摘要

This study presents the development of a new fault zone identification scheme for busbar using logistic regression binary classifier by utilising one cycle post-fault current signals of all the bays connected to the busbar. Practicability of the presented scheme has been verified by modelling an existing 400 kV Indian power generating station in power systems computer-aided design/electro-magnetic transient design and control software package. The presented scheme has been tested on enormous cases (38,400) which were generated by varying system and fault parameters. The proposed scheme provides effective discrimination between internal and external faults with a very high (99.69%) overall accuracy. Moreover, it remains stable in case of heavy through fault conditions particularly with current transformer saturation during which the conventional differential protection scheme mal-operates. Furthermore, it provides equally compatible accuracy for unknown system/unseen data set as well as for double/one-and-half breaker busbar arrangement. In addition, performance of the proposed scheme has been verified on the laboratory prototype and results are found to be satisfactory. The average tripping time is of the order of 23 ms in case of internal faults. At last, comparative evaluation of the proposed scheme with recently presented schemes in the literature indicates its superiority.

  • 出版日期2017-1-5