【作者】 唐万梅；

【导师】 杨新民；

【作者基本信息】 内蒙古大学 ， 应用数学， 2006， 博士

【摘要】 预测是决策的前提，任何成功的决策都离不开科学的预测。预测对象千千万万，预测方法种类繁多，不同的预测对象适用不同的预测方法，所以，对预测方法本身的研究就成为预测研究的重要内容之一。由于传统的预测方法存在其局限性，预测或范围有限，或精度不高。对预测精度不断的追求，预测适应范围的扩大，特别是过去无法预测的领域，就成了现代预测技术追求的目标。 随着BP神经网络、灰色系统理论及支持向量机等新的理论和技术应用于预测领域，预测技术得到了很大的发展。本论文通过对BP神经网络、灰色系统理论和支持向量机的深入研究，针对其中的不足，提出了新的解决方法，建立了新的预测模型。 论文分为九章，前一、二、四章为基本理论和方法；第三、五、六、七章研究新的预测模型；第八章研究多层次综合评价模型；第九章利用因子分析法分析比较了我国西部各地区的人口素质。 第一章介绍了常用预测技术及国内外研究现状，阐述了对预测方法本身研究的重要性，确定了本论文研究的主要内容。第二章介绍人工神经网络的基本模型、工作原理、学习过程等，重点介绍BP神经网络的研究现状及目前研究中存在的问题。第三章以MATLAB提供的神经网络工具箱为基础，设计了一个三层BP神经网络判别模型，提出了一更多还原

【Abstract】 Forecasting is the premise of decision and any successful decision can not be separated from the scientific forecasting. There are various forecasting objects and methods, but different methods are suitable to different objects. Therefore the study of forecasting itself is one of important contents in forecasting research. Because of the limitation and inferior precision of the traditional method, pursuing higher precision and broader range of application has become the goal of the modern forecasting technology, especially in the domain we did not carry on forecasting in the past.With the new theories and techniques of the BP neural network, gray systematic theory and the SVM applying to forecasting domain, forecasting technology has greatly developed. After the thoroughly study of the BP neural network, gray systematic theory and SVM, aiming at their insufficiencies, new methods are raised and new forecasting model is created.This paper is divided into nine chapters. Chapter 1,2 and 4 present basic theories and methods. Chapter 3,5,6 and 7 research new forecasting model. In Chapter 8, multi-hierarchical synthetic evaluation model is given and researched. Each chapter has a brief introduction. Chapter 9 uses the factor analysis to analyse and compare the population quality in the western area of China.Chapter 1 introduces forecasting technology which is in common use and current research status in domestic and abroad. This chapter also gives the profile of the whole paper. Chapter 2 introduces the basic model, operation principles and the study course of the manual neural network, emphasizing on the current status and problems in research of BP neural network. In chapter 3 based on neural network toolbox offered by MATLAB, a judging model of three-layered BP neural network is designed, hence a new method of judging and analysis is raised, and a contrast with the traditional Fisher’s judging method. Chapter 4 introduces the fundamental theories of the gray system. In更多还原