Abstract: Survivability controller is widely adopted into critical information system.A main function of the controller is to give policies which select corresponding sequence of multiple actions in a user-defined action set based on the reward assessment.Therefore，the quality of policy given by the controller determines its ability.The objective of this paper is to find out an efficient solution approach to determine Action Sequence（AS） achieving the maximal rewards.AS is a mixture of Knapsack Problem（KP） and Traveling Salesman Problem（TSP）.This work has developed a methodology based on Genetic Algorithm（GA） for effective solving combination optimization.Special coding and effective genetic operators are designed for this genetic algorithm.Comparing with greedy algorithm simulation-based experimental results demonstrate the validity and practicability of the Genetic Algorithm.

Key words: survivability controller, Action Sequence（AS）, Knapsack Problem（KP）, Traveling Salesman Problem（TSP）, Genetic Algorithm（GA）

摘要： 生存控制器被广泛地应用在关键的信息系统中。生存控制器的一个重要功能是做决策，也就是基于收益评价从用户给出的行动集合中选择相应的行动序列。因此，决策的质量决定了控制器的能力。寻找一个有效地解决方案来确定获得最大收益的行动序列（AS）。AS是一个背包问题（KP）和旅行商问题（TSP）的混合体。以GA有效解决组合优化问题的方法论为基础，针对AS问题设计了特殊的编码和有效的遗传操作。通过与贪心算法进行比较，模拟实验结果证明了遗传算法的有效性和实用性。

关键词: 生存控制器, 行动序列, 背包问题, 旅行商问题, 遗传算法