Research On Multidisciplinary Optimization Based On Collaborative Optimization Algorithms And Robustness

Nowadays,huge engineering problems often involve complex systems,which contain many disciplines and have coupling relations among them.The relations often show the characteristics of high coupling or even nonlinear.Meanwhile,under the influence of uncertainty factors,the optimization problems of multidisciplinary systems is relatively complex,therefore,the multidisciplinary design optimization(MDO)and its uncertainty have received wide attention of domestic and overseas scholars.Collaborative optimization method(CO)and its robustness have been widely used in MDO due to their excellent discipline autonomy and high adaptability to complex systems.However,in order to overcome the inherent defects of the collaborative optimization,its algorithm structure and optimization process need to be further improved.In this paper,aiming at solving the problems that the collaborative optimization algorithm itself is difficult to meet the consistency constraint conditions,which leads to convergence difficulties and the result that the the standard collaborative optimization algorithm are easy to fall into local optimum,an improved collaborative optimization and robust collaborative optimization(RCO)based on the mixed and dynamic penalty function are proposed.In this method,adaptive relaxation factor and dynamic penalty function are introduced,and the construction of the mixed penalty function is divide into global and local segments.This paper incorporates this method into the CO and RCO model frameworks.The improved algorithm has excellent global search ability and the method not only obtains better objective function and robust design results,but also significantly reduces the inconsistency information between system level and discipline level.In this paper,the basic static and dynamic characteristics of the ship engine room structure are studied and the improved CO and RCO algorithm is applied to the multidisciplinary design optimization of ship engine room.The Isight optimization software is used to build the collaborative optimization process of the cabin model,including the sensitivity analysis,construction of the approximate model,intelligent algorithm strategy etc.The convergence,accuracy and high efficiency of this paper's improved CO and RCO algorithms are verified by comparing with the optimization results of the improved collaborative optimization based on existing penalty function method.The engineering example also proves that the improved algorithm has some reference significance for practical engineering application.