Soil-structure interaction analysis using neural networks optimised by genetic algorithm

ABSTRACT

The soil-structure systems are infinite in nature regarding the solid medium. This geometrical infinity has been tackled by devising different remedies in the shape of limiting the system dimensions to consistent or transmitting boundaries. Yet, an exact soil-structure system is too difficult and time consuming to analyse especially when nonlinearities are involved in the problem. Moreover, the mentioned boundaries have mostly been introduced only for simple geometries. In recent years, use of smart data-based methods for simulation and analysis of complex engineering problems has attracted many relevant research works. In this paper, application of optimised neural networks, as an important branch of data-based procedures, for solving the soil-structure problem is examined. Classification based on the cross validation and K-fold validation approaches and optimising inclination and weight values using the genetic algorithm are utilised to optimise performance of the devised neural network. For this purpose, available centrifuge experimental results are manipulated to predict the natural period, damping ratio, and structural responses. The results revealed the fact that between the examined procedures, the neural network optimised by the genetic algorithm has performed better than the other two approaches in terms of accuracy and computation time, for solving a soil-structure interaction problem.

KEYWORDS:

• Soil-structure interaction
• neural networks
• genetic algorithm
• optimised
• cross validation
• K-fold

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

Some or all data, models, or code that support the findings of this study are available from the corresponding author upon reasonable request.