| Summary: | Engineering design optimization problems are formulated as large-scale mathematical programming problems with nonlinear objective function and constraints. Global optimization finds a solution while satisfying the constraints. Differential evolution is a population-based heuristic approach that is shown to be very efficient to solve global optimization problems with simple bounds. In this paper, we propose a modified differential evolution introducing self-adaptive control parameters, modified mutation, inversion operation and modified selection for obtaining global optimization. To handle constraints effectively, in modified selection we incorporate global competitive ranking which strikes the right balance between the objective function and the constraint violation. Sixteen well-known engineering design optimization problems are considered and the results compared with other solution methods. It is shown that our method is competitive when solving these problems.
|
|---|