关键词: 复杂产品, 方案设计, 偏好差异, 交叉效率

Abstract: Efficiency evaluation of complex product designs emphasizes the selection of the optimal design solution from a set of similar or comparable alternatives under given input and technical conditions. In practice, the evaluation process of complex product designs involves multiple subjects, which are different in knowledge background, personal experience and so on, and make differentiated judgments about the importance orders of the designs. In response to the preference differences in the multiple subjects, scholars have proposed and applied subjective or additional parameters to measure preference differences. It is not difficult to find that those methods reduce the objectivity of the results. Based on this, this paper proposes an efficient method of complex product designs considering the preference differences of the multiple subjects, and applies the data distribution characteristics to portray the preference differences from the analysis perspective of the data set.
Specifically, this study firstly constructs the initial efficiency matrix of complex product designs on the basis of maintaining the advantages of the classical DEA cross-efficiency method; subsequently, the second-order center distance (standard deviation) and the third-order center distance (skewness) in statistics are introduced to portray the subject’s differentiated preferences, and this is used to revise the initial efficiency matrix; finally, the efficiency matrix with preference differences information is assembled by OWA operator to derive the optimal design.
The efficiency evaluation of the designs for the CC series maglev DC inverter centrifugal compressor is carried out by applying the proposed method. Further, two commonly used multi-attribute evaluation methods, SAW and TOPSIS, are applied to evaluate the efficiency of the designs, and the results are compared with those obtained by the proposed method. From the perspective of overall ranking, the changing trends of ranking obtained by the three methods are similar; from the perspective of specific ranking, the efficiency value obtained by the proposed method is more discrete. In short, the proposed method takes into account the preference differences of the subject and increases excellent differentiation of the designs, which makes the decision makers understand better and clarify the differences in the quality of the designs.

Key words: complex products, solution design, preference differences, multiple subjects