人机交互的遗传算法及其在约束布局优化中的应用

复杂工程布局（如卫星舱布局）方案设计问题在理论上属带性能约束的布局优化问题（NPC问题），很难求解，因而目前研究得尚少。为解决此类问题，该文提出了一种人机交互的遗传算法。该算法首先将人工设计的个体作为染色体群体的组成部分，然后在遗传运算中，把人工适时设计的新个体加入到染色体群体中，以代替群体中的较差个体。从而构成人机交互的遗传算法，这样可以充分发挥人和计算机各自的特长。文后通过3个算例（其中一个为作者构造的已知最优解的算例）的数值计算，验证了该算法的可行性和有效性。

Human-Computer Interactive Genetic Algorithm and Its Application to Constrained Layout Optimization

Scheme design and packing problems with behavioral constraints (inertia, balance, stability and vibration etc.) and constrained layout optimization problems belong to NPC. They are concerned more and more in recent years and arise in a variety of application areas such as the layout design of spacecraft, shipping, vehicle, machine tool, and robot etc. Taking the layout design of artificial satellite cabins as background, a human computer interactive genetic algorithm is proposed for solving the two dimensional constrained layout optimization problems in this paper. Firstly, the algorithm makes the artificial individuals (AIs) as a part of the chromosome population and divides the population into some subgroups. Secondly, each subgroup, whose values of crossover and mutation operators are different from other subgroups, operates independently. After copy, crossover, and mutation operations, the better individual in each subgroup is transferred to adjacent subgroups. Thirdly, human expert examines the locally optimal solution that can be obtained through the loops of many generations and designs new AIs with visualized technology. By the way, he finds these new AIs according to the value of fitness function, then to many actual engineering factors. Finally these new AIs are copied in order to ensure that they play an important role in chromosome population. Then they are placed into the chromosome population to replace the worse individuals. The four steps mentioned above are repeated until the human expert finds the satisfied solution. And then human computer interactive genetic algorithm is formed to solve the practical engineer layout problems and the specialties of human and computer can be exerted to the utmost respectively. The results of three examples(one of them is proposed by the authors, and its optimal solution is known)show that this algorithm is feasible and efficient. The human computer interactive genetic algorithm provides an effective approach for the practically complex layout problems in engineering, as well as for the problems of maneurerability of human computer interaction.