Research On Model And Hybrid Algorithm Of Power System Dynamic Reactive Optimization

The dynamic reactive optimization problem has considerable dimension and various constraints including successive constraint, discrete constraint and dynamic constraint, and there is not a general algorithm so far. In this dissertation, the optimal algorithm based on the predictor-corrector primal-dual interior point method and genetic algorithm is preparatory explored and the correlative problems including static reactive optimization and power flow are studied. A layer-by-layer back/forward sweep power flow algorithm for radial distribution system is proposed. In the algorithm, the power loss and voltage loss of the branches on the same layer are computed without special numbering of buses and branches. To avoid disconvergence of power flow brought by the large shunt conductance in traditional ∏ model of transformer, the voltage transforming model of transformer is proposed and applied. So the algorithm has reliable convergence as well as fast speed.A three-phase power flow algorithm based on the back/forward sweep of branch current phase components is proposed. The transformer winding connection, the phase-shifting between primary and secondary windings, and the off-normal tapping are considered and the corresponding phase voltage equation and phase current equation of transformers are deduced in the algorithm, which make the algorithm generally applicable. A novel hybrid algorithm, which combines predictor-corrector primal-dual interior point method and genetic algorithm, is presented for dynamic and static reactive optimization in this dissertation. The algorithm fully make use of the advantage of interior point method in solving successive optimization problem and the advantage of genetic algorithm in solving discrete optimization problem. The original problem is divided into three parts: total successive problem, discrete problem and successive problem. The effect of convergence criterion on convergence speed is fully considered and feasible criterion and optimal criterion are applied in total successive problem and successive problem respectively, so the whole performance of the algorithm is increased. Moreover, by solving the correction equation instead of iteration equation, the iteration calculation of interior point method is simpler and faster, while the result is equivalent. The algorithms presented in this dissertation have been verified by IEEE standard systems and practical power systems, the simulation results proved that the algorithms are correct and effective.