检索结果-内蒙古大学图书馆

On-line adaptive control of non-linear plants using neural networks with application to liquid level control

INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING 1998年第1期12卷 13-28页

作者： Ng, GW Cook, PA Univ Manchester Inst Sci & Technol Dept EEE Control Syst Ctr Manchester M60 1QD Lancs England

In this paper, we present two on-line adaptive control algorithms for non-linear plants using neural networks. The architecture used is based on the concept of specialized learning, which was first proposed by Psaltis et al.(4) and suffers from two main problems, namely lack of knowledge of the plant Jacobian and slow training speed if the standard backpropagation algorithm is used. Specialized learning has been tested successfully by several researchers using the sign of the plant Jacobian, chosen on the basis of prior qualitative knowledge of the plant. It has also been proposed to calculate the plant Jacobian through a pretrained model. However, if off-line training of the model is not possible and qualitative knowledge of the plant Jacobian is not available, then specialised learning may not be feasible for an on-line neurocontroller. We propose that it is possible to estimate the plant Jacobian through the model on-line for a slowly varying plant. Some analysis for calculating the plant Jacobian on-line is discussed. For rapidly varying plant, we propose using the recursive least-squares (RLS) training algorithm instead of the standard back-propagation (BP) algorithm in the specialised learning architecture. We show that RLS training is faster than the BP algorithm. The proposed fast algorithm is then tested on simulated non-linear plants as well as in a real-time application to a coupled-tanks test rig. (C) 1998 John Whey & Sons, Ltd.

关键词： Distress network applications Neural network adaptive control backpropagation algorithms Liquid level control knowledge IN-LINE level control Jacobian

来源：评论

学校读者我要写书评

暂无评论

Neural network models for a resource allocation problem

引用

IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS 1998年第2期28卷 276-284页

作者： Walczak, S Univ S Alabama Sch Comp & Informat Sci Mobile AL 36688 USA

University admissions and business personnel offices use a limited number of resources to process an ever-increasing quality of student and employment applications. Application systems are further constrained to identify and acquire, in a limited time period,those candidates who are most likely to accept an offer of enrollment or employment. Neural networks are a blew methodology;to this particular domain. Various neural network architectures and learning algorithms are analyzed comparatively to determine the applicability of supervised learning neural networks to the domain problem of personnel resource allocation and to identify optimal learning strategies in this domain. This paper Focuses on multilayer perceptron backpropagation, radial basis function, counterpropagation, general regression, fuzzy ARTMAP, and linear vector quantization neural networks. Each neural network predicts the probability of enrollment and nonenrollment for individual student applicants. backpropagation networks produced the best overall performance. Network performance results are measured by the reduction in the counselor's student case load and corresponding increases in student enrollment. The backpropagation neural networks achieve a 56% reduction in counselor case load.

关键词： Neural networks Resource management Personnel Employment backpropagation algorithms Algorithm design and analysis Supervised learning Multilayer perceptrons Fuzzy neural networks Vector quantization

来源：评论

学校读者我要写书评

暂无评论

Discovery of backpropagation learning rules using genetic programming

Discovery of backpropagation learning rules using genetic pr...

引用

IEEE International Conference on Evolutionary Computation

作者： A. Radi R. Poli School of Computer Science University of Binningham Birmingham UK

The backpropagation learning rule is widespread computational method for training multilayer networks. Unfortunately, backpropagation suffers from several problems. The authors have used genetic programming (GP) to overcome some of these problems and to discover new supervised learning algorithms. A set of such learning algorithms has been compared with the standard backpropagation (SBP) learning algorithm on different problems and has been shown to provide better performances. The study indicates that there exist many supervised learning algorithms better than, but similar to, SEP and that GP can be used to discover them.

关键词： Genetic programming backpropagation algorithms Supervised learning Computer networks Humans Computer science Electronic mail Nonhomogeneous media Artificial neural networks Multi-layer neural network

来源：评论

学校读者我要写书评

暂无评论

Speeding up the convergence of backpropagation networks

Speeding up the convergence of backpropagation networks

引用

IEEE Asia-Pacific Conference on Circuits and Systems

作者： S. Sureerattanan Huynh Ngoc Phien Asian Institute of Technology Thailand

A new algorithm is proposed for speeding up the convergence of backpropagation (BP) networks. This algorithm is obtained by applying the momentum term and adaptive neuron model with temperature momentum term to the Kalman filter (KF) algorithm. It is found that this algorithm can perform satisfactorily in all cases considered. Not only the convergence rate can be improved, but also the sum of squared errors can be further reduced.

关键词： Convergence Neurons Temperature backpropagation algorithms Multilayer perceptrons Multi-layer neural network Equations Electronic mail Nonhomogeneous media Supervised learning

来源：评论

学校读者我要写书评

暂无评论

Fluctuation analysis of a two-layer backpropagation algorithm used for modelling nonlinear memoryless channels

Fluctuation analysis of a two-layer backpropagation algorith...

引用

Asilomar Conference on Signals, Systems & Computers

作者： N.J. Bershad M. Ibnkahla G. Blauwens J. Cools A. Soubrane N. Ponson Department of Electrical and Computer Engineering University of California Irvine CA USA National Polytechnics Institute of Toulouse ENSEEIHT GAPSE Toulouse France Department of Electrotechnology-ESAT Catholic University of Louvain Belgium

Neural networks have previously been used for modelling the nonlinear characteristics of memoryless nonlinear channels using the backpropagation learning (BP) with experimental training data (Ibnkahla et al. 1997). The mean transient and convergence behavior of a simplified two-layer neural network have also been studied (Bershad et al. 1997). The network was trained with zero mean Gaussian data. This paper extends these results to include the effects of the weight fluctuations upon the mean-square-error (MSE). A new methodology is presented which can be extended to other nonlinear learning problems. The new mathematical model is able to predict the MSE learning behavior as a function of the algorithm step size /spl mu/. Linear recursions are derived for the variance and covariance of the weights which depend nonlinearly upon the mean weights. As in linear gradient search problems (LMS, etc.), there exists an optimum /spl mu/ (minimizing the MSE) which is the trade-off between fast learning and small weight fluctuations. Monte Carlo simulations display excellent agreement with the theoretical predictions for various /spl mu/.

关键词： Fluctuations Algorithm design and analysis backpropagation algorithms Neural networks Training data Convergence Mathematical model Search problems Least squares approximation Displays

来源：评论

学校读者我要写书评

暂无评论

Thai printed character recognition by combining inductive logic programming with backpropagation neural network

Thai printed character recognition by combining inductive lo...

引用

IEEE Asia-Pacific Conference on Circuits and Systems

作者： B. Kijsirikul S. Sinthupinyo A. Supanwansa Department of Computer Engineering Chulalongkom University Bangkok Thailand

Several approaches to Thai printed character recognition have been proposed such as comparing heading of character, backpropagation neural network (BNN), fuzzy logic and syntactic method, etc. This paper presents a new approach that combines two learning algorithms, i.e. inductive logic programming (ILP) and backpropagation neural network. After features of character images are extracted, they are employed to construct examples for training an ILP algorithm to learn rules that define the characters. The learned rules are then used to classify the unseen data. However, since some character image, especially the noisy image, may not exactly match with any rule, we then employ BNN for approximately matching the image with the rules. Experimental results demonstrate that the accuracy of rules learned by ILP without the help of BNN is comparable to other methods. Moreover, combining BNN with ILP achieves higher accuracy than the other methods tested in our experiment.

关键词： Character recognition Logic programming Feature extraction Neural networks Fuzzy logic backpropagation algorithms Data mining Testing Noise reduction Computer networks

来源：评论

学校读者我要写书评

暂无评论

Use of dynamic tunneling with backpropagation in training feedforward neural networks

Use of dynamic tunneling with backpropagation in training fe...

引用

IEEE Region 10 International Conference TENCON

作者： P.R. Chowdhury Y.P. Singh R.A. Chansarkar Defence Terrain Research Laboratory New Delhi India IT-BHU Varanasi India Former Adviser Terrain Research Defence Research and Development Organisation New Delhi India

An algorithm is proposed for supervised training in multilayer feedforward neural networks. The proposed algorithm consists of a gradient descent technique with a dynamic tunneling method for global optimization. Relative to the gradient descent technique like the backpropagation algorithm, the proposed technique gives a faster convergence, and the convergence properties of the proposed algorithm is demonstrated through three problems.

关键词： Intelligent networks Neural networks Feedforward neural networks Tunneling backpropagation algorithms Containers Multi-layer neural network Optimization methods Convergence Quantum computing

来源：评论

学校读者我要写书评

暂无评论

Assigning Initial Weights in Feedforward Neural Networks

引用

IFAC Proceedings Volumes 1998年第20期31卷 1053-1058页

作者： Karl Gustafson Guido Sartoris Department of Mathematics University of Colorado Boulder CO 80309 USA and International Solvay Institute University of Brussels CP 231 B-1050 Brussels Belgium Numerical Modelling GmbH Alte Landstrasse 88 CH-8800 Thalwil and Swiss Federal Institute of Technology Inst. for Structural Mechanics GH-8093 Zurich

Departing from the common procedures where rules of thumb are widely used, here a different approach is explored: the initial weights are to be chosen so as to obtain wide directional search during the initial stages of the network's iterations. In this way one increases the chance of correcting a wrong sign in the randomly initialized weight values. Also explored is a related question of choosing an optimal initial gain parameter.

关键词： Neural networks backpropagation algorithms Initial states

来源：评论

学校读者我要写书评

暂无评论

Improved Rolling-Mill Automation by Means of Advanced Control Techniques and Process Models

引用

IFAC Proceedings Volumes 1998年第23期31卷 305-313页

作者： Helge Frank Rudolf Pichler Markus Schoisswohl Wolfgang Staufer VOEST-ALPINE Industrieanlagenbau GmbH Electrical Engineering and Automation Turmstrasse 44 P.O. Box 4 A-4031 Linz Austria

This paper describes the application of advanced control strategies and process models to hot-strip mills. Both controlled system structures and control functions are explained. Finally, improvements resulting from th... 详细信息

关键词： Process Automation Feedforward Kalman Filters Neural Networks Identification algorithms Process models backpropagation algorithms

来源：评论

学校读者我要写书评

暂无评论

Fuzzy Neural Network for State Estimation and Optimal Control in L-Isoleucine Fermentation

引用

IFAC Proceedings Volumes 1998年第8期31卷 301-306页

作者： Xu Wenbo Zhang Xinghua Wuxi University of Light Industry Wuxi Jiangsu P R C 214036

A state pre-estimation model of L-isoleucine fermentation is established by using neural network. The chemotaxis BP algorithm that can effectively overcome the local minimum is adopted for training the net. A neural network that equals a fuzzy system is applied for getting air flow to control dissolved *** membership functions are determined by the net learning. A new control method based on neural networks and fuzzy logic is proposed. This scheme can fully make use of expert knowledge and is robust for uncertainty in model and accurate with noise immunity.

关键词： neural networks fuzzy logic backpropagation algorithms optimal control fermentation processes

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：