines physical conservation constraints and monitoring data fitting error terms to ensure that the model has good data fitting ability while satisfying physical laws.

[0064] In application scenarios, this model can be embedded into a dam digital twin system to simulate processes such as concrete aging, crack propagation, and seepage path evolution under long-term service. It is particularly suitable for identifying hidden defects and predicting structural response under extreme conditions, such as earthquakes and sudden drops in reservoir water levels.

[0065] The technical advantage of this step lies in its ability to effectively overcome the limitations of traditional finite element models in modeling complex nonlinear behaviors through adaptive evolution of local rules and emergent mechanisms of global behavior, while avoiding the lack of physical interpretability inherent in purely data-driven models. Its innovation lies in the dynamic evolution mechanism of multi-physics coupled modeling and the continuous-time modeling of material behavior using neural differential equations, providing high-precision and robust technical support for the full life-cycle safety assessment of dams.

[0066] S3 utilizes a gauge field neural network to perform constrained optimization of the evolution model, introducing SU(2) symmetric group representation learning and a differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data.

[0067] Specifically, in this application, the step of "using gauge field neural networks to perform constrained optimization of the evolutionary model, introducing SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data" is a key link in realizing high-precision modeling and dynamic simulation of the dam's operational performance. This step integrates gauge field theory and deep learning architecture to construct a neural network model with physical conservation characteristics, thereby introducing strict symmetry constraints and geometric consistency in the optimization process of the evolutionary model.

[0068] At the technical implementation level, the Gauge Field Neural Network (GFNN) uses the SU(2) symmetry group as the basis for its representation learning. The SU(2) group is a classic Lie group describing the spin symmetry of particles, which is used in this application to model the local symmetry of material properties and stress fields in dam structures. Specifically, GFNN maps the state variables (such as displacement field, stress field, and seepage field) in the evolution model to the representation space of the SU(2) group, ensuring that the transformation of physical quantities satisfies the invariance under group action during the model evolution process. This symmetry constraint helps to suppress non-physical numerical oscillations and improve the stability and generalization ability of the model.

[0069] Furthermore, this step introduces a differential geometry framework, defining the evolutionary model on a manifold space rather than the traditional Euclidean space. By constructing a Riemannian metric tensor, the model can maintain the continuity and smoothness of the physical field in its spatiotemporal evolution. For example, in the cellular automata evolution process, a Neural ODE solver combined with a Symplectic geometric integrator is used to ensure the numerical stability of energy and momentum conservation. A topological conservation term is introduced into the model's objective function, with its weighting coefficient typically set to 0.01–0.1 to balance physical constraints with data fitting accuracy.

[0070] At the parameter level, the input dimension of GFNN is the spatiotemporal tensor of sensor data (e.g., 64×64×T, where T is the time step), and the output is the multiphysics evolution state of the dam structure. The SU(2) group representation used in the model has a dimension of 3×3 unitary matrix, and is obtained through... The transformation projects it into hyperbolic space to enhance the expressive power of hierarchical features. During optimization, a hybrid optimization strategy is used: first, a global coarse search is performed using quantum annealing with a search step size of 0.1–0.5; then, a quasi-Newton method is used for local fine-tuning with a convergence threshold of 1e-5.

[0071] In practical applications, this step is mainly used for predicting the dynamic response and detecting anomalies of dam structures under complex working conditions. For example, in scenarios of sudden drop in reservoir water level or seismic disturbance, GFNN can effectively maintain the continuity of stress and displacement fields, avoiding the boundary discontinuity problem commonly found in traditional physical information neural networks (PINN). At the same time, by introducing SU(2) symmetry constraints, the model can more accurately capture the nonlinear behavior of materials and the evolution of local damage, thereby improving the matching degree with monitoring data and reducing the uncertainty of inversion analysis.

[0072] In summary, this step, by introducing gauge field theory and differential geometry methods, constructs a deep learning model with physical conservation properties, achieving high-precision constraint optimization of the evolutionary model under complex working conditions, and providing a solid mathematical and computational foundation for the performance analysis of dams.

[0073] S4, based on topological data analysis technology, constructs a simple complex representation of monitoring data, calculates the time-series variation curve of Betti number and generates a sustained homology significance index, which is used to identify hidden defects and abnormal patterns in dam structures.

[0074] Specifically, this step utilizes Topological Data Analysis (TDA) technology to perform high-dimensional topological modeling of dam monitoring data. By constructing a simple complex representation, the time-series variation curve of the Betti number is calculated, and a sustained cohomology significance index is generated, thereby identifying hidden defects and anomaly patterns in the dam structure. This method overcomes the limitations of traditional point-based monitoring and linear statistical analysis, achieving dynamic perception and anomaly detection of the structural state from a geometric-topological perspective.

[0075] In some implementations, the multi-source heterogeneous monitoring data of the dam (such as displacement, stress, temperature, and seepage pressure) are first preprocessed, including denoising, normalization, and time alignment. Then, a dynamic time window (DTW) or sliding window mechanism is used to divide the time-series data into several time slices, with the data points within each time slice forming a point cloud. Further, an adjacency graph is constructed using Euclidean distance or dynamic similarity metrics, and then the Vietoris-Rips simplex construction algorithm is used to transform the adjacency graph into a simplex, forming a hierarchical topological space.

[0076] When calculating the time-series variation curve of the Betti number, the persistent homology method is used. By gradually increasing the adjacency radius ε, the appearance and disappearance processes of topological features (such as connected components, loops, and voids) at different dimensions (0-dimensional, 1-dimensional, and 2-dimensional) are tracked. The time-series variation of the Betti number reflects the topological evolution characteristics of the structure at different scales. For example, an abnormal increase in the 0-dimensional Betti number may indicate local cracks or abrupt displacement, while the continuous increase in the 1-dimensional Betti number may reflect ring-shaped anomalies inside the structure (such as the formation of seepage channels).

[0077] Optionally, a Persistent Homology Significance Index (PHSI) can be constructed. By comparing the current Betti number curve with the historical baseline curve and combining statistical significance tests (such as p-value analysis or confidence interval assessment), the confidence level of structural anomalies can be quantified. The calculation of PHSI can incorporate a dynamic threshold mechanism, such as setting the Betti number change rate threshold to 0.15 / time step, or setting the significance p-value threshold to 0.05, to distinguish between normal fluctuations and potential diseases.

[0078] This step, as a core module of the decision output layer in this application, works in conjunction with the ontology evolution model library and the environment-driven analysis engine to provide topological criteria for assessing the health status of dam structures. Its technical value lies in the fact that it can extract high-dimensional topological features from data without pre-setting a structural damage model, enabling early identification of hidden defects and global perception of abnormal patterns, significantly improving the sensitivity and robustness of dam safety monitoring.

[0079] The multi-physics dynamic simulation method for dams based on data and mechanism fusion in this application embodiment effectively integrates multi-source monitoring data and physical mechanism models, improves the accuracy of multi-physics dynamic simulation of dams and the ability to identify defects, and enhances the level of safety situation awareness and early warning throughout the entire life cycle.

[0080] Example 2

[0081] Figure 2 This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0082] like Figure 2 As shown, S1 further includes:

[0083] S11, Deploy a dynamic threshold comparator to trigger pulse event recording when the sensor signal change exceeds 0.05%.

[0084] Specifically, the step of "deploying a dynamic threshold comparator to trigger pulse event recording when the sensor signal changes by more than 0.05%" in this application is a key link in building an event-driven data acquisition system. Its technical implementation is based on the asynchronous processing mechanism of Spiking Neural Network (SNN), which aims to improve the real-time performance and data efficiency of the dam monitoring system.

[0085] At the technical implementation level, the dynamic threshold comparator employs a combination of differential detection and adaptive filtering to monitor continuous signals from a distributed sensor network in real time. Specifically, the system first normalizes the raw sensor signals to eliminate dimensional differences and environmental drift. Then, the comparator continuously calculates the relative rate of change between the current signal value and the signal value at the previous moment, ΔS / S0, where ΔS is the signal change and S0 is the reference signal value. When this rate of change exceeds a set threshold of 0.05%, the system determines it as a valid event and triggers a pulse event recording mechanism. This mechanism encapsulates the event's timestamp, sensor ID, and signal change into a pulse signal, which is then converted into a sparse event stream using Inter-Spike Interval (ISI) coding to reduce data redundancy and improve transmission efficiency.

[0086] At the parameter level, the 0.05% threshold is set based on the typical sensitivity range of the dam structure response. Combined with statistical analysis of long-term monitoring data and identification of critical changes in structural health, this ensures that key anomaly signals are captured while avoiding false triggers caused by environmental noise or equipment drift. Furthermore, the system supports dynamic threshold adjustment. It can adaptively optimize the threshold using online learning algorithms (such as the sliding window averaging method) based on external factors such as seasonal temperature and humidity changes and water level fluctuations, maintaining a high signal-to-noise ratio and a low false negative rate.

[0087] At the application level, this step is widely used in real-time monitoring systems for key parameters such as dam displacement, stress, and seepage pressure. Especially under extreme conditions (such as rainstorms and earthquakes) or in the early stages of hidden defects, the structural response may exhibit subtle but critical changes that traditional periodic sampling struggles to capture in a timely manner. This application, however, utilizes an event-driven mechanism to achieve immediate response and recording of abnormal signals, providing highly timely data support for subsequent mechanistic model updates and early warning decisions.

[0088] In terms of technical effectiveness, this step effectively improves the spatiotemporal resolution and data utilization of the monitoring system, reduces the pressure of invalid data transmission and storage, and enhances the sensitivity to minor structural anomalies. Combined with subsequent spiking neural network processing and hyperbolic space feature learning, it lays a solid foundation for building a high-precision, low-latency digital twin system for dams, demonstrating significant engineering practical value and innovative significance.

[0089] S12 uses the STDP (Pulse Timing Dependent Plasticity) learning rule to adaptively train the spiking neural network in order to enhance its response sensitivity to abnormal signals.

[0090] Specifically, in this application, the use of the STDP (Spike-Timing-Dependent Plasticity) learning rule for adaptive training of the Spiking Neural Network (SNN) is a key technical step in improving the system's sensitivity to abnormal signals. STDP is a learning mechanism based on the synaptic plasticity of biological neurons. Its core principle is that the adjustment of synaptic weights depends on the time difference between presynaptic and postsynaptic pulses; that is, when the current synaptic neuron fires a pulse before the postsynaptic neuron, the synaptic weight is enhanced, and vice versa. This mechanism can effectively capture the temporal dynamic characteristics of signals, and is particularly suitable for processing asynchronous, sparse pulse event streams.

[0091] In its implementation, this invention constructs an event-driven SNN architecture, where each neuron receives pulse input from distributed sensors and dynamically adjusts connection weights using the STDP rule. In some implementations, the learning window function of STDP adopts a double exponential form, defined as: ΔW = A + ·exp(-Δt / τ + )-A - ·exp(Δt / τ - ), where Δt is the time difference between the preceding and following pulses, A + =0.01, A - =0.005 is the gain coefficient, τ + =20ms, τ - =30ms is the time constant. This parameter setting can enhance the ability to identify short-sequence abnormal signals while suppressing background noise interference.

[0092] Furthermore, this step is used in the dam monitoring system to process asynchronously acquired structural response data, such as strain, displacement, and seepage pressure, in real time. Through the STDP mechanism, the SNN can adaptively enhance the response path to anomalous signals (such as abrupt changes and high-frequency oscillations), thereby improving the system's early identification capability for potential structural damage. This technology integrates the biological rationality and adaptive learning capability of spiking neural networks, providing high-quality dynamic feature inputs for subsequent hyperbolic space feature extraction and topological data analysis, significantly enhancing the real-time performance and robustness of the entire system.

[0093] Example 3

[0094] Figure 3This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0095] like Figure 3 As shown, S2 further includes:

[0096] S21 decomposes the dam into 10^6 level micro-cells and defines a rule set for each micro-cell containing 128 types of local evolution rules, including material degradation and seepage diffusion.

[0097] Specifically, this step involves discretizing the dam structure into micro-cells on the order of 10^6, and defining a rule set for each micro-cell containing 128 types of local evolution rules such as material degradation and seepage diffusion. This is the core implementation step of the "ontology evolution model library" module in this application.

[0098] At the technical implementation level, this step employs a multi-scale cellular automata (MCA) modeling method to divide the dam structure into hexahedral or tetrahedral meshes in spatial dimensions, forming approximately one million micro-cells. Each micro-cell represents the material properties and physical state of a local region of the dam, and its evolution is driven by a set of rules based on physical mechanisms and data. The rule set covers 128 types of local evolution mechanisms, including material performance degradation (such as decrease in elastic modulus and compressive strength), seepage diffusion (such as changes in pore water pressure and evolution of permeability coefficient), stress-strain response, crack propagation, and temperature field changes. Each rule is modeled based on constitutive equations, damage mechanics models, or empirical formulas, and is parameterized using neural differential equations (Neural ODEs) to achieve dynamic updates and adaptive evolution of the rules.

[0099] At the parameter level, the micro-cell partitioning accuracy must meet Δx ≤ 0.5m to ensure high-resolution modeling of key structural components (such as dam foundations and seepage barriers). Each rule in the evolution rule set has thresholds for state variables (such as stress, strain, pore water pressure, etc.) and their rates of change. For example, the material degradation rule sets the rate of decrease in elastic modulus to a threshold value. Updates are triggered on demand. The update frequency of the rule set is controlled by an event-driven mechanism, with a response time not exceeding 10 seconds. -3 s, to adapt to the needs of real-time simulation.

[0100] At the application level, this step is applicable to digital twin simulation of the entire life cycle of dams, playing a crucial role in long-term service performance evaluation, identification of hidden defects, and simulation of extreme operating conditions. Through the dynamic evolution of local rules, the system can simulate the response process of dams under different environmental loads (such as water level changes, earthquakes, and temperature fluctuations), achieving high-precision prediction of structural behavior.

[0101] In terms of technical effectiveness, this step, by introducing local evolution rules coupled with multi-physics fields, effectively replaces the complex constitutive models and solvers in the traditional finite element method, significantly improving computational efficiency and model adaptability. Simultaneously, the diversity and dynamism of the rule set enhance the ability to characterize nonlinear and time-varying structural behavior, providing high-fidelity evolutionary trajectory data for subsequent topological data analysis and early warning decision-making, thereby improving the accuracy and robustness of dam safety assessment.

[0102] S22 employs a neural network constant differential equation solver for iterative calculations to output the evolution trajectory of the multiphysics field and simulate its time-varying behavior.

[0103] Specifically, in the ontological evolution layer of this application, a Neural Ordinary Differential Equations (NODE) ​​solver is used for iterative calculations to output the evolution trajectory of multiphysics fields and simulate their time-varying behavior. This step is the core link in realizing the dynamic simulation and performance prediction of the entire life cycle of dam structures, integrating the advantages of physical mechanisms and data-driven modeling.

[0104] At the technical implementation level, the NODE solver constructs a continuous-time dynamic model by combining neural networks with ordinary differential equations (ODEs). Specifically, the state variable x(t) of each cell evolves over time, and its rate of change is determined by a neural network function. Definition, that is This function not only incorporates local physical evolution rules (such as stress-strain relationships and seepage diffusion coefficients), but also adaptively learns the nonlinear response characteristics of the dam under complex working conditions through a neural network. During the iteration process, the Runge-Kutta method with an adaptive step size (such as the Dormand-Prince algorithm) is used for numerical integration to ensure a balance between computational accuracy and efficiency.

[0105] At the parameter level, the NODE model has an input dimension of d = 128, corresponding to the 128 local rules defined in the dam ontological evolution model library. The time step Δt is set to 10^{-3} seconds, and the integral error tolerance ε is controlled within 10^{-6} seconds to meet the high-precision requirements of engineering simulation for time-varying behavior. The hidden layer of the neural network adopts an LSTM structure with 256 units and the activation function is Tanh to enhance the modeling ability of historical dependencies.

[0106] At the application level, this step is used to simulate the dynamic response of a dam under the coupled effects of multiple physical fields such as water load, temperature change, and seismic disturbance. Through the collaborative calculation of continuous-time cellular automata and the NODE solver, high-fidelity evolution simulation of the internal stress field, seepage field, and temperature field of the dam can be achieved, which is particularly suitable for the identification of hidden defects and the prediction of structural behavior under extreme conditions.

[0107] In terms of technical effectiveness, this method effectively overcomes the numerical rigidity problem caused by the discrete time step in traditional finite element calculations. At the same time, through the nonlinear fitting capability of neural networks, it improves the model's adaptability and generalization ability to complex physical processes, providing a more accurate and robust dynamic simulation method for dam safety assessment.

[0108] Example 4

[0109] Figure 4 This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0110] like Figure 4 As shown, S3 further includes:

[0111] S31, construct a neural network layer with a SU(2) symmetric group structure for feature learning under symmetry constraints.

[0112] Specifically, this application constructs a neural network layer with a SU(2) symmetric group structure, which is a core component of the environment-driven analysis engine. This layer aims to achieve feature learning under symmetry constraints, thereby improving the physical consistency and generalization ability of the dam's operational behavior analysis. This step, based on gauge field theory and differential geometry, embeds physical symmetry into the neural network structure, ensuring the conservation of the dam's structural and material behavior during feature extraction.

[0113] At the technical implementation level, this neural network layer adopts the SU(2) Lie group as the symmetry structure. By constructing the feature transformation function under the group action, the network possesses rotation invariance and local symmetry preservation capabilities. Specifically, the input feature tensor is mapped to a vector field under the action of SU(2) group elements in the local coordinate system. The network weight parameters are designed as group representation matrices, and the group structure constraint of the parameters is achieved through Lie algebra exponential mapping. The activation function of each layer must satisfy covariance under the group action, and typically adopts... The group-invariant form of a nonlinear activation function or a hyperbolic tangent function.

[0114] In terms of parameters, the input dimension of this layer is usually a three-dimensional vector field (such as displacement, stress, strain), and the output dimension is a four-dimensional spinor field to support the two-valued representation of the SU(2) group. The network parameters include the group representation matrix (3×3 or 4×4), whose updates need to satisfy the orthogonality / unitary constraints of SO(3) or SU(2), usually achieved through orthogonal regularization terms (such as Frobenius norm constraints) or parameterization methods (such as Rodrigues formulas). During training, the loss function introduces a symmetry bias term to quantify the inconsistency between feature transformation and group action. Its weight coefficient is usually set to 0.01 to 0.1 to balance data fitting and physical constraints.

[0115] In application scenarios, this layer is used to process multi-physics monitoring data of dam structures (such as temperature field, seepage field, and stress field), extracting feature representations with physical conservation properties in non-Euclidean geometric space. By maintaining SU(2) symmetry, feature distortion caused by sensor noise or local disturbances can be effectively suppressed, improving the robustness and interpretability of the model under complex working conditions.

[0116] In terms of technical effectiveness, this step achieves a deep integration of physical symmetry and deep learning, ensuring the consistency of the feature space and the dam ontology evolution model in terms of mathematical structure, thereby significantly improving the accuracy and stability of the model in long-term service status prediction and disease identification.

[0117] S32 introduces a symplectic geometric integrator to maintain the differential structure and conservation properties during the field evolution process.

[0118] Specifically, in the ontological evolution layer of this application, the introduction of a symplectic geometric integrator is a key technical step to achieve the preservation of differential structure and conservation characteristics during the evolution of field quantities. This step, based on differential geometry and dynamical systems theory, aims to perform structure-preserving numerical integration on the evolution process of multiple physical fields of the dam (such as stress field, seepage field, temperature field, etc.), thereby improving the physical consistency and long-term stability of the model.

[0119] In some implementations, symplectic geometric integrators employ explicit or implicit symplectic schemes, such as the Verlet algorithm, Runge-Kutta type symplectic integrators (e.g., the RKMK method), or the Lie-Trotter splitting method, to ensure that the symplectic structure of the system is preserved during discretization. Specifically, the fields in the dam evolution model are modeled as generalized coordinate and momentum variables in a Hamiltonian system, and its evolution equations can be expressed as the Hamiltonian equations: Where H is the Hamiltonian of the system. By using a symplectic integrator to discretize and solve the above equations, the physical properties such as energy conservation, momentum conservation, and constant phase space volume can be effectively preserved.

[0120] In terms of parameter specifications, the time step Δt of the symplectic integrator is typically set to 0.01–0.1 seconds to balance computational efficiency and numerical stability. Within the framework of continuous-time cellular automata, the evolution rule of each microcell is expressed in differential form. Where F is the nonlinear field function jointly defined by the mechanistic model and data-driven approach. In iterative calculations, the symplectic integrator uses a structure-preserving numerical method to ensure that the field quantities do not exhibit numerical divergence or energy drift during long-term evolution.

[0121] At the application level, this step is suitable for dynamic simulation of dams under complex operating conditions, such as extreme events like sudden drops in reservoir water levels, seismic excitation, and sudden temperature changes. In a digital twin system, the symplectic integrator can be embedded with a multi-scale evolution model to achieve continuous and conserved evolution simulation from microscopic material degradation to macroscopic structural response.

[0122] In terms of technical effectiveness, this step significantly improves the physical fidelity and long-term predictive ability of the dam simulation model, effectively avoids the numerical instability problem that occurs in the nonlinear evolution of the traditional finite element method, and provides a solid mathematical and physical foundation for dam safety situation perception and early warning.

[0123] Example 5

[0124] Figure 5 This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0125] like Figure 5 As shown, S4 further includes:

[0126] S41, using continuous cohomology theory, constructs a simple complex representation of the monitoring data and calculates the time-series variation curve of its Betti number.

[0127] Specifically, this step constructs a simple complex representation of the monitoring data using persistent homology theory and calculates the time-series variation curve of its Betti number. This is a key step in achieving structural anomaly identification and early warning in the decision output layer of this application. Its technical implementation principle is based on topological data analysis (TDA) methods in algebraic topology, aiming to extract physically meaningful topological features from high-dimensional, nonlinear, and non-Euclidean monitoring data, thereby revealing the morphological evolution and potential defects of the dam structure at different service stages.

[0128] In some implementations, this step first synchronizes the asynchronous pulse signals from the distributed sensing network to form a time-series point cloud dataset. Then, using the Vietoris-Rips simplicoid construction method, with the dynamic neighborhood radius ε(t) as a parameter, the point cloud is progressively connected into simplicoids of different dimensions (0-simplicoids are points, 1-simplicoids are edges, 2-simplicoids are faces, etc.), forming a time-evolving sequence of simplicoids. This process can be represented as follows: for each time step t, based on the spatial distance matrix of the monitoring points, a VR complex K(t) is constructed in the neighborhood of ε(t), where ε(t) can be adjusted according to the density distribution of the monitoring data or a preset dynamic threshold strategy, for example, using an adaptive K-nearest neighbor algorithm (k = 5~10) to determine ε(t).

[0129] Furthermore, by calculating the homology groups H0(t), H1(t), and H2(t) for each K(t), the Betti numbers β0(t), β1(t), and β2(t) are extracted, corresponding to the number of connected components, the number of cavities, and the number of voids, respectively. In practical applications, β0(t) is used to monitor changes in the overall connectivity of the structure, β1(t) is used to identify the formation and expansion of cracks or seepage channels, and β2(t) is used to detect the occurrence of three-dimensional cavities or local instability regions. During the calculation process, open-source topology libraries such as GUDHI or DIPHA are used, combined with GPU acceleration to ensure computation on large-scale data (such as 10^64 GHz). 6 Real-time performance and stability at the micro-cellular level.

[0130] Regarding specific parameter settings, the time step Δt is typically set to 1 hour to 1 day, depending on the monitoring frequency and data volume; the initial value of ε(t) can be set to 1.5 to 2 times the average spacing between monitoring points, and then dynamically adjusted according to the structural response. Furthermore, to enhance the robustness of the Betti number, a persistence diagram and a persistent barcode can be introduced for feature enhancement, and a significance index based on Lipschitz continuity can be constructed for anomaly detection.

[0131] This step is of great value in the whole life cycle health monitoring of dams, especially in the identification of hidden defects, early warning of crack propagation and detection of abnormal seepage. It can provide global topological features that are difficult to capture by traditional statistical methods, thereby improving the accuracy and foresight of structural safety assessment.

[0132] S42, based on the variation of the Betti number, constructs a persistent cohomology significance index to quantify the topological significance of structural anomaly patterns.

[0133] Specifically, the step of “constructing a persistent cohomology significance index based on the change of Betti number to quantify the topological significance of structural anomaly patterns” is one of the core algorithms in the topological data analysis module of this application. It aims to extract the topological feature changes of the dam structure at different scales through algebraic topology methods, thereby identifying potential structural anomaly patterns.

[0134] At the technical implementation level, this step first constructs a point cloud dataset from the multi-source monitoring data of the dam (such as displacement, stress, and seepage), and then builds its simplex complex using the persistent homology method. Specifically, a Vietoris-Rips complex construction strategy is adopted, gradually expanding adjacency relationships with a dynamic threshold ε to form topological structures at different scales. At each scale, the 0th, 1st, and 2nd order Betti numbers (β0, β1, β2) are calculated, corresponding to the number of connected components, annulus, and cavity, respectively, reflecting the global connectivity, local porosity characteristics, and three-dimensional cavity evolution of the structure. Furthermore, by tracking the time-series curve of the Betti number changing with ε, a topological barcode or persistence diagram is constructed to quantify the persistence and saliency of structural features.

[0135] At the parameter level, in this step, the initial value of ε is set to 0.5 times the standard deviation of the sensor data, the maximum expansion threshold is 5 times the standard deviation, and the step size is 0.1 times the standard deviation. The significance threshold of the Betti number is determined using a statistical significance test method (such as Bootstrap resampling), with features having a p-value less than 0.05 considered significant anomalies. Furthermore, persistence length is introduced as a significance indicator, defined as the lifetime (birth-death difference) of points in the persistence plot, and its threshold is dynamically adjusted in conjunction with the historical baseline of structural health.

[0136] At the application level, this step can be embedded in the decision output layer of a dam health monitoring system to analyze topological anomalies of the structure in real time under different operating conditions (such as water level changes, temperature fluctuations, and seismic response). For example, in the early stages of seepage anomalies or crack propagation, the abnormal increase in the number of local annular cavities β1 can serve as an early warning signal to help determine the evolution trend of structural damage.

[0137] The technical advantage of this step is that by dynamically changing the topological invariant Betti number, it can effectively capture the abnormal topological features of the structure during nonlinear evolution, overcome the sensitivity of traditional Euclidean geometry-based feature extraction methods to noise and local disturbances, improve the robustness and sensitivity of dam structure anomaly identification, and provide key support for achieving high-precision, low-false-alarm-rate health assessment.

[0138] The multi-physics dynamic simulation method for dams based on data and mechanism fusion in this application effectively integrates multi-source monitoring data with physical mechanism models, improving the accuracy of multi-physics dynamic simulation of dams and the ability to identify defects, thereby enhancing the level of safety situation awareness and early warning throughout the entire life cycle.

[0139] Example 6

[0140] Figure 6 This is a flowchart of a multiphysics dynamic simulation method for dams based on data and mechanism fusion, according to one embodiment of this application.

[0141] Steps S1-S4 have been described in the foregoing embodiments and will not be described in detail here.

[0142] S5 employs a quantum annealing optimizer to map the parameter calibration problem to the Ising model, and combines it with the classical quasi-Newton method for local fine-tuning, thereby improving the global search capability and convergence accuracy of the model parameter optimization.

[0143] Specifically, in this application, the quantum annealing optimizer is used to map the parameter calibration problem to the Ising model, and combined with the classical quasi-Newton method for local fine-tuning. This is a key technical means to achieve efficient and high-precision parameter optimization of the dam performance analysis model. In some implementations, this step first transforms the parameter calibration problem into a combinatorial optimization problem by constructing the Hamiltonian H = ∑ i＜j J ij s i s j +∑ i h i s i Mapping the objective function to the spin variable s of the Ising model i Above, where J ij h represents the spin coupling coefficient. i This is the bias term. The mapping process must satisfy the equivalence between the energy function and the objective function. It is usually transformed using a quadratic unconstrained binary optimization (QUBO) form and solved using quantum annealing hardware platforms such as D-Wave or their simulators.

[0144] Furthermore, after quantum annealing, the system obtains a set of approximately optimal parameter solutions. To improve convergence accuracy and stability, this application introduces a classical quasi-Newton method (such as BFGS or L-BFGS) to locally optimize the quantum annealing results. The quasi-Newton method iteratively updates the approximation of the Hessian matrix, using gradient information to guide the parameters towards a better convergence direction, and its convergence speed is usually better than the traditional gradient descent method. In this scheme, the initial point of the quasi-Newton method is taken from the optimal solution of quantum annealing, the iteration step size is set to α∈[0.01,0.1], and the gradient tolerance is set to ε. gThis is to ensure high-precision convergence within a reasonable computation time.

[0145] This hybrid optimization strategy offers significant advantages in parameter calibration for dam body evolution models. Because dam structures involve multi-physics coupling and the parameter space is highly non-convex, traditional optimization methods are prone to getting trapped in local optima. Quantum annealing achieves global search through quantum tunneling, while quasi-Newtonian laws provide precise gradient correction in local regions, thus significantly improving the convergence accuracy and robustness of parameter optimization while maintaining global exploration capabilities. This approach is suitable for the joint calibration of key parameters such as dam material parameters, boundary conditions, and load factors, and exhibits superior performance, especially in handling nonlinear, multi-constraint, and high-dimensional optimization problems.

[0146] The multiphysics dynamic simulation method for dams based on data and mechanism fusion in this application transforms the parameter calibration problem into an Ising model by introducing a quantum annealing optimizer and combining it with the classical quasi-Newton method for local optimization. This significantly improves the global search capability and convergence accuracy during the model parameter optimization process, thereby further enhancing the reliability of multiphysics simulation analysis of dams and the accuracy of dam defect identification.

[0147] To achieve the above embodiments, this application also proposes a multi-physics dynamic simulation device for dams based on the fusion of data and mechanisms. Figure 7 This is a schematic diagram of a multiphysics dynamic simulation device for dams based on data and mechanism fusion, provided as an embodiment of this application. Figure 7 As shown, the device includes:

[0148] Distributed spiking neural network sensing construction module 10 is used to build a distributed spiking neural network sensing system, which collects dam monitoring data in an event-driven manner and converts continuous signals into sparse pulse sequences to improve spatiotemporal resolution.

[0149] The multi-scale ontology evolution model establishment module 20 is used to establish a multi-scale ontology evolution model, which adopts a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process.

[0150] The gauge field neural network constraint optimization module 30 is used to perform constraint optimization on the evolution model using the gauge field neural network, and introduces SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data;

[0151] The topology data analysis and anomaly identification module 40 is used to construct a simple complex representation of monitoring data based on topology data analysis technology, calculate the time series variation curve of Betti number and generate a continuous homology significance index, which is used to identify hidden defects and anomaly patterns in the dam structure.

[0152] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0153] To implement the above embodiments, this application also proposes an electronic device, including: a processor and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0154] To implement the above embodiments, this application also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided in the foregoing embodiments.

[0155] To implement the above embodiments, this application also proposes a computer program product, including a computer program that, when executed by a processor, implements the methods provided in the foregoing embodiments.

[0156] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0157] It should be noted that personal information collected from users should be used for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. Furthermore, such collection / sharing should only be conducted after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign an agreement / authorization that includes authorization of relevant user information before the user uses the function. In addition, any necessary steps must be taken to protect and safeguard access to such personal information data and ensure that others with access to personal information data comply with their privacy policies and procedures.

[0158] This application is intended to provide an implementation scheme for users to selectively prevent the use or access to their personal information data. Specifically, this disclosure is intended to provide hardware and / or software to prevent or block access to such personal information data. Once personal information data is no longer needed, risks can be minimized by restricting data collection and deleting data. Furthermore, where applicable, such personal information is de-identified to protect user privacy.

[0159] In the foregoing descriptions of the embodiments, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0160] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0161] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0162] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0163] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0164] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0165] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0166] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

[0167] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

[0168] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A multi-physics dynamic simulation method for dams based on data and mechanism fusion, characterized in that, include: S1. Construct a distributed spiking neural network sensing system to collect dam monitoring data in an event-driven manner and convert continuous signals into sparse pulse sequences to improve spatiotemporal resolution. S2, establish a multi-scale ontology evolution model, and use a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process; S3, using a gauge field neural network to constrain and optimize the evolution model, introduces SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data; S4, based on topological data analysis technology, constructs a simple complex representation of monitoring data, calculates the time-series variation curve of Betti number and generates a sustained homology significance index, which is used to identify hidden defects and abnormal patterns in dam structures.

2. The method as described in claim 1, characterized in that, The construction of the distributed spiking neural network sensing system, which collects dam monitoring data in an event-driven manner and converts continuous signals into sparse pulse sequences to improve spatiotemporal resolution, also includes: S11, Deploy a dynamic threshold comparator to trigger pulse event recording when the sensor signal changes by more than 0.05%; S12 employs the pulse timing-dependent plasticity STDP learning rule to adaptively train the spiking neural network, thereby enhancing its response sensitivity to abnormal signals.

3. The method as described in claim 1, characterized in that, The establishment of a multi-scale ontology evolution model, which combines cellular automata and neural differential equations to simulate multi-physics coupling processes, also includes: S21 decomposes the dam into 10^6 level micro-cells and defines a rule set for each micro-cell containing 128 local evolution rules including material degradation and seepage diffusion; S22 employs a neural network constant differential equation solver for iterative calculations to output the evolution trajectory of the multiphysics field and simulate its time-varying behavior.

4. The method as described in claim 1, characterized in that, The method of using a gauge field neural network to constrain and optimize the evolutionary model, introducing SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data, also includes: S31, construct a neural network layer with a SU(2) symmetric group structure for feature learning under symmetry constraints; S32 introduces a symplectic geometric integrator to maintain the differential structure and conservation properties during the field evolution process.

5. The method as described in claim 1, characterized in that, The method based on topological data analysis, which constructs a simple complex representation of monitoring data, calculates the time-series variation curve of Betti number, and generates a sustained homology significance index for identifying hidden defects and abnormal patterns in dam structures, also includes: S41, construct a simple complex representation of the monitoring data using continuous cohomology theory, and calculate the time-series variation curve of its Betti number; S42, based on the variation of the Betti number, constructs a persistent cohomology significance index to quantify the topological significance of structural anomaly patterns.

6. The method as described in claim 1, characterized in that, Also includes: S5 employs a quantum annealing optimizer to map the parameter calibration problem to the Ising model, and combines it with the classical quasi-Newton method for local fine-tuning, thereby improving the global search capability and convergence accuracy of the model parameter optimization.

7. A multiphysics dynamic simulation device for dams based on data and mechanism fusion, characterized in that, include: The distributed spiking neural network sensing module is used to build a distributed spiking neural network sensing system. It collects dam monitoring data in an event-driven manner and converts continuous signals into sparse pulse sequences to improve spatiotemporal resolution. The multi-scale ontology evolution model building module is used to build multi-scale ontology evolution models. It adopts a combination of cellular automata and neural differential equations to simulate the multi-physics coupling process. The Gaussian field neural network constraint optimization module is used to perform constraint optimization on the evolution model using the Gaussian field neural network. It introduces SU(2) symmetric group representation learning and differential geometry framework to maintain the continuity of physical quantities and improve the matching degree between the model and monitoring data. The topology data analysis and anomaly identification module is used to construct a simple complex representation of monitoring data based on topology data analysis technology, calculate the time-series variation curve of Betti number and generate a continuous homology significance index, which is used to identify hidden defects and anomaly patterns in dam structures.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-6.

Images