For the dynamic reactive power optimization of power system with mixed integer nonlinear programming

a two-stage dynamic reactive power optimization model is proposed to reduce the calculation scale. The first stage is nonlinear programming and the minimizing power system loss and node voltage deviation are taken as target. The second stage is mixed integer programming and the minimum increment of of power system loss as the target on the basis of meeting the constraints such as adjustment number of equipment in the whole day. In consideration of such situation that the conventional penalty function method is used to deal with the constraints will affect the final result due to improper selection of penalty factors

an improved state transition algorithm mixed filter technique is proposed. In this algorithm

two population parallel search method is used and filter pairs composed of individual fitness and constraint violation are used as evaluation indexes instead of penalty function. Finally

taking IEEE 14 node system and a real power system of local level network as examples

the proposed algorithm

traditional state transition algorithm and multiple ant colony algorithm are used to solve the model respectively. The correctness of the proposed model and the algorithm is verified by the calculation results.