ARTIFICIAL NEURAL-NETWORK AND GENETIC ALGORITHM FOR THE DESIGN OPTIMIZATION OF INDUSTRIAL ROOFS - A COMPARISON

An expert system is applied to the design of industrial roofs. All the codal knowledge are considered in the design. The VP EXPERT shell is used to represent the knowledge in the form of rules. A database conta ining different sections and their properties are used in the design. Five different types of trusses are designed, using expert system and grouping the members into six regions. Three loading cases are conside red in the analysis of the truss. The same five types of trusses are o ptimized using genetic algorithm. Design variables are the area of mem bers and six variables with three bit length for each variable are tak en. The stress and displacement constraints for the three loading case s are considered. The design results using the expert system compare f avorably with results of genetic algorithm.