Title: Fixture Layout Optimization in Multi-stage Manufacturing Process by using Genetic Algorithm Selection and Hammersley Sequence Sampling Direct Search   

Sponsor: NSF, NIST-ATP

Collaborators: R. Kumar, J. Kong, Y. Zhou, S. Gogoneni (Dimensional Control Systems)

The previous research in fixture layout design optimization is mostly focused on constraining a single workpeice. There is a lack of comprehensive methodology for the optimization of the multi-fixture production systems.  However, the current research in Steam-of-Variation (SOVA) based on state space methodology allows us to model multi-stage assembly processes. Based on SOVA model, this paper presents the fixture layout optimization technique for 3-dimensional assembly in multi-stage manufacturing that can ensure the predefine  threshold yield level of the production. In addition, the static equilibrium condition of workpiece is also considered to ensure that fixture layout can maintain resting stability of workpiece. The proposed optimization technique employs Genetic Algorithm to initially estimate the design space of interest in the global design space. Then, direct search optimization method is conducted by using Hammersley Sequence Sampling selects. The Hammersley Sequence Sampling method uniformly candidate from the overall design space of interest. Each design sample and fixture layout is assessed for its yield and resting stability performance. The design samples that can satisfy both yield and resting stability requirements are collected for using in the next stage of fixture design process. The proposed optimization methodology was applied to the automotive underbody assembly. The optimal set of fixture layouts illustrates that yield of the process can be improved by 10 percent from the original design without adjusting any tolerance in the process.