Decomposed predictor-corrector interior point method for dynamic optimal power flow

In this paper, a decomposed predictor-corrector interior point method (DPCIPM) is proposed for solving the dynamic optimal power flow (DOPF) problem, which is a large-scale nonlinear optimization problem. The Karush-Kuhn-Tucker (KKT) system in DPCIPM is decomposed into many subsystems based on its special block structure, where the size of each subsystem depends on the network size only. In the iterative process, slack variables and Lagrange multipliers of dynamic constraints are first predicted and corrected, and then other variables in each time interval are predicted and corrected. The parameters, such as step length and barrier parameter, are independently estimated in each subsystem. Besides, an inequality iteration strategy is introduced to avoid unnecessary computation. Implementation of the proposed DPCIPM is described in detail. The effectiveness of the proposed method has been demonstrated on the IEEE 14-bus and IEEE 118-bus systems with up to 24 time intervals. It has been found that compared with a decomposed pure primal dual interior point method (DIPM), the proposed DPCIPM is more attractive, especially when dynamic constraints become active.