An improved genetic algorithm-based damage identification method is presented that accurately identifies both the location and severity of damage in a simulated simply supported beam. Damage identification based on finite element model updating is often used and damage is identified by minimizing the error between measured and analytical results. But whether model updating can be carried out successfully mainly depends on accuracy of model, quality of vibration testing, definition of optimization problem and calculation performance of optimization algorithm. In the paper, an improved genetic algorithm (IGA) is presented firstly, the construction technique of objective function based on different residuals and their weight coefficients is introduced secondly, in the end through the simulation of a simply supported beam, and damage identification is studied comparatively. From the results it can be seen that the damage identification method developed using IGA provides greater accuracy in identifying the location and severity of damage, and the performance is better when frequency residual has a larger weight coefficient.