主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2025, Vol. 33 ›› Issue (7): 117-127.doi: 10.16381/j.cnki.issn1003-207x.2022.1333

• • 上一篇    

基于CPSO改进的TOPSIS三维空间组合定权投影动态综合评价研究

张侃, 刘思施, 魏华, 余鹏, 梁新()   

  1. 海军工程大学管理工程与装备经济系,湖北 武汉 430033
  • 收稿日期:2022-06-21 修回日期:2023-01-05 出版日期:2025-07-25 发布日期:2025-08-06
  • 通讯作者: 梁新 E-mail:0911031004@nue.edu.cn
  • 基金资助:
    国家社会科学基金军事学项目(2022-SKJJ-C024);部委级(军工)技术基础项目212HJ51001

Dynamic Comprehensive Evaluation Research on TOPSIS Three-dimensional Space Combined Fixed Weight Projection Based on CPSO Improvement

Kan Zhang, Sishi Liu, Hua Wei, Peng Yu, Xin Liang()   

  1. Department of Management Engineering and Equipment Economics,Naval University of Engineering,Hubei 430033,China
  • Received:2022-06-21 Revised:2023-01-05 Online:2025-07-25 Published:2025-08-06
  • Contact: Xin Liang E-mail:0911031004@nue.edu.cn

摘要:

针对同时考虑时间维、对象维和指标维的三维动态评价问题,指出传统TOPSIS(technique for order preference by similarity)方法的应用弊端,提出了一种扩展TOPSIS理论下的三维空间组合定权投影模型,阐明其投影降维原理与算法实现。在此基础上,引入指标维存在非线性映射关系的普适性假设和混沌系统设计思想,分别选择ANP(the analytic network process)结构模型和CPSO(chaos particle swarm optimization)寻优算法来确定指标体系的对象维与时间维权重,测算出最终的三维空间组合权重与评价排序结果。实证研究结果表明,本文提出的动态综合评价模型能够较好地解决三维空间组合定权问题,与PSO(particle swarm optimization)、EGA(elite genetic algorithm)等算法相比,CPSO具有权值全局寻优、搜索速度快、定权方法简便的优点,综合评价结论具有较强的可信度。

关键词: 三维空间组合定权, 混沌粒子群, 理想算法, 投影算法, 网络层次分析

Abstract:

The issue of comprehensive evaluation widely exists in various fields such as society, military, economy, and management, and the conclusions of comprehensive evaluation serve the goals of scientific decision-making. Aiming at the problem of three-dimensional dynamic evaluation considering time dimension, object dimension and index dimension at the same time, it points out the application disadvantages of the traditional TOPSIS method, puts forward a three-dimensional space combined weight projection model under the extended TOPSIS theory, and expounds its projection dimension reduction principle and algorithm implementation in this paper. On this basis, it is assumed that the index dimension has a nonlinear mapping relationship and the design idea of chaotic system are introduced. ANP structure model and CPSO optimization algorithm are selected to determine the object dimension and time dimension weight of the index system, and the final three-dimensional spatial combination weight and evaluation ranking results are calculated. The empirical results show that in terms of accuracy, the optimal fitness of CPSO algorithm is -1.344, better than that of SAG algorithm - 1.032, and avoids the defect of reverse optimization. In terms of convergence, CPSO shows a significant gradient decrease around the 5th generation, with a convergence speed three times that of PSO, while the EGA algorithm had relatively poor stability during the convergence process. Empirical research shows that the dynamic comprehensive evaluation model proposed in this paper can better solve the problem of three-dimensional spatial combination and weighting. Compared to PSO, SGA and EGA,CPSO has the advantages of global optimization of weights, fast search speed, simple weight determination method, and the comprehensive evaluation conclusion has strong credibility.

Key words: three-dimensional space combined weight, chaos particle swarm optimization, technique for order preference by similarity, projection algorithm, the analytic network process

中图分类号: