Bottleneck Detection of Manufacturing Systems Using Data Driven Method

Abstract

Bottlenecks in a production line have been shown to be one of the main reasons that impede productivity. Correctly and efficiently identifying botdeneck locations can improve the utilisation of finite manufacturing resources, increase the system throughput, and minimize the total cost of production. Current bottleneck detection schemes can be separated into two categories: analytical and simulation-based. For the analytical method, the system performance is assumed to be described by a statistical distribution. Although an analytical model is good at long term prediction, this type of model is not adequate for solving the bottleneck detection problem in the short term. On the other hand, the simulation-based method has disadvantages, such as long development time and decreased flexibility for different production scenarios, which greatly impede its wide implementation. Because of all these problems, a data driven bottleneck detection method has been constructed based on the real-time data from manufacturing systems. Using this new method, bottleneck locations can be identified in both the short term and long term. Furthermore, the proposed data driven bottleneck detection method has been verified using the results from both the analytical and simulation methods.