Abstract: The basic artificial bee colony algorithm has a slow convergence speed, and easily gets trapped in local optimum. An improved algorithm given in this paper combined chaotic operator and inverse operator and then produced initialization population to improve the diversity of initial population. The parameter adaptive search operator was put forward and applied to the position updating of employed bees and onlookers bees. The improved algorithm (Improved ABC, IABC) had been experimented by five standard unimodal or multi-peak functions. The experimental results showed that the IABC algorithm is superior to the ABC algorithm in the search efficiency, the quality of the optimal solution and the stability