Taking into account the basic artificial bee colony algorithm converges slowly and prematurely, an improved artificial bee colony algorithm based on local search is proposed. The method makes full use of the stochastic dynamic local search to optimize the current best solution to speed up the convergence rate. In order to maintain the population diversity and avoid premature convergence, the selection probability based on ranking is used instead of depending on fitness directly. Through the simulation experiment on a suite of standard functions, the results show that the algorithm has a faster convergence rate and higher solution accuracy.