Popular science description
A principle challenge in production is geometrical deviations of the produced products from the designed product. The functionality and aesthetic qualities of the product can be affected by these deviations besides the additional costs that they impose on the production. These deviations can be reduced by employing higher quality production machines and tools, but this solution increases the production cost and may not be reasonable.
Traditionally the deviations of production have been treated as uncertainties and noises. Therefore, most of the solutions have focused on minimizing the sensitivity of produced products to these noises. However, thanks to robotized production lines, scanning technologies, and machine learning techniques, a new opportunity has arisen to identify and treat these deviations for every product individually.
The geometrical deviation of each part can be scanned by taking several pictures of the parts. This thesis evaluates the means of utilizing the scanned forms to achieve the highest geometrical quality in non-rigid assemblies. These assemblies are ubiquitous in the automotive and aerospace industries. Predicting behaviors of non-rigid assemblies is far more complicated than rigid assemblies due to the variety of factors involved.
Two techniques of selective assembly and individualized locator adjustments are developed and evaluated to be used in individualizing the assembly process of sheet metal assemblies. The results manifest the techniques developed are promising in achieving a significant geometrical quality improvement.
The effects of other production factors including the assembly fixtures are evaluated on the potential improvements. Thus, the improvements can further increase by the specific design of fixtures for individualized assembly processes.
Public thesis defence
2021-03-12 14:00 -- 17:00
Examiner: Rikard Söderberg, IMS
Opponent: Professor Darek Ceglarek, University of Warwick, United Kingdom.