Abstract:

Complexity, tremendous and interactivity are main characters of system of systems (SoSs).  A  requirement analysis of SoSs faces some questions, such as great uncertainties and huge space of solutions. The heuristic algorithm could settle some NP-hard problems, but efficiency of heuristic algorithm is lower when the complexity of the problem becomes higher. Capabilities solutions of SoSs have their own traits. Each heuristic algorithm is expert in computing different kinds of capabilities solution. This article proposes an algorithm based on intelligent agent by choosing a 3-dimension probabilities matrix. Using the selflearning of agent, the method stores the history experience which is applied to solve such kind of SoSs requirement solutions. The history experience of Agents could be stored in the 3-dimension matrix. When dealing with huge complex SoSs requirement solutions, the Agent can choose the most efficient algorithm to solve the proper problem. This is illustrated with a case study of military SoSs, and the result shows greatly robust and efficient advantages under this context.