About the project

Calculating the manufacturing costs of complex components usually involves a great deal of work. This project used data from a SAP database, AI models and mathematical methods to investigate the possibility of automatically estimating manufacturing costs and other parameters such as the CAD hours or the material required for different liquid distributors. This should simplify the quotation phase and improve estimates.

The experiments show that actual and target values of quotation generation collected in the past can be used in combination with manufacturing data to estimate manufacturing costs and other attributes in advance. The results show that mathematical methods, such as linear regression or decision trees, would be better suited to this problem compared to predictions from artificial neural networks, as they do not require extensive training and provide more accurate predictions and more comprehensible results.

We also found that the installed components have a significant influence on the predictions and that these can vary considerably. Therefore, the influence of different components was analyzed. For example, the CAD hours required could be estimated using the estimated number of screws to be installed. These findings were then used to improve the prediction models.