【作者】 肖良清；

【导师】 乐晓波；

【作者基本信息】 长沙理工大学 ， 计算机软件与理论， 2010， 硕士

【摘要】 并行测试技术是ATS (Auto Test System)在进一步降低测试的时间、削减测试的成本的趋势下新兴的一项技术，它正以不可比拟的优势成为下一代ATS发展的热点。在运用并行测试技术时，要求分析系统执行过程中可能产生的冲突和竞争情况，特别是如何生成并行测试序列一直是一个复杂的、难于优化的NP难题。Petri网作为一种图形化建模工具，可以很直观的表示系统并发、异步等系统常见现象，而引入时间Petri网，可以更便捷地描述实时系统。因此，进一步发展并行测试技术理论和拓宽并行测试应用领域的关键在于：如何建立准确、可靠的并行测试时间Petri网模型，并在所建模型的基础上，探讨并行测试的任务调度算法。本文在探讨时间Petri网构建过程中的约简步骤，并深入分析现今并行测试任务模型不足的基础上，提出了一种基于时间Petri网的并行测试建模方法。为说明该方法的应用步骤，选用一雷达接收机的实例建立相应的时间Petri网，并对其进行了动态性质分析。在此基础上，详细研究了群智能的各类优化算法，结合时间Petri网的特点和现有并行调度算法的局限，提出了一种基于遗传—蚁群算法的时间Petri网变迁序列求解算法，针对一雷达接收机的具体实例，快速地求得了最优调度方案。仿真实验表明，与现有研究成果相比，本文所提出的算法效率更高，具有重要的实用价值。更多还原

【Abstract】 Parallel test technique is a new technique arising under the tendency of reducing test time and lowing test cost, which is becoming hot spot of the next ATS for unparalleled advantage. The parallel test system requires the analysis of competition and conflict in run time, Worse more, the optimized parallel test task scheduling sequence has been a complicated and difficult NP problem.Petri net, as a graphical modeling tool, which can represent concurrence, asynchrony in intuitive way, has been widely used in various fields. At the same time, the introduction of Timed Petri net can be more convenient to describe the real-time system. Therefore, the key of further development of the theory of parallel test technology, and broaden the application of parallel test is how to establish an accurate and reliable Timed Petri net model for parallel test, and to study parallel test task scheduling algorithm on the basis of the model activity.In this article, based on discussing the procedure of simplifying a TPN model and analyzing the shortage of current parallel test model, proposed a modeling method for parallel test based on Timed Petri net. In order to explain the method for application of steps, used an instance of radar receiver to establish the corresponding Timed Petri net, also studied the dynamic nature of the model. On this basis, a detailed study of various optimization algorithms of swarm intelligence combined the characteristics of Timed Petri net with the limitations of existing parallel scheduling algorithms, an algorithm of exploring transition sequence of Timed Petri net based on genetic-ant colony algorithm is originally proposed. The optimal scheduling can be found in a very short period of time was presented in terms of an example on a radar receiver. Simulated experiment shows that compared with the existing research results, the given algorithm has important practical value for more efficiency.更多还原