PFI:AIR - TT: Fast Multi-Echelon Optimization via Grouping

该PFI：AIR技术翻译项目专注于翻译优化和库存细分研究，以满足复杂供应链中快速多级库存优化的需求。 美国有超过2.2万亿美元投资于企业库存;因此，公司需要计算效率高的算法，可以最大限度地减少与管理这些库存资产相关的成本。 多级库存优化根据各个节点或梯队的需求变化确定整个网络中的正确库存水平。它全面考虑整个供应链的库存水平，同时考虑到系统中任何给定点（或梯队）的库存影响。该项目将以软件服务的形式产生原型算法，从而产生多级库存细分分析器。这种多级库存细分分析器具有计算效率高的优化运行时间，接近最佳的解决方案质量，并能够执行快速的“假设”分析。这些功能提供了以下优势：更短的时间找到最佳解决方案，更高质量的解决方案，以及通过“假设”分析降低风险的能力，与目前可用的细分分析技术相比。该项目解决了以下技术差距，因为它从研究发现转化为商业应用。 首先，本研究解决了将细分分析应用于大规模工业数据集时的最佳组大小。 其次，本研究优化了与算法相关的参数设置，以确定最佳设置相对于计算速度和解决方案质量之间的权衡。 最后，本研究进行实验，以确定在细分分析中使用的最佳分组标准。 此外，参与该项目的人员，包括研究生和本科生，将通过实验研究方法，与小企业的互动和软件开发过程获得创新，创业和技术翻译经验。该项目使Invistics公司参与了从研究发现到商业现实的技术翻译工作。

This PFI: AIR Technology Translation project focuses on translating optimization and inventory segmentation research to fill the need for fast multi-echelon inventory optimization within complex supply chains. The U.S. has over $2.2 trillion invested in business inventories; thus, companies require computationally efficient algorithms that can minimize the cost associated with managing these inventory assets. Multi-echelon inventory optimization determines the correct levels of inventory across a network based on demand variability at the various nodes, or echelons. It considers inventory levels holistically across the entire supply chain while taking into account the impact of inventories at any given point (or echelon) in the system.This project will result in prototype algorithms in the form of software services that result in a multi-echelon inventory segmentation analyzer. This multi-echelon inventory segmentation analyzer has computationally efficient optimization run times, near optimal solution quality, and the ability to perform fast "what-if" analysis. These features provide the following advantages: shorter times to find optimal solutions, higher quality solutions, and the ability to mitigate risks through "what-if" analysis when compared to currently available segmentation analytics techniques.This project addresses the following technology gap(s) as it translates from research discovery toward commercial application. First, this research addresses the best group size when applying the segmentation analytics to large-scale industrial datasets. Secondly, this research optimizes the parameter settings associated with the algorithms in order to determine the best settings with respect to the trade-off between computational speed and solution quality. Lastly, this research performs experiments in order to determine the best grouping criteria to use within the segmentation analytics. In addition, personnel involved in this project, including the graduate and undergraduate students, will receive innovation, entrepreneurship, and technology translation experiences through experimental research methods, interaction with a small business, and software development processes. The project engages Invistics Corporation in this technology translation effort from research discovery toward commercial reality.