Research On Neural Network Prediction

Neural Network (NN) has been deeply appreciated by many scholars. Now it is used with great success in many fields. Prediction is one of the important application fields of NN. Most of the general predicting methods are based on linear analysis, when it comes to non-linear they met many difficulties. While NN is competent for non-linear proceeding for its excellent non-linear character. Predicting methods based on NN extend the space of predicting research.The following are what this paper has done in the field of Neural Network prediction and its application.In the chapter one, the theory of time series and kinds of prediction theory are briefly introduced. It presents the latest development and background of the world's, especially exposes the signification and purpose of Neural Network prediction research. The chapter two discusses the characteristic and development of NN. It declares operational principle and interconnection mode of NN in detail. In the chapter three, it analyses Neural Network prediction research, then presents an improved algorithm of confirm BP Network and an improved algorithm of acceleration convergence on one prediction object.In the chapter four, it firstly overviews the characteristic of chaos and chaotic time series, then discusses the feasibility of chaotic time series on NN. Due to the characteristic of chaotic time series, it presents an improved algorithm to confirm BP Network. It finally gives simulation analysis using improved algorithm on single step and multi-step prediction, and validates the characteristic of chaotic time series and the validity of improved algorithm. In the chapter five, it firstly introduces flash welding, and then discusses the principle of sample set selection. It selects corresponding sample set in virtue of acquiring data and analyses data source. It determines data properties and deals with abnormal data. It presents an improved normalization method according to the character of data source. Finally it builds a quality prediction model online of welding connector in an improved BP algorithm. The simulation result has proved the validity of the method this paper discusses.In the end the paper draws some conclusions. It lists many of problem unsolved and predicts the development of the brand.