论文部分内容阅读
摘要WEB服务体系是Internet最重要的服务之一.随着Internet的迅猛发展,人类对WEB服务器性能品质的要求越来越高.传统的方法在提高WEB服务性能方面已经表现出很大的局限性,而反馈控制方法因其良好的控制理论背景和对动态系统的鲁棒收敛分析机制研究,在解决WEB性能提高方面吸引了越来越多的专家学者的关注.本文以提高Apache WEB服务器的服务性能这一课题为背景,在深入了解WEB服务器的结构、分类、相关参数特性以及所要提高的性能品质等基本信息的基础上,将QOS的性能保障问题转化成控制系统中的反馈控制环,并针对Apache服务器模型的参数多变、存在非线性因素等特点,采用较新的控制技术——多模型自适应控制方法和T-S模糊控制方法,旨在提高WEB服务器的暂态和稳态性能.尤其针对T-S模糊控制方法展开了较深入的先期理论研究.对于模型复杂的不确定时滞互联系统和非互联系统讨论了其鲁棒控制若干问题,其中包括:状态反馈、基于观测器的输出反馈、H∞跟踪控制等诸多问题.同时也将该控制方法用于Apache服务器的建模和性能控制中,MATLAB仿真实验证明,与传统的控制方法相比,上述两种控制方法在提高WEB服务性能方面具有较强的可行性和良好的应用前景.将这两种控制方法用于实际的WEB服务器中以提高其性能品质是我们下一步的研究工作.关键词:QOS(服务质量),SLA(服务管理协议),API(应用程序接口),PDC(平行分布补偿方案),多模型控制,T-S模糊模型系统AbstractWEB service is one of the most important services of Intemet. With the miraculous development of Intemet, the performance of the WEB server is more complex and important than ever. Traditional approaches to providing the performance guarantees become more and more ineffective. Recently, feedback control method for the improvement of WEB performance has attracted more and more interests, due to its good theory foundation and mechanism of robust convergence analysis.In this paper, with the background of Apache WEB server, and based on the basic information of the structure, classify, concemed parameters characteristic and performance guarantee of the WEB server, we map the problem of QOS performance guarantee into the feedback system control loops. For the characteristic of the varying parameters and nonlinear component of the Apache models, we adopt two new methods——Multiple models adaptive control and T-S fuzzy model control to improving the transient response and stability performance. In particular, we develop the anterior theory discussion for some problems of the T-S fuzzy control for the uncertain time-delay interconnected systems and general systems, including state feedback control, observer-based output feedback control, H∞ tracking control and so on. At the same time we apply these two approaches to the model building and performance guarantee of the Apache WEB server. The simulation result in MATLAB illustrates that compared to the traditional control methods, these two approaches hasstrongly feasibility and favorable foreground. Our next research work is to applying the multiple models adaptive control andT-S fuzzy model control to the actual WEB server to improving its performance. Keywords: QOS(Quality of Service),SLA(Service level agreements),API(Application Programming Inerface),PDC (Parallel distributed compensation)Multiple models control,Takagi-Sugeno fuzzy model system