EXPERIMENTS WITH NEURAL NETWORKS AS A DESIGN-SUPPORT TOOL FOR COMPLEXENGINEERING SYSTEMS

Many civil engineering problems are based on an understanding of relat ionships between variables. Many of these variables are established Fr om experimental or numerical observations and are defined in terms of algebraic expressions involving the variables. This paper focuses on t he potential applications of neural networks for evaluating the relati onships between these variables and for modelling complex and non-line ar systems. Demonstration of the potential of this approach is illustr ated through three civil engineering examples.