Abstract: As a new type of algorithm,artificial bee colony simulates the bee behaviors to find food.Since its simple parameters and flexibility,ABC is widely used to solve engineering problems.But the premature convergence and cross-border are disadvantages of ABC.To solve these problems,a novel ABC algorithm with adaptive disturbance(IGABC) was proposed in this paper.This improved algorithm adopted symmetry axis strategy to deal with the cross-border individuals,so the search efficiency is improved.A novel global self-adaptive search equation was proposed in this paper.The new search equation improves the structure of original global search equation,and adds linear increasing strategy with threshold.The search method for onlooker bees and employed bees improves the convergence precision and speed.IGABC algorithm designs a novel method on the base of global adaptive disturbance.The simulation results on 18 benchmark functions show that IGABC algorithm enhances the exploitation capacity,and the convergence speed and accuracy have made great progress,contrasting with other six improved ABC algorithms,which were proposed in the last two years.Especially when the test function is Rosenbrock,which is very difficult to find optimum solution,the convergence precision is increased by 16 orders of magnitude.

Key words: Improved algorithm,Adaptive disturbance,Linear increasing strategy with threshold,Symmetry axis,Self-adaptive,Rosenbrock