In fast moving consumer goods (FMCG) warehouse operation, a set of orders will be grouped into a number of subsets, then each of which is finished by a single picking tour. Hence, the improvement of the warehouse operation efficiency mainly rest with the optimization criteria of the picking wave compositions. In this paper, we propose a 0-1 integer optimization model based on the coincidence optimization principle. The solution of our proposed model is obtained by adopting a static clustering algorithm with channel coincidence numbers as the similarity coefficient, of which a 26.5\% decreasing in the number of channels can be achieved compared with the wave compositions based on delivery order generation time FIFO principle. Furthermore, we select the delivery order contains the most SKUs as the seed, and both take into account the balance of optimization effects and calculating time. By using channel coincidence rate as the similarity coefficient and introducing dynamic update, a cut of 26.5\% in the number of waves and 34.3\% in the number of channels can be achieved compared with FIFO principle. The latter one has already been applied to a large online supermarket, and has shown significant effects.