QQ客服 官方微博 反馈留言

题名	智能建模方法的研究与实现
作者	陈树新
学位类别	工学博士
答辩日期	1992-05-01
授予单位	中国科学院自动化研究所
授予地点	中国科学院自动化研究所
导师	张志方
学位专业	控制理论与控制工程
中文摘要	本文根据神经网络的信息处理能力、拟人学习能力和系统建模过 程对人的知识的依赖，提出了以人工智能理论为基础的B-P神经网络 智能建模。它有别于以往描述系统的参数模型，它是用网络和与网络 结构相对应的分布参数来整体描述一个系统的外观特征、 本文的主要研究工作如下： l> 依据神经网络的非线性映射原理，提出了统一描述静志糸 统、动态系统、线性系统、非线性系统的广义神经网络方法．论证了 神经网络用于动态系统建模的理论及方法。 2> 研究了可用平稳时间序列描述的动态系统的多步预报神经 网络模型。和带趋势项的非平稳过程的广义B—P神经网络建模方法． 提出了一种局部最先逼近理论，井在此基础上实现了多结构神经刚络 多步预报模型的建模方法。还提出了单输入／单输出系统传递函数的 神经网络建模方法和多输入／多输出系统的分解建模方法，给出了利 用B—P神经网络的映射原理来确定系统延时的新方法。 3> 完成了基于知识的神经网络建模系统的设计，利用已经取 得的经验和已经获取的知识，实现了有指导的智能化建模。 本文的研究结果发展了对动态系统建模的理论及方法，对于研究 社会经济、工业生产等各种问题，提出了一套新的完整而有效的理论 分析方法，并对新方法的应用给出了多种类型的应用实例分析和实用 化应用软件．因此本文的研究工作既有较重要的理论价值又有广阔的 应用前景。
英文摘要	In this thesis, the Neural Network Intelligent Modelling method based on artifical intellig, ent theory is presented. It is much different front the conventional parameter model. It is based on the information processing ability, human- like learninging ability of the neural network, and tire relationship between the human's knowledge and the system modelling procedure, it describes the input / output character of systems in distributed parameters which is related to the network. The main research works are as the followings: 1. A generalized neural network method, which is used to describe the static, dynamic, linear and nonlinear systems, is given. It is based on the nonlinear mapping principle. The theoretic method which the neural network is applied to the dynamic system nodelling is proved. 2. The modelling methods of the system which can be described by the time series are studied, includes: (1) Generalized B-P neural network modelling method of stationary process. (2) Generalized B-P neural! network modelling method of nonstationary process with tendency term and multistep forecasting. (3) The local approximating theory of Generalized B-P neural network is put forword. On the basis of this principle, the multistructure and multistep forecasting modelling method is realized. (4) SISO system transfer function modelling method and MIMO system modelling method are studied. (5) A new metliod to determinate thg time delay of the system, based on the B-P neural network mapping principle, is proved. 3. The modelling system bused on knowledge is occomplished. The intelligent modelling system efficiently uses the obtained knowledge and the experience, and has realized intelligent dynamic system modelling. This thesis gives innovation in dynamic system modelling theory and its applied method, and has completed a new modelling method from theory to application. It is an comprehensive and efficier,,t method in researching the social economic, industrial production and other problems etc. In addition, this thesis gives some kinds of example analysis and the applied software. All the researches not only have the important theory value, but also exist widespread application prospect.
语种	中文
其他标识符	269
内容类型	学位论文
源URL	[http://ir.ia.ac.cn/handle/173211/5630]
专题	毕业生_博士学位论文
推荐引用方式 GB/T 7714	陈树新. 智能建模方法的研究与实现[D]. 中国科学院自动化研究所. 中国科学院自动化研究所. 1992.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

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。