Finding and Accepting

This thesis presents the research and development of two assembly time estimation techniques for the automotive industry. Both of these techniques are capable of being implemented early in the product life-cycle of a vehicle and require information available in this phase. Currently, assembly time estimation is performed after production has begun and is a tedious manual process requiring hours of motion-time method studies for each assembly activity. The techniques proposed in this thesis can both be performed prior to the start of production and are capable of being performed automatically after the initial mapping. The first assembly time estimation technique maps the complexity of the product connectivity graph to an assembly time. Twenty-nine metrics have been proposed for the complexity measurement of product connectivity graphs. These twenty-nine metrics are mapped to assembly times through the use of artificial neural networks. One-hundred eighty-nine different neural network architectures are evaluated to determine the most appropriate for the mapping. The application of the neural network and the resultin