摘要
借鉴机械优化设计的思想,以总费用为目标函数、以零件参数(标定值和容差等级)为自变量,建立产品性能参数模型,并提出了一种改进的遗传算法。该算法运用了随迭代次数变化的变异概率、自适应变化的交叉概率以及结合赌轮算法的精英选择策略。仿真试验证明,改进的遗传算法不但在收敛速度和搜索能力上优于简单的遗传算法,而且能够较好地避免局部最优,是较好的大规模参数寻优方法。
Learn from the design concept of mechanical optimal, with the total cost as the object function, and the part parameters （ scaling values and tolerance levels ） as independent variables, the parameter model of the product performance is established, and the improved genetic algorithm is proposed. In this algorithm, the mutation probability changing follows the variation of numbers of iterations, the adaptive crossover probability, as well as the elitist selection strategy combining roulette wheel algorithm are adopted. The simulation experiments verify that the improved genetic algorithm is better than the simple genetic algorithm upon convergence speed and searching capability, and effectively avoid local optimization; it is an excellent method for large-scale parameter optimization.
出处
《自动化仪表》
CAS
北大核心
2013年第3期17-20,共4页
Process Automation Instrumentation
基金
华侨大学科研基金资助项目(编号:11HZR02)
关键词
参数模型
优化设计
遗传算法
交叉概率
变异概率
非线性规划
Parameter model Optimal
design Genetic algorithm
Crossover probability
Mutation probability
Nonlinear programming
