Master of Engineering Science
Electrical and Computer Engineering
This thesis introduces a new approach to enhance the reliability of conventional passive anti-islanding protection scheme in distribution systems embedding distributed generation. This approach uses an Islanding-Dedicated System (IDS) per phase which will be logically combined with the conventional scheme, either in blocking or permissive modes. Each phase IDS is designed based on data mining techniques. The use of Artificial Neural Networks (ANNs) enables to reach higher accuracy and speed among other data mining techniques. The proposed scheme is trained and tested on a practical radial distribution system with six-1.67 MW Doubly-Fed Induction Generators (DFIG-DGs) wind turbines. Various scenarios of DFIG-DG operating conditions with different types of disturbances for critical breakers are simulated. Conventional passive anti-islanding relays incorrectly detected 67.3% of non-islanding scenarios. In other words, the security is as low as 32.3%. The obtained results indicate that the proposed approach can be used to theoretically increase the security to 100%. Therefore, the overall reliability of the system is substantially increased.
Sheikholeslamzadeh, Mohsen, "Enhancing reliability in passive anti-islanding protection schemes for distribution systems with distributed generation" (2012). Electronic Thesis and Dissertation Repository. Paper 893.