To improve the forecast quality (F-CoP[Total]) of the field-MOP:
Select
> > .On the Select training data tab, do the following:
Drag and drop the field quantity LOGSEQV into the Output list.
Drag and drop the five scalar parameters into the Input (selectable) list:
TEMP_env
TEMP_CPU
CTE_PCB_xy
TEMP_MCU
CTE_CPU
Using the default meta model settings, click
to build the field-MOP.Review results in the Field data models window.
When the default value of 10 shapes is used for the random field decomposition, the field-MOP explains 96.9% of the variance observed in the simulated field designs. This limits the achievable F-CoP[Total] values. Ideally, you want to see values closer to 100%.
Re-create the field-MOP using 25 shapes.
With a superposition of 25 shapes, the field-MOP explains 99.4% of the variance observed in the simulated field designs.
Note: Using more shapes, you can boost the average F-CoP[Total] from 86.7% to 88.2%, with the cost of some computation time.
Now re-create the field-MOP again but drag and drop the field quantity Temp into the Output list and use the default meta model settings.