系统可靠性的多目标优化计算

Multi-objective Optimization of System Reliability

由于多目标优化问题存在多个最优解集合 ,而传统的方法往往将其转化为各目标之加权和 ,然后采用单目标优化技术 ,这种方法存在诸多缺点和脆弱性 .作为一种并行算法 ,遗传算法能很好地解决多目标优化问题 .文中在非劣性分层遗传算法的基础上对遗传算子进行改进 ,首先获得多目标优化问题的非劣解 ,然后通过对系统进行敏感性分析 ,有效地缩小了问题的解空间 .试验对比发现 ,算法的速度和精度得到有效提高 .

The problem of multi-objective optimization has many sets of optimal solutions. The traditional method to solve the problem is to translate the multi-objectiv e into several single objectives and then resolve the invert problem by the sing le-objective optimization technique. Unfortunately, this method has many inheri ted flaws. As a parallel algorithm, genetic algorithm represents excellent capab ility in multi-objective optimization. In this paper, genetic algorithm is empl oyed to the multi-objective optimization of system reliability. Genetic operato rs are modified based on the inpessimum layered genetic algorithm, and the inpes simum solution to the optimization problem is thus obtained. Then, sensitivity a nalysis is carried out to shrink the solution space. By a comparison experiment, it is verified that the speed and precision of the proposed algorithm is effect ively improved.