Dynamic Analysis of Soil-Structure Interaction using Ensemble-based Modified Ant Lion Optimization Algorithm

Abstract

In recent decades, the analysis of dynamic characteristics of Soil-Structure Interaction (SSI) has become an emerging research topic, where the SSI is defined as the structure's motion and the soil's response. The SSI is an important problem in solid and monstrous structures, which are built on delicate ground that changes the dynamic properties of the structures. The main objective of this research article is to propose an ensemble machine-learning algorithm for predicting the dynamic response and characteristics of SSI problems. After collecting the data from 57 structures, the data pre-processing is accomplished using Min-Max Normalization (MMN) and Max Normalization (MN) techniques that superiorly rescale the unstructured data for better prediction. Further, the data optimization is carried out using the Modified Ant Lion Optimization (MALO) algorithm that effectively optimizes the dimensionality of the data, where this process reduces the computational complexity and improves the prediction accuracy of dynamic characteristics in SSI modeling. Finally, the optimized data is given as the input to the ensemble classifier, which is a combination of Support Vector Machine (SVM) and ID3 for classifying the dynamic characteristics related to SSI, which are period Lengthening (PL), Super Structure Acceleration (SSA) and Pile Head Acceleration (PHA). The simulation results confirmed that the ensemble-based MALO algorithm improved performance in predicting the dynamic response and characteristics of SSI problems by error value. Whereas the proposed algorithm, on average, reduced 0.01-to-0.5 error value compared to the existing machine learning algorithms.

Capability and Robustness of Novel Hybridized Artificial Intelligence Technique for Sediment Yield Modeling in Godavari River, India

“…These techniques include deterministic, linear and non-linear regression and conceptual models that can solve complex hydrological problems [26]. These artificial intelligence techniques have been used widely by various researchers in many interdisciplinary domains [27][28][29][30][31][32][33][34][35][36][37]. One appropriate artificial intelligence model is the ANN which has the capability to handle multiple variables of both linear and non-linear dependencies.…”