CN-122021173-A - Dam structure digital twin perception method and system based on sensor network optimization

CN122021173ACN 122021173 ACN122021173 ACN 122021173ACN-122021173-A

Abstract

The invention provides a dam structure digital twin perception method and system based on sensor network optimization, which relate to the technical field of dam safety monitoring and comprise the steps of constructing a three-dimensional finite element model of a dam structure; the method comprises the steps of performing compressed sensing on a dam structure observation process to obtain a digital twin sensing model of the dam structure, constructing quantitative indexes and optimization conditions of the digital twin sensing, optimizing a sensor network of the dam structure based on the quantitative indexes and the optimization conditions by combining a three-dimensional finite element model to obtain an optimal sensor network, and performing training optimization on the digital twin sensing model through monitoring data of the optimal sensor network to obtain a digital twin body of the dam structure, wherein the digital twin body is used for obtaining global response data of the dam structure. The invention solves the problems that the global structural state of the dam is difficult to effectively perceive under the limited monitoring condition, the digital twin perception mechanism lacks theoretical support, and the monitoring network perception capability is difficult to quantitatively evaluate.

Inventors

CHEN CHEN
TIAN JICHEN
LI YANLING
WU ZHENYU
ZHANG HAN
PEI LIANG
LU XIANG
ZHOU JINGREN

Assignees

四川大学

Dates

Publication Date: 20260512
Application Date: 20260210

Claims (10)

1. The dam structure digital twin perception method based on sensor network optimization is characterized by comprising the following steps of: Constructing a three-dimensional finite element model of the dam structure; Compressive sensing is carried out on the observation process of the dam structure, and a digital twin sensing model of the dam structure is obtained; Constructing a quantization index and an optimization condition of digital twin perception; Based on the quantization index and the optimization condition, optimizing the sensor network of the dam structure by combining the modal matrix provided by the three-dimensional finite element model to obtain an optimal sensor network; And training and optimizing the digital twin perception model through the monitoring data of the optimal sensor network to obtain a digital twin body of the dam structure, wherein the digital twin body is used for acquiring global response data of the dam structure.
2. The sensor network optimization-based digital twin perception method for the dam structure according to claim 1, wherein the compressive sensing is performed on the observation process of the dam structure to obtain a digital twin perception model of the dam structure, and the method comprises the following steps: Performing compressed sensing on the observation process of the dam structure to obtain a compressed sensing model of the dam structure; Mapping the compressed sensing model into a neural network taking a sparse observation vector as input and a high-dimensional state vector as prediction output; Acquiring a global state simulation value and a sparse observation simulation value based on a numerical simulation result of the three-dimensional finite element model; constructing an objective function of the neural network based on the global state simulation value and the sparse observation simulation value; constructing an effective measurement operator through a modal matrix and a perception network observation matrix of the dam structure; constructing cross-domain adaptation constraint and information integrity meeting conditions through an effective measurement operator; And obtaining a digital twin perception model based on the neural network, the cross-domain adaptation constraint and the information integrity meeting condition.
3. The sensor network optimization-based dam structure digital twin sensing method of claim 2, wherein the performing compressed sensing on the dam structure observation process to obtain a compressed sensing model of the dam structure comprises: decomposing the dam structure state into a mode combination form based on a control theory to obtain a high-dimensional state vector of the dam structure state, wherein the high-dimensional state vector is formed by a mode matrix and a time-varying mode coefficient of the dam structure; Performing linear projection on the observation process of the dam structure to obtain a linear relation between a sparse observation vector and a high-dimensional state vector; Substituting the mode combination form of the high-dimensional state vector into the linear relation to obtain the compressed sensing model of the dam structure.
4. The sensor network optimization-based dam structure digital twin perception method according to claim 1, wherein the constructing quantization indexes and optimization conditions of digital twin perception comprises: The method comprises the steps of constructing an observability index of each mode through a perception network observation matrix and a mode matrix, wherein the perception network observation matrix is constructed through a sensor network; singular value decomposition is carried out on the effective measuring operator, and the condition number of the effective measuring operator is obtained through construction; taking the observability index of each mode and the condition number of the effective measuring operator as quantitative indexes of digital twin perception; and combining a preset threshold value, and constructing an optimization condition of digital twin perception through a quantization index.
5. The sensor network optimization-based dam structure digital twin sensing method according to claim 1, wherein the optimizing the sensor network of the dam structure based on the quantization index and the optimization condition by combining the modal matrix provided by the three-dimensional finite element model to obtain the optimal sensor network comprises the following steps: defining an optimized objective function of the sensor network; Performing modal analysis on the three-dimensional finite element model to obtain a modal matrix of the dam structure; performing QR decomposition on the transposition of the modal matrix, and acquiring a first preset number of initially selected sensor nodes by combining a column principal component selection method and an optimization objective function to obtain a current sensor set; Constructing a candidate node set by excluding the remaining nodes after the current sensor set; Expanding sensor nodes by adopting a greedy algorithm through quantization indexes and optimization conditions, and screening out a second preset number of expanded sensor nodes in a candidate node set; Updating a perception network observation matrix and an effective measurement operator through the initially selected sensor nodes and the expanded sensor nodes; judging whether the optimization condition is met or not through the updated perception network observation matrix and the updated effective measurement operator; if the optimization conditions are not met, removing the sensor nodes which do not meet the optimization conditions, and re-selecting the expanded sensor nodes; and if the optimization condition is met, taking the current sensor set as an optimal sensor network.
6. The sensor network optimization-based dam structure digital twin perception method according to claim 5, wherein the sensor node expansion is performed by a greedy algorithm through quantization indexes and optimization conditions, and a second preset number of expanded sensor nodes are obtained by screening in a candidate node set, comprising: for each candidate sensor node, constructing a temporary sensor set through the current sensor set and the candidate sensor node, wherein the candidate sensor node is any sensor node in the candidate node set; Calculating a corresponding perception network observation matrix and an effective measurement operator based on the temporary sensor set; calculating the quantization index of each temporary sensor set through a perception network observation matrix and an effective measurement operator; substituting the quantization indexes into an optimization objective function, and calculating an optimization objective function value of each candidate sensor node corresponding to the temporary sensor set; selecting a candidate sensor node as an expansion sensor node based on the optimization objective function value maximization and the spatial distribution constraint, adding the expansion sensor node into a current sensor set, and removing the expansion sensor node from the candidate node set; and selecting the next expansion sensor node through the updated current sensor set and the updated candidate node set until a second preset number of expansion sensor nodes are obtained.
7. The dam structure digital twin perception system based on sensor network optimization is characterized by comprising: the first construction module is used for constructing a three-dimensional finite element model of the dam structure; the sensing module is used for performing compressed sensing on the observation process of the dam structure to obtain a digital twin sensing model of the dam structure; The second construction module is used for constructing quantization indexes and optimization conditions of digital twin perception; The optimization module is used for optimizing the sensor network of the dam structure by combining the modal matrix provided by the three-dimensional finite element model based on the quantization index and the optimization condition to obtain an optimal sensor network; the training module is used for training and optimizing the digital twin perception model through the monitoring data of the optimal sensor network to obtain a digital twin body of the dam structure, and the digital twin body is used for acquiring global response data of the dam structure.
8. The sensor network optimization based dam structure digital twin perception system according to claim 7, wherein the perception module comprises: the compressed sensing unit is used for performing compressed sensing on the dam structure observation process to obtain a compressed sensing model of the dam structure; The mapping unit is used for mapping the compressed sensing model into a neural network taking a sparse observation vector as an input and a high-dimensional state vector as a prediction output; the acquisition unit is used for acquiring a global state simulation value and a sparse observation simulation value based on a numerical simulation result of the three-dimensional finite element model; The first construction unit is used for constructing an objective function of the neural network based on the global state simulation value and the sparse observation simulation value; the second construction unit is used for constructing an effective measurement operator through the modal matrix of the dam structure and the perception network observation matrix; the third construction unit is used for constructing cross-domain adaptation constraint and information integrity meeting conditions through an effective measurement operator; And the fourth construction unit is used for obtaining a digital twin perception model based on the neural network, the cross-domain adaptation constraint and the information integrity meeting condition.
9. The sensor network optimization based dam structure digital twin perception system according to claim 7, wherein the second construction module comprises: a fifth construction unit, configured to construct an observability index of each mode through a sensing network observation matrix and a mode matrix, where the sensing network observation matrix is constructed through a sensor network; the decomposition unit is used for carrying out singular value decomposition on the effective measurement operator and constructing the condition number of the effective measurement operator; The sixth construction unit is used for taking the observability index of each mode and the condition number of the effective measurement operator as quantitative indexes of digital twin perception; and the seventh construction unit is used for constructing the optimization condition of digital twin perception through the quantization index in combination with the preset threshold.
10. The sensor network optimization-based dam structure digital twin perception system according to claim 7, wherein the optimization module comprises: The definition unit is used for defining an optimization objective function of the sensor network; the analysis unit is used for carrying out modal analysis on the three-dimensional finite element model to obtain a modal matrix of the dam structure; The decomposition unit is used for performing QR decomposition on the transpose of the modal matrix, and acquiring a first preset number of initially selected sensor nodes by combining a column principal component selection method and an optimized objective function to obtain a current sensor set; An eighth construction unit, configured to construct a candidate node set by excluding remaining nodes after the current sensor set; The expansion unit is used for expanding the sensor nodes by adopting a greedy algorithm through the quantization indexes and the optimization conditions, and screening out a second preset number of expanded sensor nodes in the candidate node set; The updating unit is used for updating the perception network observation matrix and the effective measurement operator through the primary selection of the sensor nodes and the expansion of the sensor nodes; the judging unit is used for judging whether the optimization condition is met or not through the updated perception network observation matrix and the updated effective measurement operator; the first processing unit is used for removing the sensor nodes which do not meet the optimization conditions if the optimization conditions are not met, and re-selecting the expansion sensor nodes; And the second processing unit is used for taking the current sensor set as the optimal sensor network if the optimization condition is met.

Description

Dam structure digital twin perception method and system based on sensor network optimization Technical Field The invention relates to the technical field of dam safety monitoring, in particular to a dam structure digital twin sensing method and system based on sensor network optimization. Background The dam is used as an important hydraulic building, is widely applied to projects such as flood control, water supply, power generation, water resource allocation and the like, and the operation safety of the dam is directly related to the life and property safety of people and the stable development of regional economy and society. In recent years, along with the continuous expansion of the scale of hydraulic engineering, the dam has the characteristics of increased dam height, complex structure, diversified operation conditions and the like, and the dam body and the dam foundation are in long-term service under the conditions of high water head, high stress and multi-field coupling, and the structural response and damage evolution process of the dam have obvious nonlinear and time-varying characteristics, so that the dam safety monitoring and operation management are challenged greatly. At present, the dam structure health monitoring mainly depends on permanent monitoring facilities such as measuring points of displacement, stress, strain, seepage and the like which are arranged according to related specifications, and a monitoring and evaluation model is built based on measuring point data. The monitoring mode based on the measuring points is mature in application in engineering practice, but is limited in the number and the spatial distribution of the monitoring measuring points due to factors such as engineering cost, construction conditions and operation and maintenance cost, and the overall perception of the integral state of the dam structure is difficult to realize. Under the condition of complex working conditions or local damage, the global response characteristics of the dam body are difficult to accurately reflect only by means of a small amount of monitoring data, and the problem of insufficient perceptibility exists. In order to improve the monitoring and analysis capability, a plurality of safety monitoring methods based on models are proposed in the prior art. The method is based on continuous medium mechanics and structural mechanics theory, and analyzes deformation, stress and stability of the dam structure by establishing a numerical model. The method has good physical meaning and engineering interpretation, but has strong dependence on model parameters, boundary conditions and working condition assumptions, complex model calibration process and high calculation cost, and is difficult to meet the requirements of quick update and real-time analysis in the operation period. Another class of methods uses statistical or machine learning models to model and predict dam structure responses based on historical monitoring data. Statistical models based on empirical assumptions and shallow machine learning methods have been used in engineering, but they have limited ability to characterize complex nonlinear relationships and multi-factor coupling effects, and have insufficient adaptability to operating conditions or extreme conditions. In recent years, a deep learning method is introduced into the field of dam safety monitoring due to strong nonlinear modeling capability, but the method is highly dependent on limited measurement point data, so that effective characterization of the global structural state of the dam is difficult to realize, and meanwhile, the method still has defects in the aspects of physical interpretability and engineering reliability. With the development of intelligent perception and information technology, a digital twin technology is used for dam safety monitoring and analysis, and mapping of a virtual model and an actual dam operation state is achieved by fusing physical entities, a numerical model and monitoring data. However, the application of the existing digital twin technology in dam engineering is still in the primary stage, the sensing capability of the existing digital twin technology depends on the existing monitoring network to a great extent, the system description cannot be carried out on the perceptibility of the dam structure under the limited observation condition from the theoretical layer, and the relation between sensor layout and sensing effect lacks a quantitative evaluation method, so that the application effect of the digital twin model under the complex working condition is restricted. In summary, the existing dam safety monitoring and digital twin technology generally has the problems of sparse monitoring data space, insufficient sensing capability of the global structural state, lack of system theoretical support of sensing mechanism and the like, and is difficult to evaluate the safety state of the whole structure of the dam in a high-precision and q