An Intelligent Hybrid Approach Using KNN-GA to Enhance the Performance of Digital Protection Transformer Scheme

Abstract

This paper presents a novel hybrid K nearest neighbor-genetic algorithm (KNN-GA)-based digital protection transformer scheme, which effectively discriminates its internal faults with noninternal faults. The internal faults include the faults within the current transformer (CT) locations on two sides of the transformer. The noninternal fault includes magnetizing inrush current, sympathetic inrush current, recovery inrush current, external faults (faults outside the CT locations), and overfluxing. In conventional differential protection of a transformer, many maloperations of differential relay have been reported under certain working conditions. A new hybrid KNN-GA algorithm is put forward to improve the performance of the differential relay of a transformer. In this paper, real-time experimental results are presented for laboratory custom built differential relay model of a transformer. The generated experimental data for one power frequency cycle for various operating conditions is used by MATLAB to test the performance of a proposed algorithm. The proposed scheme has also been implemented on a digital signal processor TMS 320C6416T for a real-world application. The performance evaluation shows that compared with conventional methods, the proposed algorithm has a good reliability in terms of discrimination between internal and noninternal faults.