排列序编码的遗传算法性能分析及应用

The Application and Analysis of Genetic Algorithm Based on Permutation Code

以背包问题为模拟实例 ,考察基于排列序编码的演化算法的性能 ,我们发现 ,如果问题的编码不具有结构性 ,此时求解的质量由个体在解空间的分布情形决定 ;在种群规模偏小的情形下 ,交叉算子的作用不容忽视 ,而在大规模种群的情形下 ,交叉算子的作用有限 ,变异算子应该在算法中占主导地位 ,这样能保证算法具有平稳的在线性能 ,可以应用于系数时变的优化问题的求解 ,对演化算法的实际应用具有一定的指导意义。

In this paper, we explain the performance of genetic algorithm based on permutation coding through experiments on Knapsack problems. It can be founded that when the solution codes don′t have constructive property, the algorithm solution quality depends on the distribution of sampled individuals. When the population is small, the crossover operator can't be ignored;on the other case,it has finite impact on the algorithm and the mutation operator should play a significant role in the algorithm. These conclusions can give some advise on the application of genetic algorithm.