Dresearch On Reactive Power Optimization Of Distribution System Considering Uncertainty

The reactive power optimization of distribution system is that, when the system parameters, active power of source, power of loads are given, all the control variables in the power system are available using intelligent optimization algorithm on the condition that all the constraints are satisfied, which can make one or more performance index achieve optimal. By adjusting terminal voltage of generators, ratio of transformers, switching group of capacitor, reactive power flow can achieve rational distribution, which can reduce network loss, improve the economy of system operation and keep the voltage of nodes stable, improve the stability of power grid.Considering the number of objective function, single objective reactive power optimization and multi-objective reactive power optimization are researched. For single objective reactive power optimization, an improved quantum behaved particle swarm optimization algorithm is proposed to solve the problem of reactive power optimization of distribution system containing distributed generation (DG). Through the algorithm, the optimal control variables are available to achieve the minimum of active power loss. At the same time, the variation of differential evolution algorithm is implemented on the algorithm which can improve the diversity of population and reduce the possibility of local optimum in multi peak problem. Moreover, DG is considered to improve the economy and power quality of power network operation. The result shows that active power loss is significantly reduced when adding DGs, and PV DG can suppress voltage fluctuation better than PQ DG. For multi-objective reactive power optimization, a quantum particle swarm optimization algorithm based on elite selection is proposed to solve multi-objective dynamic reactive power optimization including active power loss, voltage stability and cost of compensator. The algorithm uses quantum rotation gate theory on particles of feasible region to change their position, and uses quantum not gate to variate population particle by probability so as to increase the diversity of population. In addition, considering restraint of each objective function, non-dominated solution is used to express optimal solution of multi-objective function. Also, congestion ranking algorithm makes the distribution of Pareto frontier more uniform. In the iterative process, elite selection algorithm is used to select the optimal solution from individual optimal solution set and the global optimal solution set to lead the evolution of the whole population which can make Pareto frontier more uniform. At last, reactive power optimization system for distribution network is achieved based on above research theory. The system can be used to optimize power grids with different nodes from single and multiple objectives and give the final optimization results and corresponding optimization scheme.