舰船 C4I 集成的若干实时元素特性研究
文献类型：学位论文
| 作者 | 雷飞 |
| 学位类别 | 工学博士 |
| 答辩日期 | 2005-05-01 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 邹益仁 |
| 关键词 | 多 Agent 实时系统集成 实时控制 多机调度 协同设计 multi-agent integration of real-time system real-time control |
| 其他题名 | A Dissertation in Control Theory and Control Engineering |
| 学位专业 | 控制理论与控制工程 |
| 中文摘要 | 本文以海军舰船 C4I 系统集成应用中的元素特性分析为背景，针对其中若干实时元素特性展开一系列的模型与算法的研究，这些实时元素包括：实时 Agent 模型、具有切换时间的多机调度以及实时控制任务的协同设计与反馈调度问题。以上研究不仅给舰船 C4I 系统的集成应用提供方法指导，而且丰富了实时系统集成的研究内容。论文的主要内容和创新之处如下： 1．基于 Agent 的 C4I 集成框架及实时 Agent 模型研究。对舰船 C4I 系统进行功能层次划分，构造了一个基于多 Agent 的分布式集成框架，提高了 C4I 系统的灵活性和协调性。为了提高 Agent 的实时性能，引入实例检索机制，对 Agent 的模型进行改进，并建立 Agent间的协作检索和学习机制，以增加实例检索的准确性和成功率。 2．具有可变切换时间的多机实时调度问题研究。结合 C4I 系统集成应用的需求，研究具有切换时间的多机调度问题，首先对问题时间约束进行了简化，建立了可变切换时间任务多机调度问题的优化模型。文中进一步采用遗传算法和模拟退火混合优化算法对该问题进行求解，详细设计了算法的编码、初始群体产生、交叉、变异等内容。仿真结果表明该模型和算法是可行的。 3．实时控制任务的设计与调度研究。研究控制任务在实时环境中的性能损失问题，首先讨论处理器利用率对控制性能的影响，提出实时控制任务利用率的动态调整算法。其次对实时调度引起的抖动和延迟对控制性能的影响进行研究，建立相关模型，并提出对性能损失进行补偿的静态和动态方法，对提高控制任务在实时环境中的性能具有很好的效果。4．实时控制任务的模糊调度研究。分析了实时控制任务的控制性能在不同控制阶段与利用率需求间的关系，提出一种实时控制任务的模糊反馈调度系统。模糊控制器通过监测实时控制任务状态变量的误差及其变化率，查询模糊决策表决定任务的优先级，反馈调度器根据优先级分配任务的利用率。仿真结果表明在处理器资源有限时，该方法能有效改善实时控制任务的控制性能。 |
| 英文摘要 | Based on the element properties analysis in integration of the naval warship C4I system, this dissertation did some model and algorithm research of real-time elements which include real-time agent model, multi-machine scheduling (MMS) on the tasks with variable switch time, integrated design of control and scheduling and feedback scheduling of real-time control tasks. These researches are practically significant and extended the field of integration of real-time systems. Main works and innovations of this dissertation are as follows. Firstly, the dissertation presents a distributed integration frame of C4I system based on multi-agent system, and the system was modularized with agent granularity. By introducing the agent technology into it, the C4I system’s flexibility and harmony was improved. To enhance real-time performance of the agent, case base was added in agent model, further more, learning mechanism and co-search method was introduced. Secondly, the model of multi-machine scheduling on the tasks with variable switch time was proposed with simplified time constraints. GASA hybrid optimal strategy was adopted to solve this problem and some implement problem such as individual coding, initial population generation, crossover and mutation was described in detail. The simulation results show that the model and its algorithm are feasible. Thirdly, the performance of control tasks will degrade in real-time circumstance. The relation between control performance and sample period was investigated and a dynamic priority allocation algorithm was proposed. Real-time scheduling causes sampling jitter and increases computing delay, which will further cause performance degradation. Both static and dynamic compensation methods were proposed to compensate performance cost, and these methods are efficient for improving controller performance in real-time applications. Finally, the relationship between the performance index and the processor utilize of the real-time control tasks in different control phase was investigated, further more, a fuzzy feedback scheduling system was proposed. By querying fuzzy control decision-making table (FDT), fuzzy controller assigns priority to real-time control tasks dynamically according to the current error and , and processor utilize was assigned by feedback scheduler according to the priority. Simulation results verified that the fuzzy feedback scheduling system can improve the overall system performance under limited computing resources. |
| 语种 | 中文 |
| 其他标识符 | 200218014603163 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/5860]  |
| 专题 | 毕业生_博士学位论文 |
| 推荐引用方式 GB/T 7714 | 雷飞. 舰船 C4I 集成的若干实时元素特性研究[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2005. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。