Dam deformation forecasting using SVM-DEGWO algorithm based on phase space reconstruction.

Area of Science:

• Geotechnical Engineering
• Computational Intelligence
• Time Series Analysis

Background:

• Dam deformation monitoring is crucial for structural safety.
• Traditional prediction models often struggle with the complex, nonlinear dynamics of dam behavior.
• Accurate forecasting of dam deformation is essential for risk assessment and management.

Purpose of the Study:

• To develop a robust hybrid model for analyzing and predicting dam deformation.
• To improve the accuracy and reliability of dam deformation forecasting.
• To integrate chaos theory, SVM, and an optimized algorithm for enhanced predictive performance.

Main Methods:

• Identification of chaotic characteristics in dam deformation time series using Lyapunov exponent, correlation dimension, and Kolmogorov entropy.
• Reconstruction of time series in phase space to determine SVM input variables.
• Optimization of SVM parameters using a hybrid Difference Evolution Grey Wolf Optimization (DEGWO) algorithm.

Main Results:

• The hybrid model, termed PSR-SVM-DEGWO, demonstrated superior fitting and prediction accuracy.
• Validation using actual monitoring data from the Jinping I super-high arch dam confirmed the model's effectiveness.
• The DEGWO algorithm showed improved performance in optimizing SVM parameters compared to existing methods.

Conclusions:

• The proposed PSR-SVM-DEGWO model offers a significant advancement in dam deformation analysis and prediction.
• This hybrid approach effectively captures the complex dynamics inherent in dam behavior.
• The model provides a reliable tool for enhancing dam safety monitoring and management.