【作者】 侯彩虹；

【导师】 丁永生；

【作者基本信息】 东华大学 ， 管理科学与工程， 2007， 博士

【摘要】 在全球化市场竞争日益激烈的今天,以客户为中心的研发和生产模式成为很多企业发展的核心,感官评估是保证产品质量和挖掘顾客对产品的反应的重要方式,是生产和消费领域内广受关注的问题。为了解决传统的感官评估研究分散而独立的问题,本文从系统的的角度对感官评估进行深入研究与探讨。多源信息融合是对目标对象进行系统分析,并将其中各信息源整合以完成需要的评估和决策而进行的一系列处理过程。在对感官评估进行系统研究时,可以利用信息融合将客观评估、主观评估、评估专家、评估词汇等评估信息融合进行分析,并给出适时完整的评价。本课题将信息融合的一些分析思想、算法和模型,应用到感官评估的研究中,定义了一个智能感官评估融合系统,并在此系统架构的基础上完成了感官评估的多源信息融合方法,最终实现了对感官评估的系统分析和研究。首先,从评估系统的基本概念开始,对感官评估的评估方法、评估内容以及感官评估的应用现状进行了综述,指出了目前发展存在的问题以及将来的发展方向,并进一步指出了进行感官评估研究的实际需求;介绍了信息融合基本原理、概念、方法和应用,为感官评估融合系统的设计和分析提供了技术支撑。其次,从理论基础和实际需求出发,给出了一个两层感官评估结构,定义了面向设计的感官评估和面向市场的感官评估。总结了感官评估的四个主要要素,并以集合的形式进行了描述。基于感官评估的这个分层模型,定义了多评估空间,并完成了多评估空间的数学描述。然后,在感官评估的两层架构基础上,对纺织品智能感官评估系统进行了总体设计,定义了五个模块,并简单地介绍了系统各组成模块的功能以及模块之间的关系。再次,在面向产品设计开发的客观数据评估模块中,采用核聚类对织物的机械性能指标进行融合;并采用免疫聚类对客观数据进行聚类融合,提取此织物的机械性能特性,并对两种不同的方法在用于感官数据聚类上的效果进行比较。接着,在面向产品设计开发的专家评估研究中,对感官评估基础数据的标准化和感官数据的融合问题进行了研究。通过对感官评估基础数据进行标准化处理,为数据分析阶段提供了有效的目标数据;并对顾客感官评估数据进行分析,借助于信息融合算子给出了专家评估的群决策结果。再接着,在顾客和市场研究一层,分析顾客偏好心理,据此设计顾客满意度调查表,提取了顾客偏好心理;并依据专家系统建立了顾客感官偏好与织物类别间的联系;并实现智能感官评估融合系统。最后,对全文研究内容进行了总结,指出研究工作中存在的不足,明确了下一步的研究方向。更多还原

【Abstract】 In the competitive day of global market, the main work of many companies is to order the R&P only around consumer’s preference. Sensory evaluation, as a powerful tool in mining customer’s demands, is often used in quality assurance, and to build customer-oriented and customer-centred market. The sensory evaluation has become the hot point of research. In this situation, we systematically study sensory evaluation to solve the problems and difficulty in classic sensory study. The term of information fusion refers to the process of correlating and associating data from various sources in order for the information to be presented in a single view. In the paper, we generally analyze objective sensory evaluation, subjective evaluation, and evaluation system with several mathematical algorithms, to offer a whole evaluation to satisfy with the requirements of market.In the thesis, we apply concerned theories and mechanisms of multi-source information fusion technology to analyze the sensory evaluation. Our main work includes bringing forward an intelligent information fusion system for sensory evaluation, studying on sensory evaluation fusion algorithms, and finally implementing them.Firstly, we investigate the development of sensory evaluation, including basic conceptions and components of evaluation system, and how the mathematical methods to be used in analyzing sensory evaluation. Moreover, we briefly introduce multi-source information fusion technologies, including its principle, method and application fields. Multi-source information fusion technology is useful for designing and analyzing information fusion models for sensory evaluation.Secondly, we propose 2-level structure of sensory evaluation, then define the Design-Oriented Sensory Evaluation and the Market-Oriented Sensory Evaluation. And, we analyze the inner structure of two types of evaluation and the difference between them. We abstract four essential elements in the context of sensory evaluation space and presented the subdivision of each element by using the format of set expression. Based on it, we put forward the whole model of sensory evaluation fusion system, as well as its sub-models. Then the function of individual model and the relationship among them are explained.Again, in the objective data evaluation model, we use kernel clustering and immune clustering to aggregate physical parameters of fabric, and compare the clustering results.Next, we treat with the sensory date by the model of Sensory Data Aggregated Model, in which a new linguistic 2-tuple fuzzy model based approach is proposed for evaluating fabric. Then, we analyze these data with intelligent technology and use fusion operator to implement aggregating the subjective data of experts. And, the expert panel and evaluation terms are analyzed with these models.After that, according to the designed principles of consumer questionnaire and taking the characteristics of MOSE in mind, we design a questionnaire for fabric hand evaluation to get large numbers of consumer sensory information; and use expert system to build a judgement relationship between sensory evaluation and fabric kind. Based on the above work, we develop the sensory evaluation fusion system, where Java language is used to compile the program, SQL is used to manage the data system, and FrontPage is used to make Web page.Finally, a summary of the thesis is made, and the deficiency in the project and the further development are narrated respectively.更多还原