A Monitoring and Evaluation System and Method for Conjugate Soil Slope Restoration Based on Multi-Source Sensing

Abstract

This invention discloses a monitoring and evaluation system and method for conjugate soil slope restoration based on multi-source sensing, belonging to the field of civil engineering monitoring and safety assessment technology. The system includes a heterogeneous sensing layer, an edge fusion layer, a cloud evolution layer, and an autonomous decision-making layer. The method includes: acquiring multi-dimensional state information of the conjugate system through a biomimetic multi-source sensor network; constructing a four-dimensional time-series tensor model and using high-order singular value decomposition with confidence constraints for data fusion and restoration; dynamically evaluating the system's health based on a physical information element learning network; and predicting failure paths and generating graded early warnings using a micro-damage propagation graph network. This invention treats the restoration body and the original soil as a conjugate system, realizing a shift from phenomenon monitoring to mechanism inversion and from static evaluation to dynamic prediction, solving the problems of isolated data, static evaluation, and delayed early warning in traditional monitoring.

Description

Technical Field

[0001] This invention relates to the field of civil engineering monitoring and safety assessment technology, and in particular to a monitoring and assessment system and method for conjugate soil slope restoration based on multi-source sensing. Background Technology

[0002] Currently, traditional slope restoration monitoring faces three major bottlenecks: isolated data, static evaluation, and delayed early warning. Existing technologies typically treat the restoration and the original soil as independent monitoring objects, limiting monitoring to measuring single-point physical quantities, which fails to reflect the complex mechanical interactions between the restoration and the soil. Furthermore, multi-source data suffers from heterogeneity, asynchrony, and inconsistent precision, leading to difficulties in data alignment and completion. In terms of evaluation algorithms, traditional machine learning methods rely on large amounts of historical fault data and lack physical mechanism constraints, making accurate, rapid, adaptive evaluation of small samples and failure path prediction difficult. Therefore, there is an urgent need for an intelligent monitoring and evaluation system capable of comprehensively perceiving the state of the conjugate system, deeply integrating multi-source data, and performing dynamic prediction. Summary of the Invention

[0003] This invention aims to solve the above-mentioned problems by providing a monitoring and evaluation system and method for conjugate soil slope restoration based on multi-source sensing. This system establishes a unified sensing theory of "conjugate response field" and deeply integrates "temporal tensor fusion," "physical information element learning," and "micro-damage map propagation prediction" technologies to achieve a transformation from phenomenon monitoring to mechanism inversion, from static evaluation to dynamic prediction, and from isolated alarms to system-wide decision-making.

[0004] To achieve the above objectives, the present invention provides a monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing, comprising: The heterogeneous sensing layer is used to deploy a multi-source sensor network in a biomimetic manner to acquire multi-dimensional state information of the conjugate system. The conjugate system refers to the system in which the repair body and the original soil body reinforced and constrained by it are regarded as a mechanically interacting and cooperatively responding system. The core monitoring target is upgraded to sensing and analyzing the dynamic evolution of the "interaction stress field" and "cooperative deformation field" inside the system.

[0005] The edge fusion layer is used for runtime tensor fusion algorithms to generate aligned, complete, and highly reliable system state tensors; The cloud evolution layer includes a physical information element learning network and a micro-damage graph propagation model, which are used for state evaluation and trend prediction based on the system state tensor. The autonomous decision-making layer is used to generate visual diagnostic reports and tiered early warnings based on the assessment and prediction results.

[0006] Preferably, in the heterogeneous sensing layer, the multi-source sensor network adopts a "backbone-root" non-uniform encrypted network topology; The backbone is a distributed optical fiber sensor network pre-embedded in the core of the repair body, used to continuously sense strain and temperature field. The fibrous roots are a cluster of sensors wirelessly deployed in the critical interaction region of the conjugate system. The sensor cluster includes a MEMS inclinometer, an earth pressure gauge, and a moisture content sensor.

[0007] Preferably, in the cloud evolution layer, the micro-damage graph propagation model discretizes the conjugate system into a dynamic graph. ,in For a set of nodes, Let be the set of edges. Let it be the set of edge weights; The micro-damage graph propagation model uses a spatiotemporal graph convolutional network to process graph sequences, utilizes Chebyshev graph convolution to capture damage propagation between nodes, uses one-dimensional temporal convolution to capture the state evolution of nodes themselves, and outputs the damage probability and critical path saliency score of each node at future time steps.

[0008] Preferably, a four-dimensional temporal tensor is constructed in the edge fusion layer. As a unified mathematical expression for the state of a conjugate system; in, In terms of spatial dimension, the three-dimensional spatial location is mapped to a one-dimensional index through spatial encoding, representing the spatial distribution of the sensor; The modal dimension represents different types of sensing physical quantities; The time dimension represents a continuous, equally spaced time series; The confidence dimension represents the confidence level of each data point.

[0009] Preferably, in the cloud evolution layer, the physical information meta-learning network includes a physical information encoder and a meta-learning evaluator; The physical information encoder takes the joint feature vector and historical state as input, and forces the network to learn feature representations that conform to the laws of mechanics through physical residual loss constraints. The physical residual loss is the deviation between the stress and strain estimates output by the network and the constitutive relation function of the conjugate system. The meta-learning evaluator adopts a model-independent meta-learning framework, which divides the monitoring data stream into a series of continuous time-segment tasks. It achieves rapid adaptation of small samples through inner and outer layer updates, and outputs the system health index and risk evolution curve.

[0010] This invention also provides a monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing, comprising the following steps: S1. Construct a biomimetic multi-source sensor network, adaptively deploy and schedule sensor nodes, and obtain multi-dimensional state information of the conjugate system. S2. Construct a four-dimensional temporal tensor model as a unified mathematical expression of the state of the conjugate system, and perform multi-source data fusion and generation repair. S3. Construct a meta-learning dynamic evaluation model embedded with physical information, analyze the state tensor of the repaired system, and output the system health index and risk evolution curve. S4. Construct a micro-damage propagation network, predict failure paths, and generate a visual diagnostic report and graded early warning based on the assessment and prediction results.

[0011] Preferably, in step S1, the system dynamically adjusts the sensing strategy based on the real-time system status, defining each sensing node. At any moment Perception priority And schedule according to priority; Perception Priority The calculation formula is: ; in, The prior geological weakness coefficient for the location of nodes. This represents the real-time risk assessment value for a local area surrounding the node. This is a function to indicate abnormal node data. Continue working for the node The normalized value of energy consumption cost over time. This is the abnormal excitation gain coefficient.

[0012] Preferably, in step S2, the multi-source data fusion and generation repair includes: performing mode-specific normalization on each mode, and using confidence-constrained high-order singular value decomposition for tensor completion to generate the optimal complete system state tensor. The objective function of the completion process is: ; in, For the complete tensor to be found, To be related to confidence level The relevant weight tensor, It represents the Hadamah accumulation. It is the Frobenius norm. For tensor rank regularization, Its coefficient.

[0013] Preferably, in step S4, the micro-damage propagation graph network adopts a spatiotemporal graph convolutional network, which uses Chebyshev graph convolution to capture damage propagation between nodes and uses one-dimensional temporal convolution to capture the state evolution of the nodes themselves. Micro-damage propagation graph networks discretize conjugate systems into dynamic graphs. , border rights Determined by both physical connection strength and signal correlation, the calculation formula is as follows: ; in, It is spatial distance. It is the feature length. It is the time series correlation coefficient. It is an indicator function. For enhancement coefficient; The network outputs the damage probability and critical path saliency score of each node at future time, and locates the critical origin node and potential slip surface through graph attention weights and gradient backpropagation.

[0014] Therefore, the monitoring and evaluation system and method for conjugate soil slope restoration based on multi-source sensing, which adopts the above-mentioned structure, has the following beneficial effects: (1) This invention adopts the conjugate response field theory, abandons the binary discrete model of "repair body + slope", and establishes a unified mechanical field model of the conjugate system coupled with "repair body-soil body". The monitoring target is defined as the spatiotemporal distribution of interaction forces and coordinated strain energy in the field.

[0015] (2) This invention creates a four-dimensional temporal tensor model by spatiotemporal-modal tensor fusion, which expresses and repairs heterogeneous, asynchronous and different precision data in a unified mathematical space, thus solving the fundamental problem of "difficult alignment and difficult completion" of multi-source data.

[0016] (3) This invention embeds physical laws such as soil mechanics constitutive model and seepage equation as hard constraints into the neural network, and enables the model to have the ability to quickly adapt to small samples through the meta-learning paradigm, thus breaking through the dependence of traditional machine learning on a large amount of historical fault data.

[0017] (4) Based on graph neural networks, this invention combines the internal structural relationship of slopes with micro-damage signals to realize damage source tracing and prediction of slip surface formation path, transforming passive alarm into active prediction.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 The overall architecture diagram of the monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing provided in the embodiments of the present invention; Figure 2 This is a schematic diagram of the micro-damage propagation network and failure path prediction provided in an embodiment of the present invention; Figure 3 A schematic diagram of a "trunk-root" non-uniform encrypted network topology provided for an embodiment of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0022] Example like Figure 1 As shown, this embodiment of the invention provides a monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing. The system architecture follows a four-layer closed-loop logic of "heterogeneous perception - edge fusion - cloud evolution - autonomous decision-making." Specifically: A heterogeneous sensing layer is used to deploy a biomimetic multi-source sensor network to acquire multi-dimensional state information of the conjugate system. In this embodiment, the multi-source sensor network adopts a "trunk-root" non-uniform encrypted network topology to simulate the root system's soil-fixing principle.

[0023] Among them, the "backbone" is a distributed optical fiber sensor network pre-embedded in the core of the prosthesis (such as...). Figure 3 As shown in the diagram, the system continuously senses strain and temperature fields, providing long-distance, high-precision linear monitoring data. The "roots" are a cluster of sensors wirelessly deployed in the critical interaction zones of the conjugate system, including MEMS inclinometers, earth pressure gauges, and moisture sensors, used to monitor the local conditions at key points. This non-uniform deployment method can specifically capture the mechanical response of the interface between the repair body and the soil.

[0024] The edge fusion layer is used in runtime tensor fusion algorithms to generate aligned, complete, and highly reliable system state tensors. The edge fusion layer constructs a four-dimensional temporal tensor. As a unified mathematical expression of the state of a conjugate system.

[0025] in, In terms of spatial dimension, the three-dimensional spatial location is mapped to a one-dimensional index through spatial encoding, representing the spatial distribution of the sensor; The modal dimension represents different types of sensing physical quantities; The time dimension represents a continuous, equally spaced time series; The confidence dimension represents the confidence level of each data point. The edge fusion layer uses this model to solve the alignment and completion challenges caused by the heterogeneity, asynchronicity, and precision of multi-source data.

[0026] The cloud evolution layer includes a physical information element learning network and a micro-damage graph propagation model, which are used for state assessment and trend prediction based on the system state tensor.

[0027] The physical information meta-learning network consists of a physical information encoder and a meta-learning evaluator. The physical information encoder takes the joint feature vector and historical states as input and forces the network to learn feature representations that conform to the laws of mechanics through physical residual loss constraints. The meta-learning evaluator adopts a model-independent meta-learning framework to achieve fast adaptation with small samples.

[0028] Combination Figure 2 As shown, the micro-damage graph propagation model discretizes the conjugate system into a dynamic graph, uses a spatiotemporal graph convolutional network to process the graph sequence, captures the damage propagation between nodes and the state evolution of the nodes themselves, and thus predicts the failure path.

[0029] The autonomous decision-making layer is used to generate visualized diagnostic reports and graded early warnings based on the assessment and prediction results, forming operation and maintenance decision support. The system ultimately generates a dynamic cloud map of the conjugate response field, a system health evolution dashboard, a damage evolution heat map and a prediction path map, and outputs quantitative and actionable decision instructions (such as "It is recommended to start local grouting reinforcement").

[0030] This embodiment also provides a monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing. This method is based on the above-mentioned system and specifically includes the following steps: S1. Construct a biomimetic multi-source sensor network and adaptively deploy and schedule sensor nodes: The system dynamically adjusts its sensing strategy based on real-time system status. Each sensing node is defined. At any moment Perception priority And schedule according to priority. Priority awareness. The calculation formula is: ; in, The prior geological weakness coefficient for the location of nodes. This represents the real-time risk assessment value for a local area surrounding the node. This is a function to indicate abnormal node data. Continue working for the node The normalized value of energy consumption cost over time. This is the abnormal excitation gain coefficient.

[0031] The system's central scheduler periodically schedules all nodes according to... The system prioritizes nodes, increases their sampling frequency and switches them to high-precision mode, and implements energy-saving scheduling for low-priority nodes.

[0032] S2. Construct a four-dimensional temporal tensor model to perform multi-source data fusion and generation repair: In the edge fusion layer, mode-specific normalization is performed for each mode, and tensor completion is performed using high-order singular value decomposition with confidence constraints to generate the optimal complete system state tensor. The objective function of the completion process is: ; in, For the complete tensor to be found, To be related to confidence level The relevant weight tensor has a weight of 0 for missing data points and a weight greater than 1 for high-confidence data points. It represents the Hadamah accumulation. It is the Frobenius norm. For tensor rank regularization, Its coefficient.

[0033] S3. Construct a meta-learning dynamic evaluation model embedding physical information: In the cloud evolution layer, the physical information element learning dynamic evaluation model is constrained by minimizing the physical residual loss. Physical residual loss Defined as: ; in, It is the network's output of stress and strain estimates for the $k$-th key point. It is a function representing the constitutive relation of the conjugate system. It is a set of material parameters that evolve over time.

[0034] The meta-learning evaluator outputs the current system's overall health index through inner and outer layer updates. and future Risk evolution curve over a period of time .

[0035] S4. Construct a micro-damage propagation network to predict failure paths: The micro-damage propagation graph network employs a spatiotemporal graph convolutional network, utilizing Chebyshev graph convolution to capture damage propagation between nodes and one-dimensional temporal convolution to capture the state evolution of nodes themselves.

[0036] The network discretizes the conjugate system into a dynamic graph. , border rights Determined by both physical connection strength and signal correlation, the calculation formula is as follows: ; in, It is spatial distance. It is the feature length. It is the time series correlation coefficient. It is an indicator function. This is the enhancement coefficient.

[0037] The network outputs the damage probability and critical path saliency score of each node at future time, and locates the critical origin node and potential slip surface through graph attention weights and gradient backpropagation.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing, characterized in that, include: The heterogeneous sensing layer is used to deploy a multi-source sensor network in a biomimetic manner to obtain multi-dimensional state information of the conjugate system. The conjugate system refers to the system in which the repair body and the original soil body reinforced and constrained by it are regarded as a mechanically interacting and cooperatively responding system. The edge fusion layer is used for runtime tensor fusion algorithms to generate aligned, complete, and highly reliable system state tensors; The cloud evolution layer includes a physical information element learning network and a micro-damage graph propagation model, which are used for state evaluation and trend prediction based on the system state tensor. The autonomous decision-making layer is used to generate visual diagnostic reports and tiered early warnings based on the assessment and prediction results.

2. The monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing according to claim 1, characterized in that, In the heterogeneous sensing layer, the multi-source sensor network adopts a "backbone-root" non-uniform encrypted network topology; The backbone is a distributed optical fiber sensor network pre-embedded in the core of the repair body, used to continuously sense strain and temperature field. The fibrous roots are a cluster of sensors wirelessly deployed in the critical interaction region of the conjugate system. The sensor cluster includes a MEMS inclinometer, an earth pressure gauge, and a moisture content sensor.

3. The monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing according to claim 1, characterized in that, In the cloud evolution layer, the micro-damage graph propagation model discretizes the conjugate system into a dynamic graph. ,in For a set of nodes, Let be the set of edges. Let it be the set of edge weights; The micro-damage graph propagation model uses a spatiotemporal graph convolutional network to process graph sequences, utilizes Chebyshev graph convolution to capture damage propagation between nodes, uses one-dimensional temporal convolution to capture the state evolution of nodes themselves, and outputs the damage probability and critical path saliency score of each node at future time steps.

4. The monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing according to claim 1, characterized in that, In the edge fusion layer, a four-dimensional temporal tensor is constructed. As a unified mathematical expression for the state of a conjugate system; in, In terms of spatial dimension, the three-dimensional spatial location is mapped to a one-dimensional index through spatial encoding, representing the spatial distribution of the sensor; The modal dimension represents different types of sensing physical quantities; The time dimension represents a continuous, equally spaced time series; The confidence dimension represents the confidence level of each data point.

5. The monitoring and evaluation system for conjugate soil slope restoration based on multi-source sensing according to claim 1, characterized in that, In the cloud evolution layer, the physical information meta-learning network includes a physical information encoder and a meta-learning evaluator; The physical information encoder takes the joint feature vector and historical state as input, and forces the network to learn feature representations that conform to the laws of mechanics through physical residual loss constraints. The physical residual loss is the deviation between the stress and strain estimates output by the network and the constitutive relation function of the conjugate system. The meta-learning evaluator adopts a model-independent meta-learning framework, which divides the monitoring data stream into a series of continuous time-segment tasks. It achieves rapid adaptation of small samples through inner and outer layer updates, and outputs the system health index and risk evolution curve.

6. A method for monitoring and evaluating conjugate soil slope restoration bodies based on multi-source sensing, applied to the monitoring and evaluation system for conjugate soil slope restoration bodies based on multi-source sensing as described in any one of claims 1-5, characterized in that, Includes the following steps: S1. Construct a biomimetic multi-source sensor network, adaptively deploy and schedule sensor nodes, and obtain multi-dimensional state information of the conjugate system. S2. Construct a four-dimensional temporal tensor model as a unified mathematical expression of the state of the conjugate system, and perform multi-source data fusion and generation repair. S3. Construct a meta-learning dynamic evaluation model embedded with physical information, analyze the state tensor of the repaired system, and output the system health index and risk evolution curve. S4. Construct a micro-damage propagation network, predict failure paths, and generate a visual diagnostic report and graded early warning based on the assessment and prediction results.

7. The monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing according to claim 6, characterized in that, In step S1, the system dynamically adjusts the sensing strategy based on the real-time system status and defines each sensing node. At any moment Perception priority And schedule according to priority; Perception Priority The calculation formula is: ； in, The prior geological weakness coefficient for the location of nodes. This represents the real-time risk assessment value for a local area surrounding the node. This is a function to indicate abnormal node data. Continue working for the node The normalized value of energy consumption cost over time. This is the abnormal excitation gain coefficient.

8. The monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing according to claim 6, characterized in that, In step S2, multi-source data fusion and generation repair includes: performing mode-specific normalization on each mode and using confidence-constrained high-order singular value decomposition for tensor completion to generate the optimal complete system state tensor. The objective function of the completion process is: ； in, For the complete tensor to be found, To be related to confidence level The relevant weight tensor, It represents the Hadamah accumulation. It is the Frobenius norm. For tensor rank regularization, Its coefficient.

9. The monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing according to claim 6, characterized in that, In step S3, the physical information element learning dynamic evaluation model is constrained by minimizing the physical residual loss. Defined as: ； in, It is the network to the first Stress and strain estimates output at key points It is a function representing the constitutive relation of the conjugate system. It is a set of material parameters that evolve over time.

10. The monitoring and evaluation method for conjugate soil slope restoration based on multi-source sensing according to claim 6, characterized in that, In step S4, the micro-damage propagation graph network adopts a spatiotemporal graph convolutional network, which uses Chebyshev graph convolution to capture damage propagation between nodes and one-dimensional temporal convolution to capture the state evolution of the nodes themselves. Micro-damage propagation graph networks discretize conjugate systems into dynamic graphs. , border rights Determined by both physical connection strength and signal correlation, the calculation formula is as follows: ； in, It is spatial distance. It is the feature length. It is the time series correlation coefficient. It is an indicator function. For enhancement coefficient; The network outputs the damage probability and critical path saliency score of each node at future time, and locates the critical origin node and potential slip surface through graph attention weights and gradient backpropagation.

Images