Abstract
Soft computing technics have often been used in various engineering fields. A neural network, one of them, has also been noticed as one of powerful approaches that can solve complicated problems. The present research deals with that how to utilize accumulated results being yielded from structural numerical simulations or experimental tests. The neural network has a potentiality to store them without a huge space in the knowledge base systern. The neural network can also interpolate the descrete results with respect to given data, linearly or non-linearly. Furthermore, the availablity for a reasoning engine in the expert system is also investigated. The system would be so available for all designers, when the method to extract easily the digitized certain knowledges was established. As the example, the elastic-plastic buckling behaviors of single layered reticular domes whose behaviors have to be estimated with many structural parameters, are treated. Through several case studies, the performances of the neural network are discussed.
