ABSTRACT
A typical computational mechanics problem, such as finite-element analysis of solids and structural systems, includes the following tasks: discretizing the geometry of the system with a finite-element mesh; specifying the constitutive behavior of the materials in the system; applying the boundary conditions; applying the stimuli, such as forces; and, performing the analysis. In most cases, specifying the constitutive behavior of the materials is the most difficult part of the task. This is where the neural networks can play an important role. Application of neural networks in computational mechanics started in the late 1980s and the early 1990s in material modeling (Ghaboussi et al., 1990, 1991; Ghaboussi and Wu, 1998). In this chapter, we will discuss the application of neural networks in modeling of constitutive behavior of materials (Ghaboussi, 2001). First, some fundamentals of constitutive modeling are described.
