Software Design For Reactive-power Optimization Of Large-scale Power Systems Based On A Primal-dual Interior-point Algorithm

VAR Optimization is a complicated Nonlinear Mixed Integer Programming problem with discrete and continuous variables on object function and conditions. The algorithm in this paper is a new nonlinear primal-dual algorithm with quadratic Penalty Function, it realizes successive discretization of discrete variables in optimization process, solves discrete variables efficiently, and possesses stable convergence and numerical robustness of primal-dual algorithm. A predictor-corrector technology is applied in the algorithm, which effectively improves stable convergence and numerical speed of the algorithm. The program in this paper was developed include an electric power component class library ,the STL, uBlas and UMFPACK libraries based on Object-Oriented technology, so it cost less time to develop ,maintain and update. The input and output file formats of the program are same with the BPA flow input format, which brings conveniences to analyst. The numerical results for three different test systems and Ningxia power system show that the proposed method is accurate, reliable and effective, and is very promising for the discrete optimization of large-scale systems due to its fast execution time and reasonable accuracy.