Modeling and optimization of floating raft systems in submarines under different objectives by using hybrid genetic algorithm

Abstract

A systematic modeling and optimization method for analyzing typical floating raft systems in submarines is presented and optimization under different vibro-acoustic objectives is investigated. The Frequency Response Function-based (FRF-based) substructuring method and the coupled Finite Element/Boundary Element (FE/BE) method are employed to study vibration transmission from the vibrating machinery to the base structure as well as the vibro-acoustic behavior of the fluid- loaded cylindrical structure. The hybrid genetic algorithm (HGA) involving the FRF-based substructuring sensitivity analysis (SA) and the genetic algorithm (GA) is developed and applied to obtain the optimal values of stiffness of isolators under five objective functions at low frequencies. Optimization is carried out for two cases where the fluid–structure interaction is considered and neglected respectively. The modeling results demonstrate that the effect of fluid-loading reduces the driving point mobility of the connecting points on the base structure but has little influence on the isolation frequencies of the whole floating raft system. The optimized results are compared with those obtained by SA and GA. The results of SA show a regular trend of variation while those of GA show an irregular trend. Since GA is a stochastic optimization technique, the trend of variation in the results of HGA combines those of GA and SA. The optimized results demonstrate that HGA possesses the merits of SA and GA, and never gets trapped at a local optimum or goes into premature convergence. Compared with SA and GA, however, it requires more evaluation of objective functions in each iteration step and as a result is not computationally efficient.