Proceedings of the
9th International Conference of Asian Society for Precision Engineering and Nanotechnology (ASPEN2022)
15 – 18 November 2022, Singapore
doi:10.3850/978-981-18-6021-8_OR-18-0310
Optimal Grouping of Random Grouped Assembly for Three-Dimensional Precision
1Department of Mechanical Engineering, National Chung-Hsing University, 145 Xingda Rd., Taichung City 40227, Taiwan
ABSTRACT
Precision is one of the most critical indices of a precise product. Grouped random assembly (GRA) has been propose d to achieve higher precision for batch production. GRA is a method that divides components, with tolerance sorted, into serval groups. Components in each group are then randomly assembled with other components in corresponding group so that higher precision can be achieved. While current GRA deals with one-dimensional dimensional chain, thispaper investigates optimal grouping, in terms of best three-dimensional (3D) precision, for assembly of multiple componets. As there are many combinations of matching groups in GRA, we started with minimum groups of components, whose dimensions were sorted. Allocation of matching groups of different components are then optimized by the GeneticAlgorithm (GA). If the resultant precision is not satisfied, groups of parts are further increased to improve the resultant precision. The illustrated assembly example of four components, with 1000 components for each part, showed that3D errors can be reducedby 56.8% to 62.5% under four groups. If the number of groups is increased to 10, the three-axial errors can be further reduced by 31.3% to 53.3%, indicating that this method can effectively and quickly configure the GRA with the best precision. The main contribution of this research is to develop a method which can quickly and effectively allocate the best GRA grouping, with minimum three-dimensional assembly errors without modification or rework of componets. It solves the bottleneck dealing with multi-dimensional assembly error. It also provides the basis for the best GRA grouping and production planning of precision products.
Keywords: Assembly Precision, Error Stack-up, Grouped Random Assembly, Genetic Algorithm
1Department of Mechanical Engineering, National Chung-Hsing University, 145 Xingda Rd., Taichung City 40227, Taiwan
ABSTRACT
Precision is one of the most critical indices of a precise product. Grouped random assembly (GRA) has been propose d to achieve higher precision for batch production. GRA is a method that divides components, with tolerance sorted, into serval groups. Components in each group are then randomly assembled with other components in corresponding group so that higher precision can be achieved. While current GRA deals with one-dimensional dimensional chain, thispaper investigates optimal grouping, in terms of best three-dimensional (3D) precision, for assembly of multiple componets. As there are many combinations of matching groups in GRA, we started with minimum groups of components, whose dimensions were sorted. Allocation of matching groups of different components are then optimized by the GeneticAlgorithm (GA). If the resultant precision is not satisfied, groups of parts are further increased to improve the resultant precision. The illustrated assembly example of four components, with 1000 components for each part, showed that3D errors can be reducedby 56.8% to 62.5% under four groups. If the number of groups is increased to 10, the three-axial errors can be further reduced by 31.3% to 53.3%, indicating that this method can effectively and quickly configure the GRA with the best precision. The main contribution of this research is to develop a method which can quickly and effectively allocate the best GRA grouping, with minimum three-dimensional assembly errors without modification or rework of componets. It solves the bottleneck dealing with multi-dimensional assembly error. It also provides the basis for the best GRA grouping and production planning of precision products.
Keywords: Assembly Precision, Error Stack-up, Grouped Random Assembly, Genetic Algorithm
