Optimus Rev 10.5 offers faster visual insight into design dependencies

Deeper insight into dependencies between design variables & simulation response offers greater opportunity for design improvement. The new SOM Plot visualization introduced in Optimus Rev 10.5 generates a color pattern for each input variable and simulation response. By graphically comparing these individual “blueprints”, engineers instantly discover correlations that allow them to fine tune their design optimization strategy. At the same time, SOM Plot evaluation may detect design space regions of specific interest, essentially driving adaptive DOE with a human in the loop.

Deeper understanding through SOM Plots

A self-organizing map (SOM) is a type of artificial neural network that is trained using unsupervised learning. Optimus iteratively generates such a two-dimensional discretized representation of the input space of the training samples. SOM may be considered a non-linear generalization of principal components analysis, as it is representing the total dataset by a limited number of cells in the map.

Correlations found through SOM Plot comparison allow engineers to fine tune the design optimization strategy

Each cell represents a specific design space region and is characterized by a color and a transparency level. The color is a measure of the parameter value, whereas transparency gives some indication on the number of virtual experiments the data is based on. As a result, similar SOM Plot patterns reveal similar parameter behavior and more transparent areas indicate low experiment density.

Based on SOM Plot evaluation, engineers may plan a new DOE for specific regions in the design space. Additional experiments targeted in these interesting regions append valuable information to the dataset that is already available. This illustrates how Optimus enables engineers to better leverage their expertise in adapting the design optimization strategy. Insight gained through SOM Plots plot comparison allows them to better target DOE and speed up optimization to identify the optimum design.