Hardware-in-Loop Validation of a Machine Learning Algorithm with IEC 61850 GOOSE Communication to Protect a Three-Terminal Transmission Line Integrated with Inverter-Based Resources

The integration of third-terminal introduces new protection challenges. The infeed from the additional source can impair the ability of protection systems at the original terminals to detect short circuits. Additionally, the short-circuit current response of IBRs differs significantly from that of conventional generators, adding another layer of complexity in three-terminal configurations when the tapped resource is an IBR. These factors compromise the reliability of standard protection logic, compelling a fundamental re-engineering of settings to maintain system integrity.

This paper presents an ongoing project involving the integration of an 80 MVA IBR into one of the terminals of a 230 kV three-terminal transmission line. It evaluates how the infeed effect impacts distance element security and how the inverter control-loop dynamics affects the performance and dependability of diverse protection methodologies. This project employs a line current differential (87L) scheme as the primary protection, leveraging its inherent robustness in IBR-integrated networks. The study illustrates the dependable performance of the 87L scheme for both internal and external faults. In instances where 87L tele-protection channels are unavailable, a step-distance scheme provides backup protection. However, due to unreliable negative-sequence currents from IBR, distance elements face protection challenges at IBR terminal, particularly during phase faults.

To circumvent these limitations of traditional protection, we utilized DCB with transfer trip schemes and worked on a simplified method using a one-way mirror bit communication for backup protection in three-terminal lines with IBRs. While this simplified method provides a practical alternative to complex conventional schemes, it remains dependent on dedicated communication channels between terminals. To achieve a protection framework independent of these dedicated channels, this paper introduces a Machine Learning (ML) algorithm integrated with IEC 61850 GOOSE communication. This intelligent method is specifically deployed at the IBR terminal to mitigate the inherent unreliability of distance protection. The proposed strategy significantly enhances system resilience by providing a robust protection architecture for multi-terminal networks.

The proposed protection scheme was experimentally validated using a real-time Hardware-inthe-loop (HIL) setup at our laboratory. A three-terminal transmission system with an integrated

IBR was modelled within the HIL environment, with physical protective relays deployed at each terminal. This paper also details a Machine Learning (ML) algorithm developed using python programming language written within the HIL SCADA. The ML-derived decision logic is transmitted to the IBR-terminal relay via the IEC 61850 GOOSE protocol. The proposed protection scheme was comprehensively tested under both internal and external fault conditions to validate its safety, security, and reliability. This paper presents the experimental configuration and highlights the hardware-in-the-loop simulation results.