Genetic optimization method of ship strong frame based on sequential surrogate model

Abstract

ObjectivesThe optimization design of a ship strong frame structure under the requirements of the common structural rules (CSR) is a complex and time-consuming problem. Moreover, its tremendous constraints make it difficult to judge the feasibility of any design scheme. As the approach aims at global accuracy, when adopting the static surrogate-assisted evolutionary algorithm to solve this problem, the prediction of key areas will be distorted in the case of small size samples. Aiming at the above problem, a strong ship frame optimization method based on a sequential surrogate-assisted genetic algorithm is proposed.MethodsFirst, the constraints of a strong frame structure based on CSR are analyzed, and all 675 constraints are reduced to 2 positive constraints according to constraint type. Then, surrogates for objective functions and constraint functions are constructed, and a genetic algorithm based on the feasibility principle is adopted to find the optimized solution. The true response of the solution is then calculated and the surrogates updated. In addition, the expected feasibility function (EFF) criterion is applied to update the constraint surrogates in order to refine the prediction accuracy at the constraint boundaries. The above procedures are iterated several times, and the optimized global feasible solution is finally obtained.ResultsThe proposed method can obtain a better solution than the static surrogate-based algorithm with a lower computational burden, and the weight of the design area is finally reduced by 15.55%.ConclusionsThe proposed sequential surrogate-based algorithm is superior to the static surrogate-based algorithm, and possesses good application value in the optimization of ship strong frame structures under complex constraints.

Keywords

• strong frame
• structural optimization
• sequential surrogate model
• expected feasibility function
• genetic algorithm (ga)
• common structural rules (csr)