Recurrence Plot and Change Quantile-based Deep Supervised and Semi-supervised Protection for Transmission Lines Connected to Photovoltaic Plants Journal: Engineering Applications of Artificial Intelligence

Pallav Kumar Bera, Samita Rani Pani · 2025

Read Recurrence Plot and Change Quantile-based Deep Supervised and Semi-supervised Protection for Transmission Lines Connected to Photovoltaic Plants Journal: Engineering Applications of Artificial Intelligence by Pallav Kumar Bera, Samita Rani Pani (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02084.txt.

Introduction

Pallav Kumar Bera and Samita Rani Pani design a smarter protection relay for power lines that connect solar plants to the grid. Conventional relays struggle here because power electronics change how fault currents behave. Their single-ended method watches three-phase currents, builds a recurrence matrix from changes in their quantiles. It feeds the result to an InceptionTime deep learning system that spots faults, locates them, and names the faulty phase.

Simulations on two standard test grids, including the New England 39-bus system, covered switching events, high-impedance faults, sensor saturation, noise, and double-circuit lines. Semi-supervised methods such as label spreading were added so the system performs near supervised levels with little labelled data. The authors report robust performance across configurations and changing solar capacity.

Catalog record

CREATORPallav Kumar Bera, Samita Rani Pani
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE53 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-02084

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Recurrence Plot and Change Quantile-based Deep Supervised and Semi-supervised Protection for Transmission Lines Connected to Photovoltaic Plants Journal: Engineering Applications of Artificial Intelligence Pallav Kumar Bera, Samita Rani Pani Abstract Conventional relays encounter difficulties in protecting transmission lines (TLs) connected to converter-based energy sources (CBESs) due to the influence of power electronics on fault characteristics. This article proposes a single-ended intelligent protection method for the TL segment between the grid and a Photovoltaic (PV) plant. The approach utilizes a Recurrence Matrix and an InceptionTime-based system to identify faults by using the mean change in quantiles of 3-phase currents. It determines the fault position and identifies the faulty phase. ReliefF feature selection is applied to extract the optimal quantile features. The scheme’s performance is assessed under abnormal conditions, including faults and capacitor and load-switching events, simulated in Power Systems Computer Aided Design / Electromagnetic Transients Program (PSCAD/EMTDC) on the Western System Coordinating Council (WSCC) 9-bus system, with various fault and switching parameters. The scheme is also validated on the New England IEEE 39-bus system and in presence of partially rated converters. Additionally, the validation of the proposed strategy takes into account various conditions, including double-circuit line configuration, noise, series compensation, high-impedance faults, current transformer (CT) saturation, evolving and cross-country faults, remote and local faults, as well as variations in PV capacity, sampling frequency, and data window size. To address label scarcity and improve generalization, semi-supervised learning paradigms including label spreading, label propagation, and self-training are integrated with the InceptionTime framework, enabling near-supervised performance with limited annotated fault data. The results demonstrate that the approach is effective in handling different system configurations and conditions, ensuring the protection of TLs connected to large PV plants. 1 INTRODUCTION In recent years, sustainable power generation (SPG) have gained substantial traction as an alternative to fossil fuels in electricity generation. This growth is expected to continue as governments, corporations, and individuals increasingly embrace SPG for their environmental and economic benefits. A considerable portion of SPG, including solar photovoltaic (PV) systems and wind farms (WFs), is integrated into the power grid by means of high-voltage transmission lines (TLs), using power electronic converters to facilitate energy transmission from remote locations. The introduction of SPG has fundamentally altered traditional power system topologies, leading to bidirectional power flows and varied fault current levels. To maintain grid stability, modern grid codes impose fault ride-through (FRT) requirements on these converter-based energy sources (CBESs), which influence fault current characteristics due to the intermittent nature of the resources and their reliance on power electronic interfaces [ 1 ] . Unlike traditional synchronous generators, the fault current behavior in CBES is governed by specific control strategies, inverter parameters, and the overall fault conditions in the system, among other factors. The increasing integration of CBESs, such as large-scale PV and wind installations, presents substantial challenges to existing protection schemes, including distance, directional, differential, and overcurrent protection. These conventional schemes struggle to maintain dependability and security due to the new fault dynamics from CBESs, including lower fault currents to reduce thermal stress and lower negative-sequence currents to prevent bus capacitor overvoltage. Studies have highlighted the impact of PV integration on traditional distance protection used in TLs. The low and unique fault current characteristics from PV can lead to misoperations in distance relays [ 2 ] . Oscillations in apparent impedance caused by CBES-injected currents can result in overreach and underreach issues for phase distance elements [ 3 ] . The impedance behavior of CBESs under different FRT requirements is analyzed highlighting differences between measured and actual fault impedances due to control strategies in [ 4 ] . The presence of CBESs also impacts negative-sequence directional elements, which are crucial for phase and ground distance protection [ 5 ] . The misoperation of negative-sequence overcurrent and directional negative-sequence overcurrent relays were studied in [ 6 ] . In differential protection schemes, the integration of large-scale PV plants introduces reliability concerns due to disparities in short-circuit characteristics between PV inverters and traditional synchronous machines [ 7 ] . Sensitivity challenges in differential protection of TLs with CBESs have been documented in [ 8 ] . Overcurrent relay performance is impacted by distortions in fault signals and shifts in angular relationships between voltages and currents, leading to challenges in fault direction determination, localization, and prevention of false tripping [ 9 ] . Further, high-impedance fault detection becomes increasingly difficult in systems with integrated PV, as highlighted in [ 10 ] . To address these challenges, alternative protection methods have been proposed. An enhanced distance protection scheme based on zero-sequence impedance and time delay was suggested in [ 11 ] to address impedance measurement distortions in CBES-integrated systems. A current differential protection approach was developed in [ 12 ] using phase current signs at both ends of the TL connected to WFs. Transient current correlation-based differential protection [ 13 ] and positive sequence-based distance protection independent of plant parameters [ 14 ] are explored for CBES-integrated systems. Fault detection, classification and location are performed using the magnitude ratios of the line end phase currents and apparent power in [ 15 ] . These methods aim to enhance fault detection sensitivity and reliability in TLs connected to CBES. To further support protection reliability in complex scenarios, recent efforts have turned to machine learning (ML)-based protection schemes, which adapt dynamically to system conditions. A hybrid fuzzy and random forest (RF) using combined linear trends (CLT) was used in [ 16 ] to detect and classify faults in grid-connected PVs. However, its application is limited by high sampling frequency requirements. Techniques involving positive-sequence current and empirical mode decomposition (EMD) with RF classifiers have been applied for fault detection in lines with TCSC compensation [ 17 ] . However, it primarily focuses on DFIG-based wind farms and may not readily extend to PV-integrated systems or hybrid renewable sources. Support Vector Machines (SVM) have been applied to estimate impedance in PV-integrated TLs under variable infeed conditions [ 18 ] . Fault detection using discrete wavelet transform combined with SVM-based classification in renewable-energy-penetrated microgrids has been reported in [ 19 ] . Similarly, ensemble tree-based approaches for fault detection in inverter-based generator systems have been presented in [ 20 , 21 ] . A deep learning–based protection algorithm employing two-dimensional spatial current trajectory imaging for PV-connected transmission lines has also been proposed in [ 22 ] . However, these studies explore only a limited set of features and do not address validation under many critical scenarios, thereby restricting the robustness and generalizability of the methods. Due to the unique fault current characteristics that challenge existing protection techniques, it is essential to develop new protection schemes or enhance existing ones with a more comprehensive exploration of feature sets. This article proposes a novel integration of recurrence matrix representations of change quantiles (RMCQs) with an InceptionTime-based classifier to achieve intelligent protection of transmission lines connected to PV plants. The primary contributions of this study are as follows: 1. A unique RMCQ-based intelligent technique is proposed for fault detection, fault localization, and identification of faulty phases. The method is tested across various scenarios, accounting for parameters influencing fault current behaviors. 2. The RMCQ-based protection approach is validated on the IEEE

Read the complete text