Controlled islanding strategy for power systems based on flexible semi-supervised spectral clustering
Many blackout occurrences such as those in USA, Canada, and Italy (2003), Brazil and Paraguay (2009), and India (2012) are some evidences proving the vulnerability of current electrical power systems. Having a preventive plan is necessary to protect systems from experiencing blackout. Intentional is...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2014
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/47945/1/FK%202014%203RR.pdf http://psasir.upm.edu.my/id/eprint/47945/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Many blackout occurrences such as those in USA, Canada, and Italy (2003), Brazil and Paraguay (2009), and India (2012) are some evidences proving the vulnerability of current electrical power systems. Having a preventive plan is necessary to protect systems from experiencing blackout. Intentional islanding is a self-healing method with the main goal is to prevent the system from cascading outages which lead to blackout. Islanding strategy is based on splitting power systems by means of cutting lines into several smaller isolated ones called islands, so that the cascading effects and disturbances flowing in the grid are stopped. However, without considering specific constraints, these islands will not be stable and will collapse soon and even the stability of the grid worsens. Previous methods can minimize partitioning cutsets (either power imbalance or power disruption) while fully satisfying only one constraint (slow coherency). Thus,there is a possibility that by not considering other factors during islanding, the final suggested islands are not stable enough. The framework proposed in this research is capable of handling multiple constraints applied to the system. Furthermore, unlike prior spectral clustering methods which are not capable of satisfying a constraint partially, here it is possible to define degree of satisfaction. It is a value defined for the combined constraint specifying how much satisfied constraints should be. The combined constraint is the combination of all constraints built based on preferred factors such as slow coherency and minimal power imbalance. Hence, the proposed method is called flexible semi-supervised spectral clustering for controlled islanding. In this work, slow coherency is chosen as the first and most preferred constraint, so that generators are categorized in slowly coherent groups. To generate stable islands,minimal load-generation imbalance is computed which results the second constraint. As the final step, lines with lower power flow are discovered and chosen to find minimum power flow disruption. In order to verify applicability of the proposed framework, it is applied to two IEEE test cases: 39-bus and 118-bus. By using this framework containing several constraints, it is shown that this method of islanding generates more stable islands by causing as few as possible power flow disruption and load shedding. The obtained results clearly confirm that the proposed framework is able to find several cutsets based on the defined constraints. This new method generates islands while considering different factors of power systems simultaneously which is expected to lead to the most stable islands, and therefore save systems from blackouts. |
|---|