CAJ | 학술논문

为了提高遗传算法对航路规划问题的求解效率,提出了一种约束引导的航路规划遗传算法（CD-GA）.与传统GA不同的是,该算法在优化过程中使用航路节点间的关联约束来实时限定基因值的准确变化范围.为了使染色体与航路的表达方式更加接近,采用定长实数的矩阵编码方式;采用一种分步递归初始化策略生成初始种群,保证其中均是非劣个体;在算法迭代过程中,分别采用一种连续多点分步交叉策略和扰动连续修复变异策略进行交叉和变异,使得算法搜索空间逐步减小,从而加速算法收敛.仿真实验结果表明,该算法能够显著提高遗传算法的全局搜索性能,并且算法收敛速度快,稳定性好.
위료제고유전산법대항로규화문제적구해효솔,제출료일충약속인도적항로규화유전산법（CD-GA）.여전통GA불동적시,해산법재우화과정중사용항로절점간적관련약속래실시한정기인치적준학변화범위.위료사염색체여항로적표체방식경가접근,채용정장실수적구진편마방식;채용일충분보체귀초시화책략생성초시충군,보증기중균시비렬개체;재산법질대과정중,분별채용일충련속다점분보교차책략화우동련속수복변이책략진행교차화변이,사득산법수색공간축보감소,종이가속산법수렴.방진실험결과표명,해산법능구현저제고유전산법적전국수색성능,병차산법수렴속도쾌,은정성호.
To improve the efficiency of path planning solved by Genetic Algorithm(GA), a CD-GA (Constraint Driven GA) for path planning was proposed. Compared with traditional GA,the association constraints among path nodes were applied to immediately limit accurate variation-rang of genetic value in the process of optimization by the algorithm. To make the chromosome be close to the characteristics of path, fixed-length real-number matrix encoding method was applied. The initial populations were generated by an initialization strategy with sequential recursion to ensure the individuals to be superior. In the iteration process of algorithm, the crossover and mutation were carried out using successive-multipoints sequential crossover strategy and disturbance-sequential-restoration mutation strategy respectively, and the search space of the algorithm decreased gradually,thereby the convergence of the algorithm was accelerated. The result of simulation test shows that the proposed algorithm can improve the overall searching ability of GA obviously,and the algorithm has quick convergence and good stability.