HYBRID GENETIC ALGORITHM FOR OPTIMUM DESIGN OFADVANCED GRID COMPOSITE CIRCULAR CYLINDERS

Based on the characters of optimal design for the composite grid-stiffened st ructures (A GS) , such asmulti-loading cases , multi-const raints , mixed discrete-continuous design variables , etc , a st rategy of optimalanalysis was proposed. Some improvement s about genetic and simplex algorithms were made , and an optimal problem with the buckling and st rain const raint s was t ransformed into a non-const rained optimal problem by the exteriorpenalty function scheme. Based on the foregoing measures , a new hybrid genetic algorithm by conjunction of geneticapproach with simplex algorithm was developed successfully. Comparison of the benchmark result s given by t raditional genetic algorithms showed effectiveness of the present algorithm. Some numerical examples were discussedabout the effect s of st rength const raint s and grid configurations upon the optimal result s. It is concluded that globalbuckling const raint is the key factor for identifying safety of AGS , and the algorithm provided is convenient and efficient .