An accurate state estimation plays an essential role in power system operation and planning in energy management systems. However, existing multi-area state estimation researches have not focused on the importance of system clustering. The clustering mechanism divides or partitions a system according to user-defined criteria. Few published research works have mentioned the importance of considering the electrical properties of a power system while devising their partitioning methods.
To the best of our knowledge, these publications have not considered the application of such a concept to multi-area state estimation. This research attempts to model a partitioning technique of the power system whose purpose is to ensure the sub-system observability prior to the multi-area state estimation. Hence, the accuracy of the multi-area state estimation could be improved. A modified genetic algorithm-based phasor measurement unit (PMU) placement is introduced in this thesis, which includes the electrical distance-based additional PMU installation technique. The modified partitioning method is introduced in this thesis based on a genetic algorithm partitioning algorithm. In the modified partitioning method, a proposed genetic index is proposed aiming to include the consideration of system observability, which employs the proposed PMU placement technique to represent the sub-system observability.
In addition to the partitioning method, few publications have considered the state estimation by only employing PMU measurements. Furthermore, few publications have considered employing the noise statistics estimation technique to state estimation to improve the convergence of the estimation process. A cubature Kalman filter-based algorithm (CKF) is used in the thesis to solve the state estimation problem where only PMU measurements data are employed. An online noise statistic estimation technique is incorporated into the CKF to improve convergence. A modified two-level MASE is introduced to implement the modified CKF. The modified partitioning method is applied to the multi-area state estimation algorithm. By employing all the techniques introduced in this thesis, a considerable improvement of accuracy and convergence can be achieved.