Reactive Power Optimization And Planning Based On The Adaptive Immune Algorithm And Predictor-Corrector Interior Point Method

Reactive power optimization and planning are two of the essential problems in the power system. Reactive power optimization is to minimize the power loss by adjusting the terminal voltage of generators, the tap position of transformers and the reactive compensations (capacitors or reactors), with the constraints of the security and operation under the conditions of the configuration parameters and loads given. While reactive power planning is to identify the optimal locations and capacities of new VAR sources with the constraints of voltage profile and stability at different load levels in the most economical way.Reactive power optimization and planning is the mixed problem of discrete, constraint, nonlinear combination optimization. Interior point method and stochastic methods having globally searching capability have been extensively applied in the field duing to their characteristics. Aimed at the problem of reactive power optimization and planning, the contents of this thesis are as follows:Firstly, Adaptive Immune Algorithm (AIA) is proposed to apply to reactive power optimization. By automatically adjusting the parameters: selecting rateα, cloning radius r and mutation radius R with the help of the measure of distance between the antibodies, and simultaneously searching in both larger and smaller regions, AIA achieves the balance between the fast convergence and diversity of things which avoid the local minimum point; thus AIA is excellent at computation speed and convergence. The test results show that AIA owns obviously advantage comparing with other methods.Secondly, a new method–the combination of AIA with predictor-corrector interior point method is proposed to apply to reactive power optimization. Firstly, the method makes use of the capability of the global optimization of AIA to find the candidate optimal solutions, which are taken as the initial points of interior point method; then, the predictor-corrector interior point method is used to perform the local optimal search in the neighborhood of candidate optimal solutions to further improve the accuracy of the solutions; at last, according to the variation of the duality gap during the iterating process, a new selecting methods of central parameterσand relevant obstacle parameterμare given. The test results indicate that the new method is effective.Finally, a new model of reactive power planning is established, which takes the sum of investment of reactive equipments and active power loss as objective function, and the limit of voltage profile, voltage stability to be constraint, with respects to credible contingencies at different load levels. Moreover, a new decomposition strategy for this new model is also proposed: (1) The coupling between Planning Problem and operation problems is decomposed, planning problem optimizes the locations and capacities of new VAR sources; operation problem, which processes various contingencies, is considered to verify and regulate the optimal operation point produced by planning phase; (2) The decomposition solution is also applied to different load levels, a set of candidate sub-optimal compensated locations and capacities in the condition of the maximal load level are calculated by the combination of AIA with predictor-corrector interior point method, then the optimal compensated position and capacity is picked up by taking into account of the minimum losses at minimu load level and middle load level at each candidate compensated location and capacity.