| The reinforced concrete frame structure is designed to meet the structural requirements of which safety, applicability and durability, and also the economic needs that reasonable structure stress, as low as possible material consumption. Design is obtained by current design method "trial- verifymodify" is not necessarily the optimum scheme which should meet the specification requirements. Structural optimization design is an important development in the theory of structural design, its connotation is not only to get the smallest or lightest weight, but also to achieve the optimal allocation of a resource reasonable, to reconcile the contradiction between the development of construction industry and economy, resources and environment in urbanization process. The theory of structural optimization has a long history, and has been applied in many fields successfully, and this contribute to the developments of many large finite element software with optimization capability. While, the research and application of optimization in reinforced concrete frame structures are relatively short. There are two aspects of challenges of the establishment of a comprehensive, practical optimal algorithm for reinforced concrete frame structure. Firstly, optimal design of reinforced concrete frame structure is multi state, multi variable, multi constraint and multi object complex optimization problems, and is a statistic problem with a lot of uncertainties; secondly, traditional optimization algorithms are struggling in the global exploration and local exploitation ability, and are hogtied by the characteristics of the large amount of calculation in complex structures.To address a series of problems in the optimization design of reinforced concrete structure, on comprehensive analysis of reinforced concrete structure optimization design model, the research of hybrid optimization algorithm based on CMAES algorithm is carried out. The main innovative work completed in this paper is included as follows:(1) Relationship between building structural displacement response and design variables is explicitly expressed using the principle of virtual work. Based on the principle of displacement based seismic design, two kinds of constraints: the target displacement constraints and constraint displacement constraints are proposed. The optimization models are solved by a nonlinear programming algorithm and the CMAES algorithm simultaneously. A structure optimization design model based on the displacement based seismic design is derived by comparing the effects of different target displacement shapes on the optimization results.(2) Combined with the performance of DE algorithm and CMAES algorithm, a self-adaptive sub population scheme is constructed, which benefits from the fact that CMAES group assists DE group to exploit the optimal solution, DE group assists CMAES group to explore the potential field. The self-adaptive sub population hybrid algorithm(Sa S-MA) is proposed, and it is applied to the numerical experiment platform and the linear optimization problems of reinforced concrete structures. Comparing with currently acknowledged algorithm, the effectiveness of Sa S-MA is demonstrated, the influences of the parameters on the optimal performance of the algorithm are analyzed, and the proposed values of these parameters are given out.(3) The design variables are divided into discrete and continuous variables, two phase of adaptive hybrid algorithm(AHA) is proposed to optimize this kind of variables. A switch operation is designed to achieve the design process of two stages and the reduction of variable dimension. A constraint which is used to handle the case of failing nonlinear analysis is built to avoid the singular point obstruct the performance of the algorithm. A strain constraints which can enhance the stability of the nonlinear analysis of the reinforced concrete structure is setup. By means of two nonlinear optimization design examples of reinforced concrete frame, the effectiveness of the algorithm is verified.(4) The self-updating kriging model is presented based on the approximate characteristic of kriging model. The refining operation of the self-updating Kriging model is implemented with the aid of CMAES algorithm. The disadvantage of large amount of calculation of the nonlinear analysis of reinforced concrete structure is overcome by replacing the nonlinear analysis program of reinforced concrete structure by self-updating Kriging model. The effectiveness of the algorithm is verified by means of two RC structure nonlinear optimization design examples, On the basis of comparative study of some key parameters the proposed value of the algorithm is obtained.(5) Using CMAES algorithm with large population size and the double cycle RBDO frame, the reliability optimization design method of reinforced concrete frame structure is solved. In order to overcome the problem of large computation in double cycle reliability optimization design method, RBDO-kriging model is proposed, unified approximation of design variable and random variable is implemented. CMAES algorithm is used to determine the refining region of design variables, reliability index method is used to determin the refining region of random variables, by this the refining operation of RBDO-kriging model in CMAES search region is implemented. Based on the comparative study of some key parameters the proposed value of the algorithm is obtained... |
PDF Full Text Request