A predictor-corrector interior point method for optimal reactive power

Reactive optimal power flow is a nonlinear large-scale optimization problem. Predictor-corrector interior pointer method (PCIP) shows excellence performance on this problem. In an iteration in conventional PCIP, the affine scaling direction is obtained in predictor step so as to evaluate the second order terms of taylor expansions for complementarity functions and obtain the corrector step further. In this paper, an improved partner of PCIP is proposed. The method is in a new rectangle coordination model. In this model, all KKT equations are transformed to quadratic or linear ones. The proposed method works in the similar way as conventional PCIP in predictor step, but it evaluates the second order terms of all the functions instead. Then it calculates the corrector step. Numerical tests are carried out and the results confirms the priority of the proposed method to conventional one.