Improved differential evolution algorithm (IDEA) is used to solve the reactive power optimization of distribution network problem. an anti-learning population initialization method is introduced to The algorithm. the method makes the initial population diverse and is able to fully extract the information of the search space; the method introduces Gauss perturbation mechanism to the interlace operation,which improves the diversity of the population in the dimension scale;also,the Artificial colony search thoughts and the Bees accelerated evolution and reconnaissance operations strategy are added into the evolution process so that the algorithm can quickly jump out of the local optimum and avoid premature. based on the above ,a distribution network reactive power optimization model is established and solve the model with IEEE30 node system adopting IDE algorithm.then compare it with the basic DE algorithm. simulation results prove that the IDE algorithm mentioned in this article has a better performance,which can effectively solve reactive Power optimization problem.