Study of Optimum Design of Single layer Reticulated ShellsBased on Niche Genetic Algorithm

摘要

This paper expresses an improved genetic algorithm, and the theory of Niche technique and fuzzy control arernintroduced into the GA, for the purpose of enhancing the population diversity and maintaining of the best parts ofrneach generation. The Simple GA itself has the problem of premature and converging before its time, while lowrnconvergence speed in the anaphase. In order to avoid premature convergence and occurrence of minimal deceptivernproblems, which is caused by the niche technique, fuzzy control is presented for the controlling of the crossoverrnprobability, c P and mutation probability, m P . Above all, that is the new type of algorithm—fuzzy controlled nicherngenetic algorithm (FNGA). The FNGA program is used for structural optimum calculation, and the complex finiternelement software ANSYS is used for structural internal force calculation. The two programs both need the resultrndata of each data, while the results of the two programs are not compatible. Data transferring can not be executedrnautomatically and it causes a big problem. In the paper, data transferring and automatic execution are realizedrnbetween the two programs—FNGA and ANSYS, and the automation level is highly promoted. Finally, a numericalrnexample of single layer reticulated shell is presented to demonstrate reliability and feasibility of the new algorithmrnand further establishes the base of the optimum design of space frame structures.