In shape mining, data from engineering design across different tools and disciplines are integrated and analysed. Unstructured surface meshes serve as meta-design representations on which we base sensitivity analysis, design concept retrieval and learning as well as methods for interaction analysis of heterogeneous engineering design data.
We apply the formal methods of shape mining to passenger car design. Sensitivities and sensitive cluster centers are mapped onto the car shape, which results in a very intuitive visualization. Furthermore, we are able to identify conceptual design rules using tree induction and to create interaction graphs that illustrate the interrelation between spatially decoupled surface areas.
The intensive use of computational tools in development, test, manufacturing, and service has resulted in a tremendous increase of data that is managed in an engineering context. The integration and combined analysis of data from multiple disciplines and the large scale integration of cyber physical systems will increase the efficiency and quality of the overall product lifecycle. Expensive and safety-critical tests e.g. for structural stability or autonomous driving are augmented by computer simulations. The collected data are of high value and their detailed analysis is required for the next generation system design.
Heterogeneous data sets have to be merged in a unified representation. Consistency, security and integrity of data have to be guaranteed. The engineer is the centre of data analytics: new methods for the visualization of high dimensional interactions and dependencies need to be researched.
For more information
L. Graening and B. Sendhoff, “Shape mining: A holistic data mining approach for engineering design”, Adv. Eng. Inform., vol. 28, issue 2, pp. 166-185, 2014.