A Global Convergent Algorithm for OPF Problems With Regularity Condition Not Satisfied

Optimal power flow (OPF) essentially belongs to nonlinear nonconvex optimization problems. With the complexity of power system operation, current optimization algorithms have several shortcomings in solving OPF problems. First, the performances of the existing optimization algorithms degrade when the OPF problems cannot meet the regularization assumption required by the optimization algorithms. Second, existing optimization algorithms either have global convergence but relatively slow convergence speed, or fast convergence speed but local convergence. To overcome these shortcomings, this paper proposes an efficient global convergence method without the assumption of regularity. This method is derived from the framework of the sequential quadratic programming-interior point method (SQP-IPM), where the equality constraints of the subproblem generated by SQP method are modified to avoid the assumption of regularity. The method is also combined with the inexact line search method, thus the method has the property of global convergence. The results of several numerical experiments demonstrates that for the OPF problems with regularity condition not satisfied, the proposed method performs best regarding both solution quality and solving time when compared with state-of-the-art method.