Target matching problems and an adaptive constraint strategy for multiobjective design optimization using genetic algorithms

In multiobjective design optimization problems, the designer may know that some objectives are harder to extremize than others or that some regions of the objective space are more desirable/important. Such useful information can be incorporated into the genetic algorithm optimization procedure by treating the more challenging/important objectives as constraints whose ideal values are adaptively improved/tightened during the procedure to guide the search. Employing this adaptive constraint strategy and a morphological representation of geometric variables, a genetic algorithm was developed and evaluated through special ‘Target Matching’ test problems which are simulated topology/shape optimization problems with multiple objectives and constraints.