The high penetration of distributed energy resources (DERs) introduces considerable challenges to the operation of distribution grids. To facilitate the management of DERs, distribution grids are divided into multiple cells. However, for cell coordination, existing methods require extensive time to solve the operation constraints. To accelerate the speed of solving problems, this paper proposes a decentralized optimization method for cells based on the steady-state security region (SSSR) in power injection spaces. Based on the SSSR, this paper decomposes the entire optimal problem model into several subproblems. Meanwhile, the node voltage constraints and line current constraints of the optimization model both become a set of inequalities of linear combinations of variables and the variables in the objective and constraints are identical for each cell. Furthermore, a decentralized optimization model is established for the minimum power loss, and coordination among cells is realized by Karush-Kuhn-Tucker (KKT) conditions. Therefore, the entire optimization process requires less time and few iterations. To verify the effectiveness of the proposed method, simulations for a modified PG&E 69-bus distribution grid are performed in the MATLAB environment.The optimization results reveal that the average calculation time is improved to 2.84 seconds compared to 27.32 seconds for the centralized method and 88.9 seconds for the alternating direction method of multipliers (ADMM), and the number of iterations is reduced to 3 compared to 35 with ADMM.