BP Neural Network Algorithm Improvement And Application Research

Artificial neural network is a hot research field in recent years, involving electronic science and technology, information and communication engineering, computer science and technology, control science and technology, and many other disciplines, its application areas include: modeling, time-series analysis , pattern recognition and control, and constantly expand. Learning algorithm of neural network is always an important research contents in neural network theory research and application field, learning algorithm about the feed-forward neural network has not been a very satisfactory solution in particular. In this paper, BP neural network with the broadest applications and most representative in feed-forward neural network is been as an research object, do some research on BP algorithm, propose two kinds of improved algorithm: FAGABPNN(Factor Analysis Genetic Algorithm Back Propagation Neural Network)algorithm and IAPSOBPNN(Particle Swarm Optimization with Immunity Algorithm Back Propagation Neural Network)combined training algorithm.FAGABPNN algorithm is proposed because BP neural network has complex network structure and low classification ability in complex sample classification, the basic idea is: reduce the dimensions of input samples using Factor Analysis (FA), improve BP algorithm using Genetic Algorithm (GA), then form BP neural network algorithm with strong classification ability.IAPSOBPNN combined training algorithm is proposed because BP neural network is easily trapped into local minima, and has low generalization ability, information-processing mechanism of immune system is used to improve PSO, then form particle swarm optimization with immunity algorithm(IAPSO), particle swarm optimization with immunity algorithm and BP algorithm are combined to form combined training algorithm, then research parameters of BP neural network effectively by it.This paper is divided into seven chapters, the first chapter introduces research background, research contents and research significance about this paper; the second chapter introduces BP neural network; the third chapter introduces genetic algorithm; the forth chapter introduces particle swarm optimization; the fifth chapter does some research aiming at BP neural network has complex network structure and low classification ability in complex sample classification, reduce the dimensions of input samples using factor analysis, improve BP algorithm using genetic algorithm, propose FAGABPNN algorithm; the sixth chapter does some research aiming at BP neural network is easily trapped into local minima, and has low generalization ability, algorithm of combining particle swarm optimization with immunity algorithm with BP algorithm researches parameters of BP neural network, propose IAPSOBPNN combined training algorithm; the seventh chapter summarizes main work about this paper, and prospects further research work.