Deterministic Design Optimization (DDO)

Thank you so much for the useful Uqlab software. My question is a general one and related to the RBDO module, is it possible to carry out a deterministic design optimization (DDO) within the module. Appreciate the steps in doing this @moustapha

Hi @aokoro

You can run deterministic optimization using the command

RBDOOpts.Method = 'deterministic'.

This was however not documented in the release so it was not thouroughly tested and valiadated so it may fail to work.

If it does’n’t work, I would recomment directly using matlab optimizers for DDO. Please have a look at fmincon and ga. You can get more info about the set-up for these functions by calling doc fmincon or doc ga in the matlab command line.


Thank you so much @moustapha for the response. Just a further question on DDO. A case where your hard constraints are only the environmental variables and are not part of the cost function; how can this be accounted for in a DDO.

For instance, the cost function contains only design variables and the hard constraint contains only environmental variables.
Cost function C = X1 + X2 + X3
Bounds of design variables X1 (3,5), X2 (4,9) , X3 (0.1 , 0.7)
Hard Constraint:(LSF): X4 +X5 # environment variables (load on the system)
Soft Constraint: X1>X2 ; X2>X3

Hi @aokoro

I am not sure in this context you would have such a case. This wouldn’t be a well-defined problem. If the cost and constraints are independent, there is no point of doing constrained optimization in the first place.
In this example your optimal solution would be (4,4,0.1) assuming first soft constraint is X1>=X2. You have no mean to control the hard-constraint so the system is either in a failed or safe state and there is nothing you could do about it.

I hope this helps.