Multi-Objective Optimization for Curvilinearly Stiffened Porous Sandwich Plates Reinforced with Graphene Nanoplatelets

Abstract

With the development of innovative manufacturing technology, multi-objective optimization algorithms for optimal design of advanced composite structures have gained increasing attention. An effective and high-accurate prediction on the mechanical behavior of structures is the basic core of optimization algorithms. Thus, a novel refined sinusoidal higher-order theory (NRSHT) combined with isogeometric analysis (IGA) is developed as the high-precision solver. A novel curvilinearly stiffened porous sandwich plate reinforced with graphene nanoplatelets (CSP-GPL) is proposed as the research object. Compared with previous higher-order theories, the proposed NRSHT can more accurately forecast the natural frequencies of CSP-GPL through several numerical and experimental tests. Subsequently, the shape and material distribution design of CSP-GPL are studied with multi-objective optimization. The random forest regression (RFR) is utilized as the high-fidelity surrogate model to construct the objective function in the improved Nondominated Sorting Genetic Algorithm (NSGA-II), which can significantly accelerate the integration of NRSHT-IGA and NSGA-II. Finally, the Pareto-optimal solutions, optimizing for fundamental frequency and total mass of CSP-GPL, are obtained from the present platform, which can give effective suggestions for the future designer to meet specific requirements.