Modeling And Prediction Of Fire Cost Based On Neural Networks

Artificial Neural Network (ANN) theory and methods have been developed rapidly in the past several decades and have been applied in diverse areas, such as engineering, computer science, physics, biology, economics and managements, etc.ANN is a nonlinear system that simulates the information processing method of the human brain, with strong ability to handle nonlinear problems, and adapts to the modeling for such problems as with complex information, dark background knowledge, or indefinite inference rules, for example, the modeling on prediction of fire cost in this thesis.In this thesis, the prediction model of fire cost is built mainly based on the nonlinear approximation property of neural networks, which can be used to investigate our country's fire cost. To establish neural network model, you must have rich program experiences, which hampers the development of neural networks in a way. An available Neural Network Toolbox (NNT) in MATLAB, however, solves the problem.Back Propagation (BP) Neural network and Radial Basis Function (RBF) Neural network are adopted in the model building in this thesis. The learning and training of Neural Network are adopted in this paper after the primitive numbers are handled smoothly. They are used in the prediction of fire cost. This simulation result shows that it is successful for Neural Network to be used in it. So a new method and theory are provided in the prediction of fire cost.The following works are mainly done in this thesis:(1) The theory and phylogeny of the Artificial Neural Networks are introduced, and the fundamental information of BP Neural Networks and RBF Neural Networks are introduced, which provide the foundation for the modeling in the thesis.(2) The fundamental theory of model ing and prediction of fire cost based on neural networks is discussed. The structure and the principle of them are analyzed. How to use the training functions in the MATLAB toolbox to train neural networks to solve specific problems is explained.(3) The prediction models of fire cost built in the thesis are used to predict our country's fire cost. The results show that it is successful for Neural Network to be used in it.This thesis is organized as follows:The first chapter is the preface, in which the theory and phylogeny of the Artificial Neural Networks are introduced, the background, the purpose, the meanings of the study and the innovation point are stated.In Chapter 2 and Chapter 3 some fundamental information of BP Neural Networks and RBF Neural Networks are introduced, which will be used in the 4^th chapter. The structure and the principle of them are analyzed. How to use the training functions in the MATLAB toolbox to train neural networks to solve specific problems is explained.Chapter 4 is the empirical part, the prediction models of fire cost is built using BP Neural network and RBF Neural network. The program is designed in MATLAB language. The simulation and prediction results are achieved.In the last chapter, the conclusion and further work are talked out.