Contact : 9561854529 Email : submitpaper@ijaea.com editorinchief.ijaea@gmail.com Working Hours: 9am to 9PM

International Journal of Advanced Engineering Application

ISSN: 3048-6807

Submit Paper Apply for Reviewer

Intelligent Fault Detection in Smart Distribution Grids Using Bidirectional LSTM-CNN Ensemble

Author(s):Archana Bhattacharya, Dinesh Kumar Palaniswamy

Affiliation: Department of Electronics, Jadavpur University, Kolkata, West Bengal, India

Page No: 44-47

Volume issue & Publishing Year: Volume 3, Issue 3, 2026/03/09

Journal: International Journal of Advanced Engineering Application (IJAEA)

ISSN NO: 3048-6807

DOI: https://doi.org/10.5281/zenodo.19351044

Download PDF

Article Indexing:

Abstract:
incorporating advanced metering infrastructure, distributed generation, and real-time supervisory control. Within this context, rapid and accurate fault detection and classification has acquired renewed urgency: the Bureau of Indian Standards estimates that uncleared distribution faults contribute approximately 31% of total technical losses on 11 kV feeders in Tier-2 Indian cities, and the average fault clearance time of 4.7 minutes in manually operated systems significantly exceeds the 0.5-minute target of SCADA-integrated digital relay systems. Traditional protective relay algorithms based on overcurrent and distance principles struggle with high-impedance faults, evolving distributed generation fault contributions, and complex multi-terminal feeder topologies that characterise modern Indian distribution networks.
This study presents a deep learning-based fault detection and classification system combining Bidirectional Long Short-Term Memory (BiLSTM) networks for temporal feature extraction with a parallel one-dimensional Convolutional Neural Network (CNN-1D) for frequency-domain feature extraction, fused through a learned attention mechanism. The model is trained on 18,500 labelled fault events — covering six fault types (no-fault, LG, LL, LLG, 3LG, and open conductor) under 48 loading and generation scenarios — generated from a validated PSCAD/EMTDC model of the 47-bus Coimbatore zone 11 kV distribution network built from actual TNEB feeder data. SHAP (SHapley Additive exPlanations) values provide feature-level explainability for utility engineer acceptance.
The proposed BiLSTM-CNN ensemble achieves 99.3% classification accuracy with 99.7% sensitivity, 99.1% specificity, and mean fault location error of 1.3% across all six fault classes, outperforming standalone LSTM (97.1%), CNN-1D (96.8%), and the SVM baseline (93.4%). Detection latency of 18.3 ms satisfies IEC 61850 GOOSE message timing requirements for digital substation protection, confirming practical deployment viability in Tamil Nadu's expanding smart grid infrastructure.

Keywords: smart grid, fault detection, fault classification, BiLSTM, CNN, deep learning, SHAP explainability, 11 kV distribution, PSCAD, TNEB, SCADA, IEC 61850, partial shading, power system protection

Reference:

  • [1] Abdel-Nasser, M., & Mahmoud, K. (2019). Accurate photovoltaic power forecasting models using deep LSTM-RNN. Neural Computing & Applications, 31, 2727-2740.
  • [2] Gers, F. A., Schmidhuber, J., & Cummins, F. (2000). Learning to forget: Continual prediction with LSTM. Neural Computation, 12(10), 2451-2471.
  • [3] Kumar, P., & Sharma, A. (2023). LSTM-based fault identification in distribution networks with high DG penetration. IEEE Transactions on Industrial Informatics, 19(3), 2847-2858.
  • [4] Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765-4774.
  • [5] Ministry of Power. (2023). Annual Report 2022-23. Government of India, New Delhi.
  • [6] Naik, R. J., Bhui, P., & Kulkarni, S. (2022). Real-time fault detection in smart distribution systems. Electric Power Systems Research, 204, 107702.
  • [7] Raghunathan, S., Senthil Kumar, M., & Palaniswamy, V. (2022). Customer interruption costs for industrial feeders in Tamil Nadu. Energy Policy, 165, 112983.
  • [8] Rai, A., Singh, S. N., & Miao, L. (2021). Fault detection in distribution feeders using SVM and wavelet. IEEE Transactions on Industry Applications, 57(6), 5782-5793.
  • [9] TNEB. (2023). Annual Statistical Report 2022-23. Tamil Nadu Generation and Distribution Corporation.
  • [10] Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30.