Researching On Shape Optimization Of The Truss With Discrete Variables Based On The Improved Genetic Algorithm

Optimization of design is one of advanced design methods at present day, which can determine the best schemes for engineering design by combining optimization theory with computer technique. Now it is difficult to find global optimum result by using typical optimization methods to solve large-scale and complicated nonlinear and discontinuous optimization problems in practices. Genetic Algorithm (GA), a fresh optimization algorithm, can undertake this responsibility. Global optimization and connotative parallel searching is one of its most important characteristics.In the paper, some modifications on typical GA were proposed based on the late research of GA and properties of truss and an improved GA was used to solve shape optimization of the truss with continuous and discrete variables. Some hybrid genetic algorithms, such as Imitative Full-stress Design and improved GA , Pseudo Parallel and improved GA and Imitative Hierarchic and improved GA, were put forward to improve searching efficiency and avoid premature convergence for genetic algorithm. Researches about multi-objective shape optimization of the truss were also done in the paper. The improved genetic algorithm was used to get effective solutions and the optimal solution was determined from effective solutions set by theory of similarity priority ratio. All jobs done in the paper are valuable for engineering application.