Abstract: Because the generation solution is unpredictable in evolutionary design, the optimal solution is generally selected by man. An improved genetic algorithm called spreading genetic algorithm is proposed. This approach can solve the problem through selecting the optimal solution by fitness. The feasible solutions under constraint conditions are confirmed by the customer requirements. The individuals belonging to the feasible solutions are gradually spread in the colony through the operation of genetic algorithm, so that the colony is full of feasible solutions that meet the requirements. Then 3D graphs are drawn by using key points selection method and NURBS technique. Thus in the evolutionary design the generation solution can be selected through fitness, and shape visualization is automatically carried out. An example analysis shows that the spreading genetic algorithm has high convergence speed, good on-line and off-line performances.

Key words: Evolutionary design, Genetic algorithm, Non-uniform rational B-spline, Spreading genetic algorithm