Slope safety factor prediction based on PSO-LSTM

In slope engineering, the safety factor of a slope is a crucial indicator for evaluating its stability. Traditional methods, grounded in physical and mechanical theories, such as the limit equilibrium method and finite element numerical method, encounter challenges such as complex model setup, intri...

Full description

Saved in:
Bibliographic Details
Published in2024 4th International Conference on Neural Networks, Information and Communication (NNICE) pp. 881 - 888
Main Authors Zeng, Jiahao, Shi, Yuhen, Yu, Peng
Format Conference Proceeding
LanguageEnglish
Published IEEE 19.01.2024
Subjects
Algorithm optimization
Finite element analysis
Model building
Neural network
Numerical models
Particle swarm optimization
Prediction algorithms
Predictive models
Safety
slope safety factor
Stability analysis
Online AccessGet full text

Cover

More Information
Summary:In slope engineering, the safety factor of a slope is a crucial indicator for evaluating its stability. Traditional methods, grounded in physical and mechanical theories, such as the limit equilibrium method and finite element numerical method, encounter challenges such as complex model setup, intricate calculations, and difficult convergence. This paper introduces a slope safety factor prediction method (PSO-LSTM) that combines Particle Swarm Optimization (PSO) and Long Short-Term Memory Neural Network (LSTM) techniques to forecast the slope safety factor. Unlike physical numerical methods, this approach eliminates the need for intricate physical modeling, significantly enhancing prediction efficiency. In comparison to traditional neural networks, the PSO algorithm optimizes the hyperparameters within the LSTM network model, minimizing the error between predicted and actual values, thereby improving the accuracy of slope safety factor prediction. Through predictions based on a set of real slope data, the accuracy rate exceeds 90%, confirming the precision, generalization ability, and effectiveness of the PSO-LSTM method.
DOI:10.1109/NNICE61279.2024.10499047