【摘要】 物理性能与环保性能是实现可持续产品设计的关键。在可持续产品设计过程中，不同类型工业企业对物理性能与环保性能呈现出异质性偏好，往往采取不同策略方案提升产品设计绩效。本文采用包括工业设计和生态设计两个模块，涵盖多个产品规格和属性的综合指标体系衡量产品设计绩效，构建异质性偏好的网络数据包络分析绩效评价模型，并提出新的线性优化算法，提高计算精度，降低计算复杂度。基于全球重要汽车制造商车辆测试数据库中的关键产品规格、产品属性和污染排放性能相关数据，对30家汽车制造商的940条汽车生产线的设计绩效进行评价。研究发现：异质性偏好并不一定改变产品设计的物理性能与环保性能；通过协调产品设计中物理性能和环保性能之间的绩效关系，可实现产品的最优设计。本文的研究结论，为中国工业企业正确地处理物理性能与环保性能之间关系并合理利用异质性偏好信息提升可持续产品设计绩效，提供有益的理论支持和建议。更多还原

【Abstract】 Physical and environmental performances are the key to sustainable product design. In the process of sustainable product design, different types of industrial enterprises have heterogeneous preferences for physical and environmental performances, and often adopt different strategies to improve the design performance. However, traditional performance evaluation methods have failed to effectively reveal performance changes made by heterogeneity preferences. In order to achieve a balance between the physical and environmental performances of products, it is urgent to make a profound study of an intrinsic mechanism of product design performance changes made by industrial enterprise heterogeneity preferences in theory and practice. The current study defines the product design performance as a comprehensive index to measure multiple product specifications and attributes, including industrial design and ecological design, and develops a network data envelopment analysis model for a performance evaluation under heterogeneous preferences. Considering the unique structure of this model, we propose a new linear optimization algorithm. This method not only improves a computational accuracy of solving nonlinear models, but greatly reduces a computational complexity. By adjusting the preference weights of industrial enterprises and analyzing the performance relationship among the industrial design performance, ecological design performance, and overall performance of product design, we can determine the optimal combination of product specifications and attributes. Based on the vehicle test database of important automobile manufacturers in the world, the design performances of 940 automobile production lines in 30 automobile manufacturers are evaluated by using key product specifications, product attributes and the pollution emission performance. The current study finds: the heterogeneity preference does not necessarily lead to changes in the industrial design performance and ecological design performance, and the variation in the system performance of product design is only caused by a change in preference weights. In certain intervals, small changes in manufacturer preferences lead to significant variations in the industrial and ecological design performance, and these intervals are the key to the sustainable design of the automobile production line. In the absence of any change in the relationship between the performance values of industrial design and ecological design, we need to ensure that the environmental performance meets the requirements for the environmental protection department and then improves the physical performance. When the performance values vary between industrial design and ecological design, we can do the optimal design of both physical and environmental performances simultaneously by coordinating the performance relationship between the two subsystems in product design. This study provides a relatively complete framework for enterprises to coordinate physical and environmental performances in sustainable product design from the perspective of the “heterogeneity preference”, and to provide solutions for industrial enterprises in China to overcome the dilemma of sustainable product design. This study has the following limitations: First, this study defines the system performance of product design as a convex combination of the industrial design performance and ecological design performance based on network DEA models, and demonstrates the variations in the design performance under different decision-making preferences by adjusting preference weights. However, decision-makers may have no preference information for the industrial design performance and environmental design performance. How to determine the reasonable weights is still a valuable issue for research. Second, for the problem of solving nonlinear models, this study proposes an approximate linearization solution method based on the unique structure of performance evaluation models about heterogeneity preferences. Although the numerical examples show the advantages of the new algorithm in the computational accuracy and complexity, the approximation error between the approximate solution and true solution has not been proved in terms of the nonlinear programming theory. Finally, for the issue of product type classification, different types of products are aimed at different consumer groups, which have varying requirements for physical and environmental performances. In the performance evaluation of product design, it is necessary to classify automobile design production lines based on product types. However, this study emphasizes the application of network models in practice, but does not classify them based on specific information of car types(sedan, sportscar, SUV, MPV, new energy vehicles, etc.), which clearly deviates from the actual situations.更多还原