Research On Method For Optimal Reactive Power Planning Based On Quasi Stepless Adjustment

The mission of Optimal Reactive Power Planning (ORPP) is to determine the installation location and capacity of reactive power compensation device based on the generation and power network planning, the ORPP can help to guarantee the secure and economic operation for power system. However, as one of the most complicated nonlinear problems, with character- istics of multi-objective, -uncertainty, -restriction, -extremum and discreteness property, many problems of ORPP are still not solved yet both in theory and practical application even heated research over thirty years. Aiming at the optimal capacitor grouping configuration in ORPP and the ORPP optimal calculation in large-scale power system, this thesis presents a new method for ORPP based on quasi stepless adjustment, in which a new capacitor configuration scheme called quasi stepless adjustment scheme is introduced come from the actual scheduling and operation planning needs in substation. Moreover, a set of solving method for ORPP optimal calculation in large-scale power system is also proposed in this thesis. The research contents are outlined as followed:(1) Quasi stepless adjustment scheme is presented. According to the results by the ORPP optimization, this scheme can provide multiple tap settings for shunt capacitor through refining and dissimilating the unit capacitor capacity, by this way, the voltage regulating performance of substation can be improved. Then, three index from the view of economy, control strategy complexity and reliability is presented to evaluate the proposed scheme.(2) Improved Cataclysmic Genetic Algorithm (CGA) is presented for solving the reactive power optimization problem. In the ICGA, an improved catastrophic model depending on the number of generations is designed. Others, it is designed a new probability algorithm of crossover depending on the number of generations, and a new probability algorithm of mutation depending on the fitness value. By this way, the searching capacity stability and convergence performance.(3) A mothed based on Multi-Agent Systems (MAS) is applied for ORPP global optimization in large-scale power systems, In order to decrease the problem's dimensions, the large-scale power network is decomposed into some small independent network through natural partitioning method. By this way, the centralized optimization of the whole system is replaced by the distributed cooperative optimization of several sub-networks via simple regionalization and communi- cation. The results of example indicate the rationality and validity of the method.It is shown that the models and the corresponding algorithms have validity, rationality, practicability and creativeness via actual measurement results in Zhongshan area. The systematic investigation of this thesis is of great importance of theory and practice and will drive the reactive power optimization planning technique practical.