Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1698588 | Procedia CIRP | 2016 | 6 Pages |
For a given electro-mechanical product, represented using assembly models and function structures, the assembly time (AT) and market value (MV) are influenced by complexity of the product. Given the AT and MV of a set of known products, complexity values can be used to predicted AT and MV for a set of unknown products using an Artificial Neural Network. This paper presents a precision analysis of four prediction models that are a combination of the aforementioned design representations and AT and MV. A sensitivity analysis of the complexity metrics was done using Multiple Linear Regression, and a set of significant metrics was identified. Lastly, a comparison of accuracy and precision for the four prediction models obtained using this set of sensitivity analysis is presented.