基于多属性群决策的政府采购供应商评价方法研究

发布时间：2018-01-12 23:12

本文关键词：基于多属性群决策的政府采购供应商评价方法研究　出处：《天津大学》2016年博士论文　论文类型：学位论文

【摘要】：随着我国政府采购体制改革的推进,“物有所值”已成为新形势下政府采购工作的原则和目标,供应商评价和选择在政府采购中的重要性也日益显现。政府采购中的供应商评价是一个新的研究领域,如何能够根据政府采购的行政性和法规化特点,研究适用的、科学的供应商评价方法,具有重要的学术价值和实践意义。本论文在对政府采购发展现状、多属性决策方法在供应商评价选择中的应用以及多属性群决策中专家权重确定方法等相关研究分析总结的基础上,对政府采购供应商的评价方法进行了较为深入的研究:针对现有政府采购评价方法中决策客观信息普遍被忽略的问题,提出了一种贯穿评价全过程的,基于动态权重系数修正的供应商评价方法。在该评价方法中,将权重系数的确定过程与供应商的评价进程紧密结合,同时兼顾了属性权重和专家权重的修正,简化了计算过程,并充分利用了初始决策数据所提供的全部有效信息。将该评价方法模型,应用到山东省某地级市一次服务类政府采购的实际案例中,利用统计学中的检验方法,对权重及评价结果的变化情况进行了科学地对比与分析,验证了评价方法的科学性和适用性,可以对当前政府公开招标采购中广泛使用的综合评分法起到很好的完善和补充作用。并在研究方法理论模型的基础上,设计实现了可用于政府采购公开招标活动的供应商评价应用系统,可以方便快捷地实现跨地域的专家评标。针对政府采购评审专家评分的公正性对供应商评价结果具有重要影响的问题,提出了基于专家历史评价数据和评价质量水平的专家权重确定方法:利用SVM分类模型在有限样本情况下的出色的分类能力,对专家历史评价数据进行分类,根据分类结果计算出专家置信度并确定出专家权重。该方法还可用于对专家评分质量进行评价,实现对政府采购评审专家信用记录及履职情况的监管。针对大样本专家群决策评价意见的集结问题,提出了一种切实可行的基于概率区间的大样本专家评价意见集结方法。利用得分区间可能发生的概率代替以往的加权平均值作为评价决策的依据,并在算例分析中借助SPSS软件具体验证和分析了该方法的有效性和可行性。
[Abstract]:With the reform of government procurement system in our country, "value for money" has become the principle and goal of government procurement under the new situation. Supplier evaluation and selection are becoming increasingly important in government procurement. Supplier evaluation in government procurement is a new research field, how to according to the administrative and regulatory characteristics of government procurement. It is of great academic value and practical significance to study the applicable and scientific supplier evaluation methods. The application of multi-attribute decision making method in supplier evaluation and selection, and the expert weight determination method in multi-attribute group decision making are summarized. The evaluation methods of government procurement suppliers are studied deeply: aiming at the problem that objective information of decision-making is generally ignored in the existing evaluation methods of government procurement, a kind of evaluation method is put forward through the whole process of evaluation. Supplier evaluation method based on dynamic weight coefficient correction. In this evaluation method, the determination process of weight coefficient is closely combined with the evaluation process of supplier, and the correction of attribute weight and expert weight is also taken into account. It simplifies the calculation process and makes full use of all the effective information provided by the initial decision data. The evaluation method model is applied to the actual case of a service government procurement in a prefecture-level city in Shandong Province. By using the test method in statistics, the scientific comparison and analysis of the weight and the change of the evaluation result are carried out, which verifies the scientific nature and applicability of the evaluation method. It can perfect and supplement the comprehensive scoring method which is widely used in the public bidding procurement, and on the basis of the theoretical model of the research method. The application system of supplier evaluation which can be used in public bidding of government procurement is designed and implemented. It can easily and quickly realize the expert bid evaluation across regions. Aiming at the problem that the fairness of the evaluation expert score of government procurement has an important impact on the evaluation results of suppliers. The expert weight determination method based on expert historical evaluation data and evaluation quality level is put forward. The expert historical evaluation data is classified by using the excellent classification ability of SVM classification model in limited samples. According to the classification results, the confidence degree of experts is calculated and the weight of experts is determined. The method can also be used to evaluate the quality of expert scoring. To achieve the government procurement review experts credit history and performance of the supervision of the situation. For the large sample of experts decision-making and evaluation of the aggregation of opinion. A feasible method based on probability interval is proposed to aggregate the opinions of large sample experts. The probability of the score interval is used to replace the previous weighted average as the basis of evaluation decision. The validity and feasibility of the method are verified and analyzed by SPSS software.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：F812.45

【相似文献】