Abstract
In this paper, the Genetic Algorithm is used to optimize the engine room structure under the static and dynamic constraints. A penalty function method is proposed to handle the complicated constraint conditions which include stress, natural frequency, and allowable magnitude of design variables. The new operator called objective elitism as well as fitness elitism is introduced to improve the efficiency of the method. Varied probability of mutation is proposed to improve the convergence of the algorithm without the risk of premature convergence to a poor solution. To handle the discrete problems, we introduced the serial numbers of cross-sections whose shape data are stored in database beforehand. Successful application of the Genetic Algorithm in the engine room optimization suggests its use to more complex engineering design problems.
