The Improvement Of Automatic Grouping Genetic Algorithms And Its Application In Structural Engineering

With the development of economy and society, engineering structure design became more and more challenging for the higher requirements of them. The satisfied design not only should meet the functional requirements, but must be safe, reliable and economical. Conventional design methods are difficult to finish such tasks, and more advanced design methods are needed. Structural optimization method is a powerful tool to help engineers improve and optimize the design or obtain a creative design. The theory and the method of structural optimization have been studied extensively and applied in the structural engineering field to some extent. However, many difficulties also exist for the complex engineering design problems. The study on optimization method to solve the complex engineering design problems is very important. In the present work, a series of researches are carried out in order to improve the automatic grouping genetic algorithms (AGGA) and apply it in the structural engineering design. With a comprehension of the basic theory of genetic algorithms (GA) and the characteristics of AGGA, several improvements are presented and studied by some benchmark tests. Three types of problems:the form-selection optimization of building structure, the pile optimization problems with discrete design variables, complicated constraints functions and grouping requirement, the frame topology optimization problems with frequency constraints, are elaborately selected to show the advantages of AGGA. According to the characteristics of such problems, well-selected design variables and constraints, appropriate optimization model, and suitable optimization method are main research features of the dissertation. The main works are as follows:1. Based on the theory of GA and the characteristic of AGGA, three improvements are proposed, i.e. controlled elitism strategy, segment crossover operator and adaptive penalty function. Controlled elitism strategy uses the concept of elitism population comprising of the best individuals. The number of individuals to be replaced by the elitisms in the current population is determined automatically. Segment crossover operator allows both the group and member chromosome to cross concurrently to improve the exploration of new schema and help produce better individuals. The proposed penalty function has three features, i.e. adaptive penalty capability; not requiring any parameter; automatically defining a different penalty coefficient which varies along the run according to the feedback received from the evolutionary process for each constraint. Moreover, it keeps those near-optimum designs with only slight constraint violation in the population and increases the probability of achieving the optimum. Five benchmark tests and two structural optimization tests demonstrate the efficiency of the proposed improvements. (see Chapter 2)2. The structural optimization problem of high-rise buildings is studied by the improved AGGA. A minimum-volume optimization model with multiple design constraints based on practical design requirements and a cardinality constraint is built to achieve the concurrent optimization of size and form-selection. Design variables are well-constructed and five assumptions are presented. Three different buildings with the heights of 6-story,10-story and 40-story are optimized by the improved AGGA. The optimal structures demonstrate that buildings of different height should select different structural form. (see Chapter 3)3. The optimum conceptual design of pile foundations at the initial design stage is studied by the improved AGGA. A modular method is proposed, which divides the foundation into modules and each module is identified by its characteristics of pile length, diameter, number and layout. Modules with the same characteristics may be packed and represented by a design variable. A minimum-volume optimization model with multiple design constraints based on Chinese code and a cardinality constraint is built to achieve the concurrent optimization of pile size and layout. The model is solved by the improved automatic grouping genetic algorithms to obtain the design with optimal variables and optimal variable grouping. A practical example demonstrates the effectiveness of the proposed approach. (see Chapter 4)4. Topology optimization problem of skeletal structures with frequency constraints is studied by the improved AGGA. The dissertation points out that the problem is greatly affected by the model choices. Using truss model leads to optimum design with some very thin members. Its actual frequencies can become very low. For the optimization problems of skeletal structures with the requirement of local bending vibration, frame model is necessary because it captures local vibration. Furthermore, it is shown that the topology optimization of frame structures with frequency constraints has the characteristic of singular optimum if the ground structure approach is used to search for the optimum topology. Frequency constraints also result in the disjoint feasible domain. Therefore, the improved AGGA is applied to treat these difficulties. To demonstrate the effectiveness of improved AGGA to obtain the singular optimum, two skeletal structures subject to frequency constraints are optimized using truss and frame model respectively. As predicted, the optimal structure based on truss model ignores severe local vibration that violates the frequency constraints. By contrast, the optimal structures from frame model reflect the local vibrations and obtain the singular optimum. The two skeletal structures based on frame model are optimized by continuous search method sequential linear programming (SLP), and neither of them are converge to the singular global optimum design. (see Chapter 5)5. The symmetry of optimum topology design of frame structures with frequency constraints is studied by the improved AGGA. For the difficulties of disjoint feasible domain and singular optimum, the improved AGGA is used to obtain the global optimum. Two benchmark tests are studied and the results show topology optimum may not be symmetric even the geometry, material distribution and ground structure are symmetric. The reasons resulting in the non-symmetric optimum solution for the frame topology optimization subject to frequency constraints are discussed. The inertial force acting on beams and the concentrated masses depend on the movement of beams and nodes and are often not symmetric, which explains the non-symmetric optimum solution to some extent. (see Chapter 6)At the initial stage of the dissertation, when the Wenchuan earthquake occurred in 2008, the author attended to the study for strengthening building structures. An anchorage-tie strengthening method was proposed and one paper has been published. This work is presented in the appendix for it is different to the main work of this dissertation.The research of this dissertation was partially supported by the National Natural Science Foundation of China (Grant No.50878038,90816025). This support is gratefully acknowledged by the author.