CN-121997679-A - AI prediction system for reservoir dam safety assessment

CN121997679ACN 121997679 ACN121997679 ACN 121997679ACN-121997679-A

Abstract

The invention relates to the technical field of reservoir dam safety management and discloses an AI prediction system for reservoir dam safety assessment, which comprises a monitoring data acquisition unit, a geometric space mapping unit, an adaptive weight distribution module, a safety risk early warning module and a safety risk early warning module, wherein the monitoring data acquisition unit is used for receiving monitoring data streams representing seepage, displacement and stress states, the geometric space mapping unit is used for determining grid nodes according to a dam space grid model, establishing a space corresponding relation between monitoring indexes and the grid nodes, and based on physical space geometric indexes for the monitoring indexes, the adaptive weight distribution module is used for extracting feature vectors according to the change rate of the monitoring indexes and locking weight change gradients of adjacent grid nodes by utilizing an adjacent incidence matrix so that the interconnected grid nodes are synchronously corrected in weight calculation, and the safety risk early warning module is used for calculating safety quantization scores and triggering risk management and control logic.

Inventors

BIAN HAIBO
ZHANG CHI
ZHANG ZHIFANG
FU PENG
YAN XINGANG
LI CONGSHI
HUANG GUIJUN
YE GUO

Assignees

湖南湘银河传感科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. An AI prediction system for reservoir dam safety assessment, comprising: the monitoring data acquisition unit is used for receiving real-time monitoring data flow of the dam, wherein the real-time monitoring data flow comprises a plurality of monitoring indexes representing seepage states, displacement states and stress states; the geometrical space mapping unit is used for determining a plurality of grid nodes according to the space grid model of the dam, and establishing a space corresponding relation between each monitoring index and each grid node so as to be based on a physical space geometrical index for the monitoring index; The self-adaptive weight distribution module is used for extracting feature vectors according to the change rate of each monitoring index relative to a preset initial value, and determining the weight of each monitoring index by utilizing a weighting algorithm, wherein an adjacent incidence matrix established based on a physical space geometric index is introduced as a weight distribution constraint factor in the calculation process, so that the weight change gradient corresponding to adjacent grid nodes is limited based on the connection state among the grid nodes, and the monitoring indexes corresponding to the interconnected grid nodes are synchronously corrected in the weight calculation process; the safety risk early warning module is used for calculating the safety quantitative score of the dam based on the calculated weight and outputting a risk early warning instruction according to the mapping result of the safety quantitative score and the preset risk interval.
2. The AI prediction system for reservoir dam safety assessment according to claim 1, wherein the adaptive weight distribution module adjusts the priority of the monitoring index in the weight distribution algorithm based on the change rate of the current moment when extracting the feature vector, and increases the response speed of the safety risk early warning module to the risk feature by increasing the corresponding component in the feature vector when the water level change amount or the seepage pressure value exceeds the preset sensitivity threshold.
3. The AI prediction system for reservoir dam safety assessment of claim 1, wherein the plurality of monitoring indicators indicative of seepage conditions, displacement conditions, and stress conditions include seepage pressure, fracture opening and closing, dam surface displacement, internal stress, and reinforcement corrosion depth.
4. The AI prediction system for reservoir dam safety assessment of claim 1, wherein the adaptive weight distribution module extracts a variation characteristic of each monitoring index based on a function of Y= (X (t)/X '-1) ×100%, wherein Y is a variation rate, X (t) is a current time monitoring value of the monitoring index, and X' is a preset initial value of the monitoring index.
5. The AI prediction system for reservoir dam safety assessment according to claim 1, wherein the geometric space mapping unit divides the geometric surface and the internal structure of the dam into interconnected mesh topology units when determining mesh nodes, and assigns unique space coordinate codes as physical space geometric indexes to each mesh node.
6. The AI prediction system for reservoir dam safety assessment according to claim 1, wherein the adaptive weight distribution module performs association constraint on the suddenly-changed monitoring index through the adjacent association matrix, and when abnormal fluctuation occurs in the monitoring index of a specific grid node, the duty ratio of the adjacent area in weight distribution is synchronously increased based on the spatial correspondence relationship so as to perform trend amplification on the local risk characteristics through the connectivity of the geometric structure.
7. The AI prediction system for reservoir dam safety assessment according to claim 1, wherein the evaluation model constructed by the safety risk early warning module comprises four management dimensions of field inspection, monitoring analysis, flood control capability and digital-analog analysis, and the safety risk early warning module generates a safety state predicted value of the dam in a subsequent service period by logically arranging quantization indexes of the four management dimensions.
8. The AI prediction system for reservoir dam safety assessment of claim 1, wherein the preset risk intervals include an extremely high risk interval, a medium risk interval, and a low risk interval, and the safety risk early warning module triggers the risk management logic of the corresponding level according to the preset risk interval in which the safety quantitative score is located.
9. The AI prediction system for reservoir dam safety assessment according to claim 1, wherein the monitoring data acquisition unit is further configured to acquire weather monitoring data and water level fluctuation data of an environment in which the dam is located, and input the weather monitoring data and the water level fluctuation data as external constraint variables to the safety risk early warning module, and the safety risk early warning module performs weighted cancellation correction on the safety quantization score based on the weather monitoring data and the water level fluctuation data to eliminate interference of environmental noise on the safety state assessment.
10. The AI prediction system for reservoir dam safety assessment according to claim 1, further comprising a self-calibration feedback unit for comparing a safety state predicted value output by the safety risk early warning module with a subsequent actual service state value of the dam, and reversely adjusting a scaling factor of the weight distribution constraint factor in the adaptive weight distribution module according to the generated comparison residual.

Description

AI prediction system for reservoir dam safety assessment Technical Field The invention relates to an AI prediction system for reservoir dam safety assessment, and belongs to the technical field of reservoir dam safety management. Background In the current reservoir dam safety evaluation, a data processing system is used for integrating multisource monitoring data and assisting management decisions, the current commonly adopted technical mode is based on an evaluation index weight system, the dam safety evaluation grade is calculated and obtained through collecting physical quantities of sensors such as water level, displacement and seepage, the technical scheme provides standardized processing logic for dam risk management, the dam safety evaluation system has certain applicability in safety evaluation under a conventional service environment, the dam safety state evolution presents complex nonlinear characteristics along with the increase of the service life of hydraulic engineering and the fluctuation of extreme environmental load, the traditional scheme generally takes monitoring indexes such as seepage, stress and displacement as mutually independent numerical variables, the processing mode presents obvious limitation under pressure working conditions, the physical structure topological relation of the monitoring data and a dam entity is stripped, the evaluation result cannot reflect the conduction path of risks in the dam structure, and the space topology perception loss in the data processing process is generated. Aiming at the limitations, the bottom contradiction cannot be solved by adopting modes such as increasing the sensor density or applying a general statistical model, mass redundant data can be generated by increasing the number of sensors, the calculation load of a system is improved, the situation that monitoring indexes are mutually isolated at a logic level is not changed, a general mathematical regression model lacks physical geometric constraint, the stress deformation rule of a dam under a specific three-dimensional boundary condition is difficult to simulate, an evaluation result generated by purely relying on mathematical fitting is often separated from engineering reality, when a dam body is locally deteriorated, the system is difficult to accurately capture the continuity influence on an adjacent structure, for example, the Chinese patent application with the publication number of CN121409344A discloses a safety monitoring method and system applied to a small reservoir dam, environmental risk influence coefficient correction is introduced from comprehensive risk evaluation of seepage flow, the core algorithm logic is dependent on the premise that the linear independence or the risk distribution has ideal statistical distribution preset, the physical geometric constraint of the dam body is absent in the prior art, the evaluation process and the physical space topology are not reflected, and the continuity influence on the internal conduction path of the three-dimensional structure is difficult to capture the continuity influence on the adjacent structure when the dam body is subjected to non-ideal local deterioration, for example, the space connectivity is lost, and the continuity influence on the adjacent structure is difficult to be captured. Therefore, how to construct a data processing logic for anchoring dynamic monitoring indexes to the grid nodes of the three-dimensional geometric model of the dam, so that the calculation of the evaluation weight is controlled by the space topology constraint of a physical entity, and the accurate prediction of the structural degradation trend is realized based on the grid connectivity, so that the method and the system are the technical problems to be solved. Disclosure of Invention In order to solve the problems in the background technology, the technical scheme of the invention is as follows, an AI prediction system for reservoir dam safety assessment comprises: the monitoring data acquisition unit is used for receiving real-time monitoring data flow of the dam, wherein the real-time monitoring data flow comprises a plurality of monitoring indexes representing seepage states, displacement states and stress states; the geometrical space mapping unit is used for determining a plurality of grid nodes according to the space grid model of the dam, and establishing a space corresponding relation between each monitoring index and each grid node so as to be based on a physical space geometrical index for the monitoring index; The self-adaptive weight distribution module is used for extracting feature vectors according to the change rate of each monitoring index relative to a preset initial value, and determining the weight of each monitoring index by utilizing a weighting algorithm, wherein an adjacent incidence matrix established based on a physical space geometric index is introduced as a weight distribution constraint factor in the calculation