The Topology Optimal Study Of Truss Structures Based On Improved Genetic Algorithm

The genetic algorithm is a kind of rising optimization algorithm recently, which belongs to bionics algorithm simulating biological evolution. Because of its simplicity and high adaptability, the application of genetic algorithm to computer science and optimization design is given more attention, for example, function optimization, combination optimization, artificial intelligence, image manipulation, and so on. The application of genetic algorithms on structural optimization design is studied based on a great deal of references and wide investigation. The fundamental conceptions and main features are introduced here. Several key factors about genetic algorithms when applied to structural optimization design are analyzed and studied, for example, the mathematical model and its constrained problem, Topology tests, and so on.Based on the analysis of the existed genetic algorithms, a different improved method is proposed for truss structure. The point-based genetic algorithm improves the accuracy and convergence speed of optimization. The improved algorithm searches the global solutions in the optimization space, maintains a diverse population, avoids the local solutions because of premature and accelerates the calculating convergence.Based on the topology optimization of trusses, some improvements of the genetic algorithm(GA) are proposed, and the improved GA is more effective and comprehensive.In the design of size and topological structure of the truss, binary encoding, crossover and mutation for size and topology variables are adopted separately. With the accuracy of the size variable coding properly reduced, the convergence of the GA is accelerated, and then the initial solution is gained. Finally, the size of section is recoded and the solution is searched in a better size precision.In order to prevent the genetic algorithm from trapping in local optimal solution, a portion of initial solution is added to the new parents.Examples show that the method is effective to the topology optimization design of truss structures with discrete variables.