Reactive Power Management in Micro Grid with Considering Power Generation Uncertainty and State Estimation

Optimal reactive power dispatch problem in power systems has thrown a growing influence on secure and economic operation, nonlinear and multi- modal problems. Used methods in this issue can be divided into two categories: First, the classical methods like linear programming (LP), nonlinear programing (NLP), quadratic programming (QP), interior point methods (IPM), Newton-bused methods, and the second, heuristic methods like genetic algorithm (GA), evolutionary programming (EP), and particle swarm optimization (PSO). In this paper, projected quasi-Newton method (PQN) is used as an optimal algorithm. This algorithm is applied on a 6-bus micro grid in medium voltage level. To make the problem more realistic, a wind turbine is put in one of the buses to consider uncertainty in power generation. Also two buses data are not available to add state estimation to the problem. For troubleshooting of power generation uncertainty, time series prediction model is used to predict wind speed. To overcome the problems of unavailability of some bases information, maximum likelihood weighted least squares estimation (MLWLSE) is used. Finally obtained information is used to optimize the reactive power in micro grid.

Reactive Power Management in Micro Grid with Considering Power Generation Uncertainty and State Estimation Keywords:

Projected Quasi-Newton method , micro grid , reactive power optimization , time series prediction model , maximum likelihood weighted least squares estimation