Recurrent Neural Networks Based Fault Detection for Synchronous Generator Stator Windings Protection.

Helal, A.; El Dessouki, H.; Hatata, A.; El-Saadawi, Magdi Mohamed Ali; Tantawy, Mohamed Abd El-Moneim Mohamed

doi:10.21608/bfemu.2020.115397

Recurrent Neural Networks Based Fault Detection for Synchronous Generator Stator Windings Protection.

Document Type : Research Studies

Authors

¹ Electrical Engineering and Control Department., Faculty of Engineering., Arab Academy for Science and Technology., Alex. Egypt.

² Electrical Engineering Department., Faculty of Engineering., El-Mansoura University., Mansoura., Egypt.

³ Professor of Electrical Engineering Department., Faculty of Engineering., El-Mansoura University., Mansoura., Egypt.

⁴ Professor of Electrical Engineering Department., Faculty of Engineering., Mansoura University., Mansoura., Egypt.

10.21608/bfemu.2020.115397

Abstract

This paper presents a proposed approach for fault detection and faulty phase(s) identification for synchronous generator protection-based om artificial neural networks. In order to perform this approach: the protection system is subdivided into different neural network modules for fault detection and classification. The proposed approach uses Recurrent Neural Network (RNN) to detect and classify the synchronous generator internal faults. The RNN uses the three-phase current measurements from both sides of the synchronous generator stator winding as its input data. RNN was trained using various sets of data available from the simulation results of the selected synchronous model under different fault scenarios (fault type, fault location, fault resistance and fault inception angle). Simulation results of the proposed RNN based synchronous generator stator winding protection provide a great performance; in terms of accuracy, speed and reliability.

Keywords