Hello experts
I am dealing with an optimization problem in which the algorithm will choose the section of the column (RC structure) between a lower and an upper bound (e.g., LB and UB).
How can I ask the algorithm to change the cross section if the condition of the period isn't satisfied?
N.B: the period value is retrieved from the sap2000 software using the API, and it has no relation with the design variables to be set as a constriant.
Any one knows how to do it please?
Hello Djedoui Nassim,
to help you I need a bit more information. When is the period value retrieved? Is it called after every evaluation or every now and then? On what does the period depend if doesn't depend on design variables?
Miha Ravber
Thanks for your reaction,
Let's say the algorithm chooses a section for the columns as (30x30)cm, here the period is verified whether to continue with this iteration or not if the condition of the period is satisfied the optimization can move on otherwise, another section should be selected.
The period is in relation to the design variables but not in a simple mathematical formulation. The sections of the columns and the other structural elements are used to build the FEM model using SAP2000 and then build the stiffness and the mass matrices (K and M) to calculate the system dyanamical properties( period, mode shape and so on).
I hope this helps you to more understand the problem.
If I understand correctly, one section represents a single solution, and if the period isn't satisfied, then the solution isn't feasible. Typically when we know a solution isn't feasible, we generate new ones until the condition is met. So when generating new solutions or when the decision variables change, you have to evaluate the period, and if the solution isn't feasible, you generate a new one. Only when the solution is feasible, then you call the objective function. You have to keep in mind that if a lot of the decision space isn't feasible (a lot of solutions don't satisfy the period), the algorithm isn't going to work very well. If some solutions violate the period more than others, then the period can be made as a constraint. You can do this by adding a constraint to the solution, which is evaluated by calling the API, which returns the period. So when you evaluate the solution, you also call the API, which will "calculate" the constraint called period.
Again an example of a badly phrased question. Study first!! And only when you have done so, pose a question! It also pays to upgrade your knowledge of the English language!
You can give a default fitness function value that is very large (in case you are in a minimization problem), the algorithm will automatically avoid such parameters in next iterations
Like this:
If period verified
Fitness = Your evaluation
If not
Fitness= large
End
- Badges
- Science topic