Constrained Multi-Objective Optimization And Its Applications In Wind-induced Vibration Mitigation Of Super High-rise Buildings

With the development of building technology and materials science,the construction heights of super high-rise buildings are getting higher and higher.The slender structure,low stiffness and damping of super high-rise buildings make them easy to induce large vibration under wind loads.The parameter optimization of the traditional wind vibration control devices used for super high-rise buildings,such as Tuned Mass Damper(TMD)and Tuned Liquid Damper(TLD),has been extensively studied.In recent years,a new type of tuned mass inerter damper formed by adding an inerter to the traditional control device has been designed,such as the Tuned Mass Inertia Damper Inerter(TMDI),which can achieve a good performance for the wind-induced vibration control with smaller mass.Since TMDIs usually have more parameters than traditional dampers,the research on optimizing TMDIs for wind-induced vibration mitigation is still limited,especially in the optimization of TMDIs with multiple conflicting objectives and constraints considered simultaneously.In view of this,this thesis focuses on the optimization of TMDIs for the mitigation of wind-induced vibration installed on super high-rise buildings by using constrained multi-objective evolutionary algorithms(CMOEAs),in order to meet the design requirements with multiple conflicting objectives and a set of constraints simultaneously.The research has significant theoretical and practical value.In fact,as early as in the 1980 s,Guangyuan Wang,an academician of Chinese Academy of Engineering,proposed a theory of fuzzy stochastic constrained multiobjective optimization design of universal structures and a practical design method based on this theory.However,due to the lack of effective CMOEAs at that time,the method can only transform a constrained multi-objective optimization problem into a constrained single-objective optimization problem by aggregating each objective multiplied by a predefined weight.Then,this transformed optimization problem is solved by using some traditional mathematical programming methods.However,it is often difficult to set appropriate weights in advance when dealing with multiple conflicting objectives.This thesis attempts to employ CMOEAs,the latest achievements in computational intelligence in the field of artificial intelligence,to optimize TMDIs installed on super high-rise buildings to control wind-induced vibration,with the expectation of better vibration control performance than those of traditional iterative optimization methods.In the course of this research,the following aspects of the science and engineering problem are explored in particular:(1)the design of constrained multi-objective optimization problems(CMOPs)which are used to evaluate the comprehensive performance of a CMOEA;(2)the design of CMOEAs which are used to solve CMOPs;(3)the application of CMOEAs on optimizing the design of a TMDI installed on a super high-rise building;(4)the application of CMOEAs on optimizing the design of an MTMDI located on a linked super high-rise building.The research work has contributed to the research community with the following novel achievements:(1)We proposed a difficulty adjustable and scalable constrained multi-objective test problem toolkit.The suggested toolkit can be used to construct a set of CMOPs by users,which can be used to evaluate constrained multi-objective evolutionary algorithms(CMOEAs)effectively on CMOPs with single-or multi-types of difficulty.This work has provided comprehensive criteria for evaluating newly designed CMOEAs.(2)We proposed a push and pull search framework for constructing CMOEAs.Under the framework of PPS,a CMOEA with MOEA/D decomposition method is implemented,namely PPS-MOEA/D.A set of CMOPs with large infeasible regions is designed by using the proposed test problem toolkit,namely LIR-CMOPs.Comprehensive experiments show that the PPS-MOEA/D achieves the best results on the LIR-CMOPs test suite,which demonstrates the superiority of the suggested PPS framework.Furthermore,the proposed PPS algorithm can be employed to optimize the TMDI and MTMDI installed on a super high-rise building and a super linked high-rise building successfully.(3)We first suggested to employ CMOEAs to optimize a TMDI installed on a super high-rise building simultaneously considering two conflicting objectives,including the extreme displacement and acceleration at the top floor,which overcomes the limitation of the traditional optimization methods.Six decision variables are optimized by adopting PPS and NSGA-II,under fluctuating across-and along-wind loads,respectively.The experimental results indicate that the optimal designed TID(when the mass ratio of a TMDI is equal to zero,the TMDI can be also called TID)outperforms TMDI and TMD in terms of wind-induced vibration mitigation under different directions,and PPS and NSGA-II have obtained better results than tuning formulae.The optimal TID is proven to be robust against variations in structural properties,and able to mitigate acceleration effect for different wind directions.(4)We formulated the MTMDI design problem as a CMOP and applied CMOEAs to solve this problem.Regarding the optimization of a MTMDI installed on a linked super high-rise building,we first formulated this problem as a CMOP with the objective of minimizing the acceleration at the top floor of each building.Several constraints,including those concerning the difference of each TMDI installation height and the peak displacement of the top floor of the two buildings,are considered.The experimental results show that PPS outperforms NSGA-II on the MTMDI optimization problem.Furthermore,both CMOEAs achieved better results than those designed by human experts,which validates the superiority of the proposed PPS for solving the optimization problem of the MTMDI located on the linked super high-rise building.Further analysis revealed that after the installation of the MTMDI obtained by the PPS algorithm in the linked super high-rise building,the wind-induced acceleration response of each building under different wind directions has decreased significantly.For Building 1,the peak acceleration response at the wind direction of90° is reduced by 50%.For Building 2,the peak acceleration responses at the wind directions of 0° and 270° are reduced by 55.6% and 50%,respectively.