The Approximation And Application Based On Double Weighs Neural Network

The artificial neural network is a kind of mathematics model set up on the basis of imitating the neuron structure and mechanism of the human brain. It involves such cross-disciplines as psychology , biology , electron , computer , mathematics , physics ,etc. In recent years, it has developed rapidly and has been widely applied to each field such as pattern recognition , image processing , automatic control, signal processing, economic research ,etc. As what has been studied is deepened constantly, The neural network structure is constantly improvement, its structure function has already more and more approached the true nerve cell function. The Double Weights Neural Network has added new vigor for the research of the neural network, it has caused the extensive concern of the research field of the neural network. It constructed a method of data fitting to Double Weights Neural Network and it alsomakes comparison between the fitting of function and that ofRadial Basis Function to show the superiority of Double Weights Neural Network in the speed and the ability to approximate. This thesis has proved in chapter 3 that Double Weights Neural Network based on single-layer can be approached unanimous continuous function of arbitrarily well in the supremum norm over a compact set and it has explained practical application with its digital instance. This research can be regarded as [43] a kind of expansion studying, In chapter 4, we apply the Double Weights Neural Network to the control field, and have designed a kind of adaptive...