Combined with GA,BP neural network is introduced into the optimization analysis of beam string structures and the imitated full-stress method for structures with discrete variables is amended to form a new type of combinational optimization method.

Combined with GA(genetic algorithm),BP neural network is introduced into the optimiza-tion design of beam string structures and the imitated full-stress method for structures with discrete variables is amended to form a new type of combinational optimization method.

A hybrid algorithm (HA) integrating the quasi-full stress algorithm and the standard genetic algorithm (SGA) is proposed to expedite the evolving process of genetic algorithm so as to apply it to the topology optimization of truss structures with discrete variables.

Regarding the rigid frame as studying object, a hybrid algorithm composed of the imitative full-stress design method and the standard genetic algorithm was proposed for solving the question of discrete topology optimization.

An imitative full-stress design method was presented for structural optimum design with discrete variables.

The paper introduces a three-dimensional elastic continuum topological optimization method with stress constraints: amendatory full stress method,namely,it makes every element of the three-dimensional elastic continuum in a full stress state through amending the element modulus of elasticity repeatedly,then deletes the elements that their modulus of elasticity is small.

Through introducing an imitative full stress design operator in genetic algorithm(GA),a optimum design technique was proposed for section optimization of frame structure with discrete variables,namely the section optimization technique of frame structure based on full stress design and GA.