Structural Reanalysis using a Neural Network-based Iterative Method

Structural design requires a knowledge of the displacements and stress resultants occurring in a structure so that the designer can be satisfied that all functional requirements and design constraints are met. It is usual to base the first analysis on an initial design and then review the design constraints and make any desirable changes in the design data. If we continue this process we encounter a circular operation of repeated design and analysis. The process is handicapped by the dissimilar computational natures of analysis and design making integration difficult. A neural network type of computational device is suitable for continuous updating of the analysis with provision for a wide range of structural design modifications including those affecting geometry, topology, nodes, loads, supports, and material properties. The convenience in reanalysis introduced will generally require more memory in the computer and faster processing compared with the conventional approach. The practicality of the method is justified by the remarkable increases in processing speeds and memory capacities of modern computers.