[freebird]

I use CFD with Genetic Algorithm to find optimum airfoil's lift and drag coefficients in viscous flow and Neural Network (NN) for estimation optimum airfoil simultaneously (to decrease time consuming related to CFD analysis).
NN works well for estimation but too time consuming of CFD to gather input information for NN arise this question that:
can we estimate lift and drag coefficient for viscous flow with both of inviscid and viscous flow? e.g. Instead of 200 CFD for the airfoil in viscous flow, use for example 50 CFD in viscous and 100 CFD in inviscid flow to prepare initial input data for NN estimation?
CFD of Inviscid flow is so fast and like this way can help to find optimum airfoil sooner.
any refrence or way would be appreciated.

[FMDenaro]

In no way drag due to the viscous stress can be deduced from an inviscid flow simulation. You can do the approximate correction from some integral method for the BL but I am not sure if this is a good choice for you.
On the other hand, you can expect an acceptable pressure field from the inviscid simulation, provided a low angle of attack is used.