According to the needs as rapid computing of large-scale power flow calculation in the smart grid dispatching and control systems, and making full use of the computing resources of existing systems, this paper presents a multi-CPU and multi-core parallel method for power flow correction equation, which is suitable for shared-memory programming model. The storage and traversal of sparse matrix is simplified by C++ standard containers, and the factor decomposition process is transformed in parallel based on the graph theory and the shared memory programming model, and the parallel solution for power flow linear correction equation is realized. Finally, the comparison between the current dispatcher power flow software of smart grid dispatching and control system and the above parallel power flow software is down, and the test results show that, the computational efficiency proposed by this paper is better with the increasing of computing scale, which verifies the feasibility of above program.