CN-122024967-A - Island engineering material performance simulation method and system

CN122024967ACN 122024967 ACN122024967 ACN 122024967ACN-122024967-A

Abstract

The invention discloses a method and a system for simulating the performance of island engineering materials, and relates to the technical field of island engineering, wherein the method comprises the following steps of S1, obtaining tidal history data and on-site monitoring data of an island engineering structure; the method comprises the steps of S2, initializing and maintaining a shared state space, wherein characterization parameters of the shared state space are micro-environment activity indexes, hysteresis memory states and virtual bone stock ground, S3, based on tide history data and field monitoring data, performing structural editing on the hysteresis memory states by using a reinforcement learning frame to update the micro-environment activity indexes, S4, taking the updated micro-environment activity indexes as conditions, inputting the updated micro-environment activity indexes into a physical constraint generation model to update the virtual bone stock ground and calculate corresponding pumping dynamics response, S5, based on the pumping dynamics response and the updated micro-environment activity indexes, combining tide driving boundary conditions to update the shared state space, and S6, outputting a material performance prediction result of an island-reef engineering structure according to the updated shared state space.

Inventors

LIU QI
Hu Wenle
JI NAN
ZHANG HUI
FENG JUNJIE
HUANG QIANG
DU ZHIGANG
ZHAO JIANJIAN
LI MINGYU
LI GUANGHUI
CHENG LONG
WANG XINWU
MA JINGJING
MA YUNLING
XIE BING
GAO FUQIANG
CAI LIPENG
FU SHUAI
WANG YUHAO
YAN SHAOYANG

Assignees

洛阳理工学院

Dates

Publication Date: 20260512
Application Date: 20260209

Claims (10)

1. The island engineering material performance simulation method is characterized by comprising the following steps of: s1, tidal history data and field monitoring data of an island-reef engineering structure are obtained; S2, initializing and maintaining a shared state space, wherein the characterization parameters of the shared state space are a microenvironment activity index, a hysteresis memory state and a virtual bone stock ground; S3, based on the tide history data and the field monitoring data, utilizing a reinforcement learning framework to carry out structural editing on the hysteresis memory state so as to update the micro-environment activity index; S4, taking the updated micro-environment activity index as a condition, inputting the updated micro-environment activity index into a physical constraint generation model to update the virtual bone stock ground and calculate the corresponding pumping dynamics response; s5, updating the shared state space based on the pumping dynamics response and the updated micro-environment activity index in combination with tidal driving boundary conditions; And S6, outputting a material performance prediction result of the island engineering structure according to the updated shared state space.
2. The island engineering material performance simulation method according to claim 1, wherein the step S3 of structurally editing the hysteresis memory state by using a reinforcement learning frame comprises: S31, constructing a plurality of hysteresis memory units corresponding to the principal tide component frequency, wherein the inversion threshold interval of the hysteresis memory units corresponding to the full-day tide is set to be a first range, and the inversion threshold interval of the hysteresis memory units corresponding to the half-day tide is set to be a second range which is larger than the first range.
3. The island engineering material performance simulation method according to claim 2, wherein the method further comprises: s32, defining the state of the reinforcement learning framework, wherein the state of the reinforcement learning framework comprises a historical sequence of the micro-environment activity index, the porosity and connectivity distribution of the virtual bone stock ground representation and the current state set of all hysteresis memory units; S33, defining the action of the reinforcement learning framework, wherein the action of the reinforcement learning framework comprises at least one editing operation of all hysteresis memory units: The method comprises the steps of carrying out numerical drift on the turnover threshold value of a selected unit according to current tidal acceleration, resetting the states of the units in a preset space area, and adding a new unit with a new turnover threshold value.
4. An island engineering material performance simulation method according to claim 3, wherein the method further comprises: S34, defining a reward function of the reinforcement learning framework, wherein the reward function is a weighted sum of at least two sub-rewards, namely a first sub-reward based on simulation and actual measurement of electrochemical impedance spectrum difference, a second sub-reward based on the physical constraint generation model to output consistency under the prediction condition and the observation condition, and a third sub-reward based on system activity change and tidal drive synchronism; And S35, executing reinforcement learning decision according to the state, action and rewarding function of the reinforcement learning framework, updating the hysteresis memory state, and aggregating to obtain an updated micro-environment activity index.
5. The island engineering material performance simulation method according to claim 1, wherein the physical constraint generation model in S4 generates a flow network, the physical constraint generation model updates the virtual bone stock ground and calculates a pumping dynamics response corresponding thereto, and the method comprises: s41, taking the updated micro-environment activity index and the spatial coordinates as condition input; s42, executing a serialization generation step, namely selecting the next generation action from action sets comprising pore growth, region closure, branching and merging by adopting Boltzmann distribution according to the gradient direction of the pressure field in the aggregate structure generated by the current part and the structure growth history, wherein the action score of the generation action is determined jointly by the gradient alignment degree of the pressure field and the history growth continuity.
6. The island engineering material performance simulation method according to claim 5, further comprising: s43, after each generating action is executed, calling a differentiable physical calculation layer, and performing single-step fluid pressure and ion concentration field simulation on the current structure; S44, adjusting a selection strategy of a subsequent generation action based on the physical field residual error output by the differentiable physical calculation layer; S45, repeating S42 to S44, generating a complete virtual bone stock ground, and synchronously outputting a parameterized response function for representing pumping dynamics.
7. The island engineering material performance simulation method according to claim 1, further comprising, between S3 and S4: the updated micro-environment activity index and target area information are sent to the physical constraint generation model and used as condition input; And receiving key physical characteristics of the virtual bone stock ground generated based on the conditions and fed back by the physical constraint generation model, wherein the key physical characteristics comprise aggregate porosity average value, characteristic transmission path length and pumping response time scale, and are used for updating the state representation of the reinforcement learning framework.
8. The method for simulating the performance of an island engineering material according to claim 1 or 7, further comprising, between S3 and S4: Calculating the space variation of the micro-environment activity index before and after updating; Transmitting the spatial variation to the physical constraint generation model; and the physical constraint generation model dynamically adjusts the Darcy permeability tensor principal diagonal component of the differentiable physical calculation layer according to the space variation.
9. The island engineering material performance simulation method according to claim 1, further comprising: after each integer number of tidal cycles, the following iterations are performed: Fixing a current virtual bone stock ground, iteratively operating S3 and S5, and optimizing a micro-environment activity index by taking a condition that the change rate of a reward function of the reinforcement learning frame is smaller than a preset change threshold or the maximum iteration number is reached; fixing the current micro-environment activity index, iterating S4, and sampling to generate a virtual bone stock ground; and verifying the currently paired micro-environment activity index and the virtual aggregate field.
10. An island-reef engineering material performance simulation system that implements the island-reef engineering material performance simulation method of claim 1, comprising: the data acquisition unit is used for acquiring tidal history data and field monitoring data of the island engineering structure; the initialization unit is used for initializing and maintaining a shared state space, and the characterization parameters of the shared state space are a microenvironment activity index, a hysteresis memory state and a virtual bone stock ground; The structured editing unit is used for carrying out structured editing on the hysteresis memory state by utilizing a reinforcement learning framework based on the tide history data and the field monitoring data so as to update the micro-environment activity index; The first updating unit is used for inputting the updated micro-environment activity index into the physical constraint generation model as a condition so as to update the virtual bone stock ground and calculate the corresponding pumping dynamics response; A second updating unit for updating the shared state space in combination with tidal driven boundary conditions based on the pumping dynamics response and the updated micro-environmental activity index; And the output unit is used for outputting a material performance prediction result of the island engineering structure according to the updated shared state space.

Description

Island engineering material performance simulation method and system Technical Field The application relates to the technical field of island engineering, in particular to a method and a system for simulating the performance of island engineering materials. Background Island engineering structures are in a severe marine environment with high salt, high humidity and strong sunlight for a long time, and the degradation of material performance of the island engineering structures is directly related to the safety and durability of the engineering. Therefore, the island engineering material performance is accurately simulated and predicted, and the island engineering material performance simulation method has important engineering significance and theoretical research value for guiding material selection, optimizing structural design, formulating scientific maintenance strategy and reducing total life cycle cost. However, existing material performance simulation methods have certain limitations in facing the characteristic, long-term cyclic tidal action of island environments. On the one hand, conventional models often have difficulty adequately characterizing the complex relationship between tidal dynamic cycles and the material's internal microenvironment response, resulting in bias in the prediction of long-term performance evolution. On the other hand, the prior method has insufficient capability in the aspect of learning and representing time-varying evolution rules of a microstructure of a material from limited field monitoring data, which restricts the restoring precision and the predicting reliability of a simulation model to a real physical process. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the application provides a method and a system for simulating the performance of island engineering materials, which are used for solving the technical problems. The application provides an island engineering material performance simulation method which comprises the steps of S1, obtaining tidal history data and field monitoring data of an island engineering structure, S2, initializing and maintaining a shared state space, S3, based on the tidal history data and the field monitoring data, carrying out structural editing on the delayed memory state by utilizing a reinforcement learning frame to update the micro-environment activity index, S4, taking the updated micro-environment activity index as a condition, inputting the updated micro-environment activity index into a physical constraint generation model to update the virtual bone site and calculate a pumping dynamics response corresponding to the virtual bone site, S5, based on the pumping dynamics response and the updated micro-environment activity index, updating the shared state space by combining with tidal driving boundary conditions, and S6, outputting a material performance prediction result of the island engineering structure according to the updated shared state space. The application provides an island engineering material performance simulation system which comprises a data acquisition unit, an initialization unit, a structural editing unit and an output unit, wherein the data acquisition unit is used for acquiring tidal history data and field monitoring data of an island engineering structure, the initialization unit is used for initializing and maintaining a shared state space, characterization parameters of the shared state space are a micro-environment activity index, a hysteresis memory state and a virtual bone stock ground, the structural editing unit is used for carrying out structural editing on the hysteresis memory state by using a reinforcement learning frame based on the tidal history data and the field monitoring data so as to update the micro-environment activity index, the first updating unit is used for taking the updated micro-environment activity index as a condition, inputting the updated micro-environment activity index into a physical constraint generation model so as to update the virtual bone stock ground and calculate a pumping dynamics response corresponding to the micro-environment activity index, the second updating unit is used for updating the shared state space based on the pumping dynamics response and the updated micro-environment activity index in combination with a driving boundary condition, and the output unit is used for outputting a material performance prediction result of the island engineering structure according to the updated shared state space. Based on the embodiment provided by the application, the key states in the simulation process can be co-evolved and consistently updated by maintaining a shared state space which uniformly characterizes the micro-environmental activity index, the hysteresis memory state and the virtual bone stock ground. Therefore, the limitation of loose coupling of different physical fields or state var