5. Constraint handling procedure
The objective of structural optimization is to develop a design
that minimizes the total structural weight while satisfying all
design requirements such as member stresses, nodal displacements
and member buckling. These requirements correspond to
the constraints for the optimization problems. Like other direct
search algorithms, the ABC algorithm was originally developed
for unconstrained optimization problems, and hence it is necessary
to somehow incorporate constrains into the ABC algorithm to
solve structural optimization problems. The advantages and disadvantages
of the well known constraint handling techniques were
presented in a review paper by Coello [38] and Kicinger et al. [17].
In the Coello’s research [38], constraint-handling techniques were
divided into five major groups (i) penalty functions, (ii) special representations
and operations, (iii) repair algorithms, (iv) separation
of objectives and constraints, (v) hybrid methods. The traditional
approach for handling the design constraints of optimization problems
is to use penalty function methods in which a constrained
optimization problem is transformed into an unconstrained problem
by adding a certain value to the objective function based on
the amount of constraint violations.