Commercialisation of an inventory management solution for intermittent demand items

计算机化的库存管理应用程序依赖于三个后续阶段的计算：首先，项目分类的基础上的需求结构（例如，“快速”和“快速”）;然后使用适当的预测方法来估计未来的需求;最后，这些预测被输入到一个库存控制模型中，该模型通过考虑库存成本和服务水平来“返回”何时订购以及订购多少注意事项。关于第一阶段，目前在工业中使用的方法利用科学上的“朴素”和/或非直观地吸引人的分类参数。关于间歇性物品（缓慢移动者），目前正在使用次优预测方法;此外，为这一类别采用的库存控制模式远远不合适，因为这些模式通常是为快速移动物品而不是缓慢移动物品制定的。商业软件在实践中提出并实施了一些创新，特别是在“预测阶段”方面。然而，结果远非最佳，因为相关的解决方案仍然没有认识到这三个阶段之间的明显相互作用，从而导致仅仅是边际的改进。多年来，主要研究员在间歇性需求的背景下，在上述所有三个领域做出了一系列贡献。他的研究反映了对库存管理问题的更全面的观点，并在EPSRC的支持下，提出了一种迭代的、理论化的、稳健的方法，其商业化构成了向EPSRC提出这一建议的目的。目前，还没有专门为间歇性项目设计的整体库存管理商业解决方案。所有软件包都试图在其功能中容纳其潜在客户持有的整个股票基础。这种对间歇性项目的缺乏专业化为我们的提议提供了一个独特的卖点。对来自英国皇家空军、英国航空航天、兄弟国际以及汽车和石油行业的数据集进行的实验表明，库存成本降低了25%，库存量高达30亿美元。PI开发了一套专有的分析程序（和Visual Basic、VB应用程序），其中包含处理经验数据和复制相应真实的系统功能所需的所有分析程序，以模拟其在（a）当前使用的方法和（B）拟议修正方法下的性能。与电子表格相关的问题涉及：i）它们的复杂性，以及它们需要大量培训才能实际应用的事实; ii）所有路径和文件名的硬编码，这使得它们的应用程序在不同的PC上不灵活; iii）与安全相关的问题; iv）由于来自客户端的SKU的数量可能容易超过100，000，因此计算性能相当慢（对于行业标准）。后续基金（FoF）将为PI提供足够的资源，以资助开发预原型BETA应用程序，旨在简化其理论的可用性。这个测试版的开发对于他的研究的最终商业实现是必不可少的。对这种解决方案的商业需求及其潜力已经通过大量科学实验的经验数据反复证明。测试版将允许在真实的业务环境中实施新的解决方案并展示其好处。

Computerised inventory management applications rely upon three subsequent stages of computation: first, items are categorised based on the underlying demand structure (say, 'fast' and 'intermittent'); then an appropriate forecasting method is being used to estimate future requirements; finally, the forecasts are input to a stock control model that 'returns' when and how much to order by taking into account inventory cost and service level considerations. Regarding the first stage, current approaches used in industry utilise scientifically 'nave' and/or non-intuitively appealing classification parameters. With regards to the intermittent items (slow movers), sub-optimal forecasting methods are being used; in addition, the stock control models employed for this category are far from appropriate since they generally have been developed for fast rather than slow moving items. Some innovations have been proposed and implemented in practice by commercial software, in particular with respect to the 'forecasting stage'. Nevertheless, the results are far from optimal since the related solutions still fail to appreciate the distinct interaction between the three stages, thus resulting in improvements that are only marginal. The Principal Investigator (PI) has generated over the years a series of contributions in all three above discussed areas in the context of intermittent demand. His research has reflected a more holistic perspective into the problem of inventory management and with the support of EPSRC has produced an iterative, theory informed, robust methodology the commercialisation of which constitutes the purpose of this proposal to EPSRC. Currently, there are not any holistic inventory management commercial solutions that have been designed specifically for intermittent items. All packages attempt to accommodate in their functionality the entire stock base hold by their potential clients. This lack of specialization on intermittent items offers our proposition a unique selling point. Experimentation with datasets from RAF, British Aerospace, Brother International as well as the automotive and oil industries has demonstrated inventory cost reductions of up to 25% in stock bases that represent up to 3 billion. The PI has developed a set of proprietary worksheets (and Visual Basic, VB applications) containing all of the analytical procedures required to process empirical data and replicate the functions of the corresponding real systems for the purpose of simulating their performance under (a) the methods currently utilised and (b) the proposed amended methodology. Problems associated with the spreadsheets relate to: i) their complexity and the fact that they would require a significant amount of training for practical application; ii) the hard-coding of all of the path and file names that renders their application inflexible on different PCs; iii) security related issues; iv) the rather slow (for industry standards) computational performance since the number of SKUs from clients may easily exceed 100,000. The Follow-on Fund (FoF) will provide the PI with sufficient resources to finance the development of a pre-prototype BETA application designed to simplify the usability of his theory. Development of this Beta version is imperative to the eventual commercial implementation of his research. The commercial need for such a solution and its potential has been repeatedly demonstrated though numerous scientific experiments on empirical data. The Beta version will allow the implementation of the new solution and the demonstration of its benefits in a real business environment.