Novel algorithms for improving manufacturing analytics

点击翻译按钮获取中文摘要

Kinaxis is a company that provides software tools to enable its customers to create, visualize, monitor and execute their supply chain. A supply chain can be thought of as a flow chart representing how all the pieces within a production process fit together from raw materials to finished product. Consider the various components involved in the building of a house as a running example. Supply chains need to identify and encode an ordering between some components. For example, before one can put the roof on a house, the foundation must be laid. Supply chains can be also be used to identify components that can be accomplished concurrently. For example, one can paint the walls in several rooms at the same time. There is no unique supply chain for a given product. As such, one needs to create a supply chain that is realizable and that optimizes certain criteria such as efficiency. The specific problems that will be addressed in this project are directly concerned with these two criteria, i.e. realizability and efficiency. One major barrier to realizability within a supply chain is the existence of cycles. If a supply chain contains the following condition: A must complete before B; B must complete before C and C must complete before A. Such a situation is a directed cycle and would render the supply chain infeasible since none of the components A, B or C can be completed. Kinaxis' software can currently identify cycles, but the process is slow and does not identify all cycles. We will explore various options to speed up the identification of cycles and enumerate all cycles that exist within a supply chain. Efficiency can be achieved in many ways. For this project, the goal will be to increase efficiency of a supply chain by identifying all components that can be completed concurrently or at the same time. Currently, Kinaxis' software cannot identify such components. We will explore various techniques in order to automate the identification of families of components that can be completed concurrently.