Dam displacement missing data completion method and system based on clustering and deep learning

By combining clustering and deep learning methods with adaptive weighted derivative dynamic time warping and hierarchical clustering, and selecting multiple influencing factors, a DS-ResNet model was established. This solved the problem of accuracy in dam displacement monitoring data completion, and enabled more accurate completion of missing displacement values ​​and dam safety monitoring.

CN117951124BActive Publication Date: 2026-06-19YANGZHOU UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
YANGZHOU UNIV
Filing Date
2023-12-22
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing methods for completing dam displacement monitoring data cannot accurately reflect the true spatial displacement field of the dam. In particular, when the length of the monitoring data is inconsistent and the monitoring frequency is different, it is impossible to accurately calculate the displacement similarity, which leads to an inability to effectively interpret the deformation behavior of the dam.

Method used

Based on clustering and deep learning methods, the similarity of displacement measurement points is calculated by adaptive weighted derivative dynamic time warping and adaptive weighted dynamic time warping. The displacement measurement points of the dam are partitioned by hierarchical clustering, and influencing factors such as water level, temperature and time are selected. A DS-ResNet model is established to complete missing values.

Benefits of technology

It can more accurately measure the similarity between dam displacement measurement points, provide more scientific and accurate dam displacement zoning results, reduce the risk of overfitting, more accurately fill in missing displacement values, improve the accuracy of dam safety monitoring, and promptly detect potential problems and risks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117951124B_ABST
    Figure CN117951124B_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for completing missing dam displacement data based on clustering and deep learning. The method includes: collecting original displacement monitoring data of a concrete dam; clustering and partitioning dam displacement monitoring points to extract spatiotemporal correlation information of dam displacement; selecting water level, temperature, timeliness, and the spatiotemporal correlation of displacement as influencing factors for dam displacement; establishing a deep learning model to complete missing values ​​of displacement monitoring points; obtaining missing displacement values ​​of the dam; and evaluating the accuracy of the model completion. This method overcomes the problem of long-sequence displacement weight values ​​tending towards extreme values. It better conforms to the operational behavior and patterns of dams and can more accurately complete missing displacement values. By completing missing displacement data, the deformation of the dam can be more comprehensively and accurately understood, improving the accuracy of dam safety monitoring, helping to promptly identify potential problems and risks in dam operation, and providing technical support for ensuring the long-term safe operation of the dam.
Need to check novelty before this filing date? Find Prior Art