Quantifying the Impact of Sustainable Product Design Decisions in the Early Design Phase Through Machine Learning
Quantifying the Impact of Sustainable Product Design Decisions in the Early Design Phase Through Machine Learning
复制标题
通过机器学习量化早期设计阶段可持续产品设计决策的影响
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Bryony DuPont
中科院分区：
文献类型：
--
作者：
Addison Wisthoff;Vincenzo Ferrero;T. Huynh;Bryony DuPont
随着越来越多的公司和研究人员有兴趣了解产品设计决策和最终的环境影响之间的关系，提出的方法已经探索满足这一需求。然而，目前有有限的方法可用于早期设计阶段，以帮助量化设计决策对环境的影响。目前的方法，主要是经过审查的生命周期评估（LCA）方法，要求设计师等到设计阶段的后期，当产品的设计是更明确的，或者，设计师辞职，依靠以前的可持续设计经验和经验知识。有一个明确的需要开发的方法，定量告知设计师的环境影响的设计决策在早期的设计阶段（特别是在概念生成），因为这允许重新审查的决定之前，他们成为昂贵的或时间密集的改变。目前的工作建立在以前的研究，涉及可持续设计知识的搜索树的发展，这在早期设计阶段应用，帮助设计师磨练在产品设计决策的影响。为了帮助量化这些设计决策的影响，目前的工作探讨了与每个潜在的设计决策相关的加权系统的发展。在本文中提出的工作的目的是量化的一般环境影响潜在的设计决策对消费产品，通过使用多层感知器神经网络与反向传播训练-机器学习的方法-与37个案例研究产品的生命周期评估影响的产品属性。通过定义LCA数据和产品属性之间的关系，设计师在早期设计阶段将更了解哪些产品属性具有最大的
As more companies and researchers become interested in understanding the relationship between product design decisions and eventual environmental impact, proposed methods have explored meeting this demand. However, there are currently limited methods available for use in the early design phase to help quantify the environmental impact of making design decisions. Current methods, primarily vetted Life Cycle Assessment (LCA) methods, require the designer to wait until later in the design phase, when a product's design is more defined; alternatively, designers are resigned to relying on prior sustainable design experience and empirical knowledge. There is a clear need to develop methods that quantitatively inform designers of the environmental impact of design decisions during the early design phase (particularly during concept generation), as this allows for reexamination of decisions before they become costly or time-intensive to change. The current work builds on previous research involving the development of a search tree of sustainable design knowledge, which, applied during the early design phase, helps designers hone in on the impact of product design decisions. To assist in quantifying the impact of these design decisions, the current work explores the development of a weighting system associated with each potential design decision. The work presented in this paper aims to quantify the general environmental impact potential design decisions have on a consumer product, by using a multi-layer perceptron neural network with back propagation training—a method of machine learning—to relate the life-cycle assessment impact of 37 case study products to product attributes. By defining the relationship between LCA data and product attributes, designers in the early design phase will be more informed of which product attributes have the largest