Six Phase Transmission Line Protection Using Bat Algorithm Tuned Stacked Sparse Autoencoder

Abstract

Six-phase transmission lines have the capability to address the continually evolving power demand. It allows upgrading the power transfer capability of the prevailing three-phase double circuit line without major changes in the transmission corridors. However, the operational performance of any six-phase system is highly dependent on its protection scheme. The possibility of larger number of faults in six-phase system complicates the protection task. Furthermore, the harmonics intrusion arising because of nonlinear loading compromises the reliability of the conventional threshold-based protection schemes. In this regard, this article addresses the above-mentioned challenges by developing a protection scheme based on the hybrid frameworks of bat algorithm and stacked sparse autoencoder-deep neural network (SAE-DNN). To overcome the limitation of SAE-DNN regarding optimal selection of architecture and tuning parameters, the selection task has been formulated as an optimization problem and solved using Bat algorithm. The use of raw voltage and current signals as input to the SAE-DNN reduces the overall complexity of the protection scheme. The efficacy of the proposed scheme has been validated for all 120 types of faults under varying operating, loading and fault scenarios. Furthermore, the proposed scheme has been validated for practical settings by performing real-time simulations on OPAL-RT digital simulator.