Neural Networks in Civil Engineering. II: Systems and Application

TLDR

The first paper of this series explained how artificial neural networks function and identified key issues in their use. This paper demonstrates the versatility of neural networks as a problem‑solving tool in civil engineering. The authors identify neural network characteristics and available systems, evaluate their relevance to various civil engineering problem classes, and demonstrate solution approaches for vector mapping, dynamic‑systems modeling, time‑varying objectives, and optimization using different neural network paradigms. The study shows that neural networks can effectively solve vector mapping, dynamic‑systems modeling, time‑varying objective, and optimization problems in civil engineering.

Abstract

The first paper of this two‐paper series developed an understanding of how artificial neural networks work and explained the primary issues involved in their use. This second paper demonstrates the versatility of neural networks as a problem‐solving tool, and shows how they can be applied to different problems in civil engineering. Initially, the basic characteristics of neural networks and the variety of systems available are identified. The significance of these characteristics in terms of solving different classes of problems is considered. A range of different types of civil engineering problems is examined (in particular, vector mapping, dynamic‐systems modeling, problems in which objectives vary with time, and optimization problems) and approaches to their solutions using different neural network paradigms are demonstrated.