复杂产品是一类结构复杂、零部件集成度高、技术含量高、研发成本高等需要根 据特殊要求定制的大型工业产品或系统，如大型船舶、高端数控设备、航空飞机、高 精尖工业机器人等。复杂产品与现代高科技产业休戚相关，影响一个国家的综合实力 和核心竞争力。由于复杂产品的整个研制周期较长，整体研制难度较高，需要先进的 技术和高超的管理能力，目前产品研制的生产模式主要采用“主制造商-供应商”，即 以主制造商为主要制造角色，负责供应链中的生产组织和系统集成以及设计和确保供 应链的正常运作，供应商则负责提供主制造商所需产品。因此，主制造商对供应商在 研制过程中的绩效表现十分关注，对供应商的绩效评价工作是否科学合理关系着最终 产品的完成质量和进度。 本文考虑到专家对供应商绩效进行评价时所采用的自然语言评价信息具有一定 的犹豫性和模糊性，采用了云模型来定量化专家的自然语言评价信息，并结合复杂产 品自身的特点，提出了一套新的评价方法模型，很大程度上降低了专家评价的模糊性 以及减少了语言信息所带来的误差。 首先在开展供应商的绩效评价工作之前需要构建一套完整的评价指标体系，作为 主制造商监测供应商研制绩效的有效观测点。本文通过广泛的文献分析和研究调研从 工程类、商务类、技术类、服务类和综合类五大方面构建了一套复杂产品供应商研制 绩效评价指标体系。 其次，分析专家在上述指标下对供应商给出的自然语言评价意见（单调型犹豫模 糊语言），通过定义单调型犹豫模糊语言术语集来系统地考虑和挖掘隐藏在显性含义 下的具有专家偏好信息的隐形含义，将其转化为精确的决策数据。在此过程中使用区 间灰数来描述专家对某个语言术语的偏好度，结合ABC分类法和指数云模型构建了专 家单调型犹豫模糊语言定量化表征模型，并对云模型参数展开敏感性分析，然后进行 II 了算例研究，并与其他文献所采用的方法进行比较研究，验证本文所提出的方法更具 优势和创新。 然后，在以上静态评价方法的基础上研究各阶段处的供应商研制绩效，通过不同 的阶段权重来集结各供应商的动态绩效。在此根据AHP分析法相关理论提出犹豫模糊 AHP分析方法，给出各供应商的综合绩效表现对比评价，探寻两类异构信息融合机理。 最后以C919商用飞机为案例研究，定量化专家给出的供应商在各阶段不同指标下的 语言评价信息，并结合供应商在最终阶段两两之间的相对绩效表现所得出的判断矩 阵，以这两类信息之间差异程度最小化为目标构建阶段权重测算模型，最终结合测算 出的阶段权重对各供应商的综合绩效进行优劣排序并选择最优供应商。最后对复杂产 品供应商研制绩效评价工作提出相应的建议和改进措施。 基于指数云模型的特点和运用方式，创新性地提出了将专家以单调犹豫模糊语言 形式给出的绩效评价信息定量化表征的方法，不仅揭示了专家评价意见在显性含义下 的隐藏偏好，还创新性地挖掘出专家对某一特定语言术语偏好程度的犹豫性和模糊 性，进一步降低专家自然语言转换成数据过程中的信息损失，从而提高决策的有效性， 为主制造商的决策工作提供更科学、合理的方法。 关键词：绩效评价；单调犹豫模糊语言术语；指数云模型；阶段权重测算 III RESEARCH ON COMPUTATIONAL METHOD OF COMPLEX PRODUCT SUPPLIERS’ PERFORMACE APPRAISAL BASED ON CLOUD MODEL ABSTRACT Complex products, such as large-scale ships, high-end CNC equipment, aircrafts and high-precision industrial robots, are large-scale industrial products or systems with complex structure, high component integration, high technology content, high R & D cost, and needs to be customized according to special requirements. Complex products are closely related to modern high-tech industry, which directly affect the comprehensive national strength and core competitiveness of a country. The entire development cycle of complex products is relatively long and difficult, requiring advanced technology and superb management capabilities. At present, the production mode mainly adopt "main manufacturer-suppliers", that is, the main manufacturer is responsible for the production organization and system integration in the supply chain and to design and ensure the normal operation of the supply chain, while the suppliers provide the products required by the manufacturer. Therefore, the main manufacturer pays great attention to the performance of the supplier in the development process, and whether the performance appraisal of the suppliers is scientific and reasonable is related to the final assembly progress and the quality of the final product. Considering the hesitation and ambiguity of the natural language appraisal information used by experts when appraising suppliers’ performance, this paper uses the cloud model to quantify the experts’ natural language appraisal information, and combines the characteristics of complex products. A new appraisal method model is proposed, which greatly reduces the ambiguity of expert appraisal and the errors caused by language information. IV First, before starting the performance appraisal of the suppliers, a complete appraisal index system needs to be constructed as an effective observation point for the main manufacturer to monitor the development performance of the suppliers. Through extensive literature analysis and research, a set of performance appraisal indicator system for complex product suppliers is constructed from five major aspects: engineering, business, technology, service and comprehensive. Secondly, this thesis analyzes the experts' natural language appraisal opinions (monotonic hesitant fuzzy language) of suppliers under the above indicators, and systematically considers and excavates the implicit information with expert preference hidden in the explicit meaning by defining the monotonic hesitant fuzzy linguistic set, turning it into precise decision data. In this process, the interval gray number is used to describe the experts’ preference for a certain language term. Combined with ABC classification method and exponential cloud model, the quantitative model is constructed, and the sensitivity analysis of cloud model parameters is carried out. And then take an example compared with other literature’s method to verify that the method has advantages and innovations. Then, based on the above static appraisal method, this paper studies the suppliers’ development performance at each stage, and aggregates the dynamic performance of each supplier through different stage weights. In this process, the hesitant fuzzy AHP analysis method is proposed according to the related theories of AHP analysis for describing comparative appraisal of suppliers’ comprehensive performance, and the fusion mechanism of two kinds of heterogeneous information are explored. Finally, taking the C919 commercial aircraft as a case study, experts’ language appraisal information of each supplier under different indicators at various stages are quantified and the judgment matrix obtained by the supplier's relative performance in the final stage is provided. With the goal of minimizing the difference between the two types of information, a stage weight measurement model is constructed. So combined with the calculated stage weights, the comprehensive performance of suppliers is sorted and the optimal supplier is selected. V Finally, some suggestions and improvement measures are proposed for the complex product suppliers’ performance appraisal and management. Based on the characteristics and application of the exponential cloud model, this thesis innovatively puts forward a method to quantify the performance appraisal information in the form of monotonous hesitation and fuzzy language, which not only reveals the hidden preferences of the expert appraisal opinion under the explicit meaning, but also innovatively excavates the hesitation and fuzziness of the expert decision, further reduce the information loss and improve decision-making effectiveness. And Provide a more scientific and reasonable method for the main manufacturer to make decisions. KEY WORDS：Performance Appraisal；Monotonic Hesitation Fuzzy Language Term；Exponential cloud model；Stage Weight Measurement VI 目 录 摘要 I ABSTRACT ................... III 目 录 ............................ VI 一、绪论 .......................... 1 （一）研究背景 ....... 1 （二）研究意义 ....... 2 （三）研究综述 ....... 4 1．国外研究现状 .... 4 2．国内研究现状 .... 7 3．文献评述 ............ 9 （四）研究方法 ..... 10 （五）创新点 ......... 11 二、复杂产品供应商研制绩效评价指标体系设计 .................. 13 （一）绩效评价指标体系的设计原则及思路 ................