Research On Distributed Algorithm Of Multi-area Optimal Reactive Power Flow

Dispatch by levels and by district and development of power market cause the infeasibility of wholly collecting and exchanging data between areas and make centralized computing on optimal reactive power flow (ORPF) difficult to realize. However, independent computing by each area based on simple equivalent may bring out notable error. How to solve ORPF of whole networks on separated data source becomes a great challenge. In this dissertation, base on Auxiliary Problem Principle (APP) and Interior Point Method (IPM), distributed algorithms of multi-area ORPF are deeply investigated and the research mainly includes following parts:Based on the nature of distributed ORPF model and APP method, the relationship between local reference buses and global bus is analyzed and a new strategy, abandoning local reference buses, is proposed, whose effectiveness is explained theoretically. In the strategy, since phase angle variables of local reference buses and angle reduction between areas are left out, the speed of APP algorithm and programming efficiency are enhanced. The effect of local IPM convergence of sub-area on global convergence is also analyzed and another new strategy, dynamically adjusting IPM's tolerance, is proposed. On the basis of ensuring convergence accuracy of the algorithm, this strategy reduces remarkably iterative times and improves the rate of convergence.A high-performance distributed IPM is proposed to solve multi-area ORPF. By duplicating border buses and introducing coupled constraints, the whole network is decomposed into sub-areas whose internal variables is independent and they couple each other only with duplicated variables. Moreover, it makes the coefficient matrix of centralized IPM's linear correction equation a block diagonal bordered matrix, which can be decomposed into the constraints equation on internal variables of independent sub-areas and coupled constraints equation. During each IPM iteration, internal variables and coupled variables are solved distributedly; thus distributed algorithm of multi-area ORPF is implemented. Data communication volume between each sub-area server which is used to calculate internal variables and the coordinator which is used to calculate coupled variables is merely related to the number of coupled variables and the size of sub-areas is unrelated. This algorithm can fully satisfy the demand of distributed computing and shows advantage of high speed of convergence, less iteration times, reliable convergence and high accurate results.These strategies and the algorithm presented in this dissertation have been verified by IEEE standard systems; the simulation results prove that they are correct and effective.