摘要

This paper proposes a process optimization method to improve the dimensional precision of riveted assemblies. The method representation and investigation use an assembly with 1093 rivets yielded from the double curved reflector. Firstly the static and dynamic finite element (FE) models respectively represent the global large-scale assembly and the local riveting process. The dimensional precision is denoted by the root mean square (RMS) of the deformations of the key points selected from the static FE nodes. Then the quantitation between RMS and process parameters equates to the iterative static FE analyses interpolating the dynamic FE analysis result and the possible former static FE analysis result. Finally the integration of the genetic and ant colony algorithms optimizes the process parameters, i.e. the rivet upsetting directions (UDs) and the assembly sequence (AS). Investigation indicates (1) both the rivet UDs and AS are the main RMS influence factors; (2) the proposed method can efficiently optimize the specific process parameters for the large-scale assembly with abundant rivets; and (3) the effective optimization prefers to solve rivet UDs and AS step by step.

全文