考虑环境社会责任下数据驱动的个性化产品配置优化研究

互联网+、大数据、环境社会责任对个性化产品定制提出了新的挑战。本项目将采用机器学习回归技术来预测个性化产品配置中顾客的偏好需求、采用支持向量机SVM来判别和分类零部件供应商的社会责任。构建以配置总成本、碳排放、社会责任为多目标的个性化产品配置优化模型，并采用基于扩展ε_constraint的多目标优化方法来求解该模型。在此基础上，进一步考虑供应链的不确定性，构建不确定情况下基于场景的多目标随机优化模型；并采用机器学习中的无监督学习算法对随机场景进行聚类，从而有效地减少模型中场景的数目。基于L-shaped方法对该随机规划模型进行求解。本项目的研究成果将促进我国中小型企业个性化定制生产的发展，更好地满足客户个性化的需求；将个性化生产和绿色低碳、社会责任相结合，从而有助于环境保护和制造业的可持续发展。同时，有助于中小型企业在实施个性化生产更好地利用大数据为个性化智能制造服务。

Internet plus, big data, and environment and social responsibility present the challenge for product configuration. Linear regression and SVM learning algorithms will be employed in the study to predict customer preferences and classify the social responsibility of suppliers, respectively. Taking configuration cost, carbon emissions, and social responsibility into account, a multi-objective optimization model is built for product configuration. Further, the uncertainty in a supply chain is considered and the scenario-based stochastic modeling is adopted to build the product configuration optimization model. An unsupervised learning algorithm is utilized to group the scenarios into clusters with the aim of effectively decreasing the amount of scenarios in the model. Finally, L-shaped algorithm is employed to solve the stochastic model. As a result, the ability of personal customization in our country will be greatly enhanced and the sustainability of production can be ensured.