Some Applications of Neural Networks in Structural Engineering

After a brief discussion of neural networks and their potential engineering applications, the author's on-going research in application of neural networks in structural engineering will be discussed. A common task in most engineering problems is the development of mathematical models. Model development with neural networks is achieved through learning. A neural network learns the significant relationships directly from the data. In computational mechanics, neural networks are being used to develop models of material behavior in which neural networks learn the material behavior directly from the experimental results. The author's current research in this area, which will be discussed briefly, is aimed at development of methods for training of neural network material models from the results of structural tests. An other example of the learning replacing mathematical modeling is in the author's research on active control of structures which will also be briefly described. In this research neural networks are trained to learn to control the structure.