一种改进的遗传算法及其在结构优化设计中的应用

An improved genetic algorithm and its application to structural design optimization

通过对遗传算法的二进制编码和实数编码的机理分析,结合2种编码的优点,从工程结构优化实际问题出发,提出了一种可以任意控制离散度的改进实数编码遗传算法.该算法利用实际工程结构问题中对尺寸设计变量精度要求的放松,在编码过程中加上"隐约束",缩小了搜索空间,减少了结构重分析次数,提高了收敛速度.该算法的优点是可以根据实际问题的需要任意选择变量的精度.实例计算表明,该算法对复杂结构的优化设计是有效的.

In combination with the advantages of two kinds of conventional coding, i.e. the binary-coding and realcoding in the genetic algorithm, an improved real coded genetic algorithm with arbitrary discrete set was proposed for engineering structure optimization in practical projects. In consideration of relatively low precision required in size design when the proposed method was applied to optimal design of engineering structures, the adding of implicit constraint during the coding process reduced the search space, decreased the number of structure reanalysis, and accelerated the convergence of calculation. Case studies show that the determination of the precision of variables according to the practical problems makes the algorithm effective for optimal design of complex structures.