Optimization Method of Product Green Design Scheme Based on Attribute Mapping

Lei Zhang, Kai Yang, Cheng Zhang, Qin Xu, Zhao XikunJournal of Mechanical Engineering 2019 · DOI

Abstract: 针对产品绿色设计方案受多因素耦合影响,不易对其优化的问题,提出基于属性映射的产品绿色设计方案优化方法。基于产品功能结构分解模型,建立产品模块实例库,采用正交试验设计安排样本空间,基于支持矢量回归(Support vector regression,SVR)理论,对优化变量与优化目标间的映射关系进行拟合,进而建立产品设计方案多绿色属性与多模块功能实例之间的映射关系;采用非支配排序遗传算法(Non-dominated sorting genetic algorithm II,NSGA-II),选取绿色属性最优的产品设计方案作为目标,对离散变量进行多目标优化,从众多可能解中快速、准确的寻出Pareto解集,从而生成产品绿色设计优化方案。以喷墨打印机产品绿色设计方案优化为例,验证了所提方法的可行性和有效性。

Introduction

Sustainable product design is increasingly critical in modern manufacturing, yet navigating the complex trade-offs between environmental attributes remains a significant challenge. This research introduces a robust framework for optimizing green product design schemes during the early conceptual phase. By integrating machine learning with multi-objective optimization, the study provides a systematic way to identify configurations that balance recyclability, material impact, and disassembly ease.

The Challenge of Green Design

Designing "green" products is inherently difficult because environmental attributes are often coupled and conflicting. For instance, a material chosen for high recyclability might complicate disassembly, or a low-energy component might involve a high-impact manufacturing process. In modular design, where products are assembled from a library of components, the number of possible combinations grows exponentially—a phenomenon known as "combinatorial explosion." Traditional trial-and-error methods or expert intuition often fail to find the optimal balance across multiple metrics. There is a clear need for a mathematical approach to map discrete design choices to continuous environmental outcomes.

Methodology

The proposed optimization workflow addresses these challenges through a four-stage process:

  1. Functional Decomposition & Library Building: The product is decomposed into functional modules (e.g., for a printer: paper feed, print head, control unit). Each module has several "instances" (specific versions) stored in a library.
  2. Sample Space via Orthogonal Design: To avoid the computational cost of testing every combination, an orthogonal experimental design is used to select a representative subset of design schemes.
  3. Attribute Mapping via SVR: Support Vector Regression (SVR) is employed to create a mapping function. SVR is particularly effective for small datasets and can model the non-linear relationships between discrete module selections and continuous "green scores."
Mapping relationship between product design attributes and module instances using SVR.
  1. Multi-Objective Optimization via NSGA-II: Finally, the Non-dominated Sorting Genetic Algorithm II (NSGA-II) searches the discrete space of module combinations to find the Pareto front—a set of optimal designs where no single attribute can be improved without degrading another.

Case Study: Inkjet Printer

The methodology was validated using a commercial inkjet printer case study. The printer was broken down into 7 functional modules, each with 4 possible instances, resulting in 16,384 possible design schemes. Training data consisted of 32 schemes selected via an $L_{32}$ orthogonal array, with 10 additional random schemes used for testing. Expert evaluations and historical data provided ground truth scores for disassembly performance, recovery/recyclability, and material environmental impact.

Results and Discussion

The SVR mapping model demonstrated high predictive accuracy. The Root Mean Square Error (RMSE) for predicting green attributes was consistently low: 0.696 for disassembly, 0.465 for recovery, and 0.628 for material impact.

Comparison of predicted vs. experimental values for recovery performance training samples.

The NSGA-II algorithm successfully converged after 600 iterations, providing a clear 3D Pareto front. This allows designers to select schemes based on specific priorities, such as prioritizing recyclability over disassembly ease for specific markets. The case study confirmed that the method is computationally efficient enough for real-world engineering workflows.

Conclusion

This research successfully bridges the gap between discrete design choices and continuous environmental performance metrics. While the current model relies on expert scoring for initial training, it provides a powerful tool for conceptual design. Future work may incorporate economic factors and market demand directly into the optimization loop to ensure that the most sustainable designs are also commercially viable.

Summary updated

Abstract

摘要: 针对产品绿色设计方案受多因素耦合影响,不易对其优化的问题,提出基于属性映射的产品绿色设计方案优化方法。基于产品功能结构分解模型,建立产品模块实例库,采用正交试验设计安排样本空间,基于支持矢量回归(Support vector regression,SVR)理论,对优化变量与优化目标间的映射关系进行拟合,进而建立产品设计方案多绿色属性与多模块功能实例之间的映射关系;采用非支配排序遗传算法(Non-dominated sorting genetic algorithm II,NSGA-II),选取绿色属性最优的产品设计方案作为目标,对离散变量进行多目标优化,从众多可能解中快速、准确的寻出Pareto解集,从而生成产品绿色设计优化方案。以喷墨打印机产品绿色设计方案优化为例,验证了所提方法的可行性和有效性。
BibTeX
@article{zhang2019optimization,
  author = {Lei Zhang and Kai Yang and Cheng Zhang and Qin Xu and ZHAO Xikun},
  title = {{Optimization Method of Product Green Design Scheme Based on Attribute Mapping}},
  journal = {Journal of Mechanical Engineering},
  year = {2019},
  doi = {10.3901/jme.2019.17.153},
}
Figure