节点文献

基于方法集的组合评价计算机支持系统研究

Study on Combination Evaluation Computer Support System Based on Method Set

【作者】 李美娟

【导师】 陈国宏;

【作者基本信息】 福州大学 , 技术经济及管理, 2004, 硕士

【摘要】 基于方法集的组合评价计算机支持系统的研究,不仅从方法论角度对完善评价理论有积极意义,同时对综合评价方法的推广应用也有重要意义。目前国内外关于综合评价方法的研究,存在两个突出问题:第一,多方法评价结论的非一致性问题;第二,理论研究与实际应用的脱节问题。而在评价支持系统研究方面,目前虽然已有一些评价软件,但都是分散的、孤立的,关于组合评价目前尚没有支持软件。本研究旨在为解决上述两大问题寻找有效途径。本研究的主要内容及成果可归结如下: 通过对几种常用的指标标准化方法进行对比分析,提出一种较为合理的指标标准化方法——改进的归一化法,旨在扬长避短。 研究单一方法评价结论的漂移性,以便为组合评价提供实验依据。把现实评价问题分为三大类分别进行仿真实验,利用 MATLAB 生成待评价对象,利用不同方法进行评价,把评价结果进行标准化处理,并判别评价方法的相容性,以相容方法的平均值为参照系,求各评价方法的漂移度。 通过计算机模拟实验对组合评价结论的收敛性进行研究,以便解决多方法评价结论非一致性问题。本实验分别运用大量的随机模拟数据和实例数据对不同类型评价方法的组合和再组合进行计算机模拟,实验结果表明组合评价具有很好的收敛性,经过若干次组合后,几种组合评价结论能迅速趋于一致,从而得到一致性的评价结果。 在计算机评价系统需求分析,总体方案设计,以及对指标标准化方法,确定权重方法,单一综合评价方法,确定相容方法集模型,不同方法漂移性测度模型,组合评价方法及组合评价收敛性进行介绍和研究的基础上开发一个基于方法集的组合评价计算机支持系统,解决综合评价研究领域亟待解决的两个问题。本系统针对不同用户不同需要提供三种不同的处理方式及相应用户界面。系统实现,以 Delphi7.0 为主,同时辅以 MATLAB6.5GUI 实现人机界面;以 MATLAB6.5 编写方法模型,编写成 M 文件,数据保存在 MATLAB 数据文件中。

【Abstract】 Study on Computer Support System for combination evaluation based onmethod set, will not only help perfect evaluation theory from methodology, butalso promote expansion of the application of comprehensive evaluation method.There are two prominent problems existing in the studies of comprehensiveevaluation both at home and abroad now: the first one is the inconsistency amongthe evaluation conclusions drawn by different evaluation methods; the second oneis the disjointedness between theoretical research and practical application.Moreover, as to the study on support system for evaluation, there are someevaluation softwares, but they are dispersed and isolated. There isn’t supportsoftware on combination evaluation. This study is intended to find an effectiveway to resolve two above-mentioned problems. The main contents and results canbe summarized up as follows: A more reasonable indicator standardization method, called as modifiednormalization method, is proposed to make full use of advantages and preventdisadvantages by comparing and analyzing the frequently-employed indicatorstandardization methods. The drift about the evaluation conclusion of single method is studied inorder to offer the experiment basis for combination evaluation. Firstly, theevaluation problems are divided into 3 categories. Then evaluation objects that aregenerated by Matlab6.5 are evaluated by using different methods. Thirdly, thecompatibility of the comprehensive evaluation methods is identified after thestandardization of the evaluation results. Finally the average of the compatiblemethods is taken as reference frame to quantitatively measure the drifting degreeof each method. Convergence of combination evaluation is studied through simulationexperiment to resolve inconsistency among the evaluation conclusions drawn bydifferent evaluation methods. In this experiment, stochastic simulated data andpractical examples data are used respectively to simulate the combinationevaluation and re-combination evaluation of different kind methods. The resultsindicate that combination evaluation converges very well. After several times ofcombination, combination evaluation results can reach consistency rapidly, thusconsistent evaluation conclusions are obtained. Based on the introduction of computer evaluation system requirementsanalysis, general design, indicator standardization methods, the methods of weightdetermination, single comprehensive evaluation methods, the model fordetermination of compatible method set, the model for measurement of drift aboutthe evaluation conclusions of single methods, combination evaluation methods 1<WP=4>Abstractand convergence of combination evaluation, a computer support system forcombination evaluation based on method set is developed to resolve two urgentproblems existing in the comprehensive evaluation research field. This systemoffers three different process models and corresponding user interfaces. Thesystem is mainly programmed using Delphi7.0, with the aid of MATLAB6.5, toimplement human-computer interface. The methods and models are composed asM-file by MATLAB6.5, and the data are saved in MATLAB data file.

  • 【网络出版投稿人】 福州大学
  • 【网络出版年期】2004年 04期
  • 【分类号】TP399
  • 【被引频次】5
  • 【下载频次】277
节点文献中: 

本文链接的文献网络图示:

本文的引文网络