几种集成预测模型及其应用-应用数学专业论文.docxVIP

PAGE PAGE IV Abstract Forecasting is the premise of decision and any successful decision can not be separated from the scientific prediction. Forecasting is very complex due to much relative factors. Although there are various forecasting methods, no one is perfect and universally applicable. Exploring new forecasting methods or system has become a research focus in the related fields. As a new type of predicting technology, the core of ensemble forecasting is integrating a number of machines rather than a single one to gain more comprehensive and reliable forecasts. The key of ensemble forecasting technology is to explore mechanisms among various learning or forecasting machines. Research shows that the method can significantly improve the generalization ability of learning systems and the forecasting accuracy. In this paper, based on further study of the grey system theory, BP neural network, rough set theory and support vector machine and comprehensive analysis of the respective advantages and disadvantages, we study how to structure an ensemble forecasting model with them, and integrate the corresponding algorithms in order to achieve ensemble forecasting. The main work and results are the following: The concept and significance of forecast are given, some conventional technologies, as well as several new types of forecasting techniques are reviewed. Some weakness and success of forecasting methods in practical applications are pointed out, the potential ensemble forecasting model based on the grey system theory, neural networks, rough set theory and support vector machine are explored. We construct an ensemble forecasting model (GANN) based on GM (1,1) model and BP neural networks model, which take full advantage of the weakening randomness and dynamic of data of Grey System and the nonlinear mapping characteristics of neural network , take the advantages of both, to improve the forecasting accuracy. Based on rough set theory and artificial neural networks,