Design Of ATM Band-width Distribution Dynamically On The Basis Of Neural Network

From the beginning of the middle of the 1980' s, the investigation of neural networks has become popular again in US, Japan and western Europe, In some aspects, such as modules of neural networks, learning algorithm, realization and practice of neural computer, etc, many achievements were obtained. About 100 kinds of modules of networks were put forward. New algorithms and improvements about them made computer more intelligible. Due to their strong abilities of non -linear mapping, error-tolerance, etc, man-made neural networks obtained wide practice in many fields.BP neural network is the widest one among all neural networks. BP algorithm (Back-propagation algorithm) is the most basic and important one among all kinds of neural networks algorithms. It is on the basis of strong theories and has a wide usage, But its shortcomings, that its learning efficiency is low, searching speed is slow and it is easy to enter the local best point, limit its use. Many people-investigate its shortcomings before, but at present some disadvantages still exist.This article puts forward a solution named divide- assemble by deducing the size of BP neural network to overcome entering the local best point, the dividing process is that a big BP neural network is divided into several small BP neural networks, every small BP neural network can study alone, after all small BP neural networks finish their study, We can assemble all these small BP neural networks into the quondam big BP neural networks; On the basis of divide- assemble solution, This article discusses the preprocessing of input species and how to deduce the size of BP neural network further to make it easy to overcome entering the local best point; For the study of every small BP neural network, This article adopts a solution named GDR-GA algorithm, which includes two algorithms. GDRæ¡®A algorithm makes the merits of the two algorithms makeup each other to increase searching speed. Finally, This article discusses the processing of ATMBand-width distribution dynamically.The paper includes the contents as follows: The second chapter introduces the basic theory of BP neural network, Chapter 3 puts forward a solution named divide- assemble, Chapter 4 introduces the preprocessing of input population, Chapter 5 introduces the basic theory of Genetic Algorithms, Chapter 6 discusses GDR-GA Algorithms, 7 part introduces the processing of ATM band-width distribution dynamically and gives several program modules among them. Chapter 8 introduces the interface, running and results about the software system, the last chapter summarizes research work of this dissertation and further researches arc prospected.