Neural network approach for datum selection in CAPP

Process planning is to convert design specifications into manufacturing instructions to make products precisely and economically. Therefore, the tolerance information from design and manufacturing processes must be carefully studied for a computer aided process planning system (CAPP) to generate a feasible and economical process plan. Tight tolerances are usually specified in design when higher accuracy of a feature (such as flatness, roundness, etc.) or a relationship (such as parallelism, perpendicularity, etc.) is required. For such relationships concerning more than one feature, especially for those geometric tolerances with specified datum(s), selection of datum of manufacturing in process planning plays a very important role to make parts within the specifications at the lowest cost. This paper presents a neural network approach for CAPP to select datums for rotational parts based on the shape of the parts and tolerance constraints. Back-propagation algorithm is used and some experiments are conducted. The results are analyzed and further research is proposed.