【作者】 唐雷；

【导师】 薛福珍； 帅建梅；

【作者基本信息】 中国科学技术大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 近年来,模糊控制的研究不断发展,模糊控制器的多种设计方法也在工业领域获得了广泛应用。同时,人们希望利用更方便的方法来设计出满意的模糊控制器,而且对模糊控制器的控制效果的要求越来越高。因此,模糊控制器的设计优化成为模糊控制的一个研究方面。其中,智能优化算法作为一种有效优化手段,逐渐被引入到模糊控制器的设计优化中,但此项工作仍是一个费时费力的过程。智能优化算法通过模拟自然界的机理来达到优化目的。其中,粒子群算法(Particle Swarm Optimization, PSO)由于其显著的全局寻优能力、局部寻优能力以及鲁棒性得到了足够的重视,在工业生产过程的优化控制中得到广泛应用。而且,有大量的研究旨在进一步提高PSO算法的寻优能力。首先,本文在研究PSO及其改进算法——量子粒子群算法(Quantum Behaved Particle Swarm Optimization, QPSO)的基础上,将免疫算子引入QPSO,提出免疫量子粒子群算法(Quantum Behaved Particle Swarm Optimization–Immune, QPSO-IM),进一步提高算法的寻优能力。多峰值函数的测试表明了本文所提算法与PSO和QPSO相比,在收敛速度,寻优结果以及高维寻优能力上的优越性。然后,本文提出用基于QPSO-IM算法的聚类方法(QPSO-IM Clustering Algorithm, QPSO-IMCA)设计模糊控制器,并提出特殊的编码方案以解决迭代聚类对初始条件较为依赖的问题。本文通过Iris和Glass测试数据集,验证了QPSO-IMCA在聚类分析能力上较基于QPSO算法以及PSO算法的聚类方法更优。基于本文所提算法,本文研发了界面友好的模糊控制器CAD平台,论文介绍了其设计原理以及模糊控制器的设计流程。为进一步提高设计出的模糊控制器的控制效果,本文提出用QPSO-IM算法来优化模糊控制器中的控制决策表。考虑到控制决策表的优化对被控对象模型的依赖,本文同时提出基于QPSO-IM算法的模糊辨识算法(QPSO-IM Fuzzy Identification, QPSO-IMFI)。在提出优化策略以及QPSO-IMFI的基础上,本文将此两项功能集成于模糊控制器CAD平台中,并介绍其设计原理以及使用流程。最后,本文通过单容水箱的液位控制实验以及烤箱的温度控制实验验证了所提方法的有效性及研制开发的模糊控制器CAD平台的实用性。更多还原

【Abstract】 During the past recent years, the research on fuzzy control has been developed a lot, and a varity of fuzzy controller design methods have been used in industry. Meanwhile, people are not only hoping to obtain satified fuzzy controller using easier design methods, but also expecting them to have better control performance. Thus, design and optimization of fuzzy controller has now been developed into a research topic in the fuzzy control research field. Among all the methods to design and optimize the fuzzy controller, intelligent optimization algorithm is a quite efficient one.Intelligent optimization algorithms achieve the optimization goal through imitating the mechanism of the nature. Among them, Particle Swarm Optimization (PSO) has received much attention for its prominent global search ability, local sesarch ability and robustness, and it has been widely used in the industrial production process. Also, numerous researches aim to further improve PSO’s optimization ability.First, this paper has done research on PSO and its improved algorithm, i.e. Quantum Behaved Particle Swarm Optimization (QPSO). On this basis, this paper draws immune operator into QPSO and proposes Quantum Behaved Particle Swarm Optimization–Immune (QPSO-IM), thus further improves QPSO’s optimization ability. Through tests using multiple hump functions, it has been improved that QPSO-IM has strong superiority in convergency speed, optimization results and high-dimension search ability, to either PSO or QPSO.Furthermore, for the design of fuzzy controller, this paper proposes QPSO-IM based Clustering Algorithm (QPSO-IMCA) which is on the basis of QPSO-IM, and also proposes a special coding method that makes the iterative clustering algorithm more independent to its initial state. Through tests using Iris and Glass data set, this paper proves QPSO-IMCA’s superiority in clustering analysis ability to either QPSO or PSO-based methods. Then, on the basis of the proposed clustering algorithm, this paper designs and develops a user-friendly fuzzy controller CAD platform, and then introduces its design principle and the design procedure.To further improve the fuzzy controller’s control peformance, this paper proposes to optimize the fuzzy control table of the fuzzy controller using QPSO-IM. Considering the dependent of the fuzzy control table optimization to the model of the plant, this paper proposes QPSO-IM based Fuzzy Identification (QPSO-IMFI) algorithm. Then, based on the proposed optimization strategy and the analysi of QPSO-IMFI, this paper integrites the fuzzy identification and optimization function into the fuzzy controller CAD platform, and then introduces its design principlen and design procedure.Finally, through the experiments of controlling the single container water tank level and the oven temperature, this paper proves the effectiveness of all the proposed methods and the practicability of the developed fuzzy controller CAD platform.更多还原