A Review on Fault Location Techniques in Distribution Networks with Distributed Generation

A Review on Fault Location Techniques in Distribution Networks with Distributed Generation

Osita U. Omeje, Umar W. Garba

Department of Electrical and Electronics Engineering University of Lagos

Lagos, Nigeria

Daniel E. Okojie

Department of Electrical and Electronic Engineering Pan-Atlantic University

Ibeju-Lekki, Lagos, Nigeria

Abstract Fault location in electric power distribution networks is critical for minimizing outages, enhancing reliability, and reducing costs. The integration of distributed generation (DG), such as solar PV and wind, introduces challenges like bidirectional power flows and variable fault currents, rendering traditional methods less effective. This narrative review synthesizes recent advancements in fault location techniques, categorizing them into impedance-based, traveling wave-based, AI-driven, and hybrid approaches. Drawing from over 20 sources published between 2017 and 2025, including reviews and novel methods, we analyze their accuracy, robustness to DG, computational efficiency, and limitations. Key findings highlight the superiority of traveling wave and graph neural network methods in DG-rich environments, with mean errors below 0.5% in simulations. Gaps include limited focus on low-voltage (LV) grids, high computational demands for AI, and the need for real-world validation. Future directions emphasize adaptive algorithms, integration with smart meters, and cybersecurity. This review provides a roadmap for developing scalable, real-time fault location solutions in modern smart grids.

Keywords Fault location; distribution network; fault detection; optimization algorithms; machine learning; smart grid; network topology

1. INTRODUCTION

   Electric power distribution networks are the final stage of electricity delivery, connecting transmission systems to end-users across residential, commercial, and industrial sectors. Their reliability is fundamental to modern economies, supporting critical infrastructures such as healthcare, education, and manufacturing. Any disruption in these networks can lead to significant economic losses, safety hazards, and reduced consumer confidence [1].

   Faults in distribution systems arise from diverse causes, including equipment aging, adverse weather conditions, vegetation interference, and human error. Studies indicate that nearly 80% of power system faults occur within distribution networks, underscoring their vulnerability compared to transmission systems [2]. These faults manifest as short circuits, ground faults, or high-resistance faults, each presenting unique detection and localization challenges. High-resistance faults, in particular, are difficult to detect due to their low fault current magnitudes, often escaping conventional protection schemes [3].

   Traditional fault location methods, especially impedance-based techniques, have been widely used due to their simplicity and low computational requirements. However, these methods rely on assumptions of unidirectional power flow and stable voltage profiles, which are increasingly invalid in modern grids [4]. The integration of Distributed Generation (DG), such as rooftop solar photovoltaic (PV) systems, wind turbines, and backup generators, introduces bidirectional current flows and dynamic voltage conditions. These factors distort impedance calculations, leading to poor accuracy and delayed fault isolation [5].

   Recent advancements in fault location research have sought to overcome these limitations by incorporating signal processing, synchronized phasor measurements, and machine learning. Traveling wave-based methods, for instance, exploit the propagation of transient signals to estimate fault distance with high precision [6]. Synchro phasor-based approaches leverage time-synchronized voltage and current measurements to achieve fault location accuracy within ±1% of line length [7]. Meanwhile, machine learning models such as Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM) networks, and ensemble learning techniques have demonstrated strong capabilities in fault classification and adaptability to evolving grid conditions [8].

   This review critically evaluates existing literatures on fault location in a distribution network with DG penetration, highlighting their strengths, limitations, and implications for future fault management strategies. The discussion emphasizes the importance of developing scalable, adaptive, and efficient fault location techniques to support the transition toward resilient, low-carbon, and intelligent distribution networks.

   The remaining part of this paper is structured as follows. Section II presents the fundamental concepts in fault analysis. Review of related existing literatures on the subject matter is presented in section III. Discussion on the findings of the

   literature review is presented in section IV, while section V conclude the paper.

   Figure 1.1: Power network infrastructure [5].

2. FUNDAMENTAL CONCEPTS

   1. Fault Location

      Fault location refers to the process of determining the exact distance or position of a fault along a transmission or distribution line, typically measured from a substation or monitoring point. Accurate fault location is critical for reducing outage durations, minimizing repair costs, and improving overall system reliability.

      Traditional impedance-based methods estimate fault distance by measuring voltage and current at one end of the line and calculating the apparent impedance. While simple and widely adopted, these methods suffer from poor accuracy in networks with Distributed Generation (DG), where bidirectional power flows distort impedance calculations [4].

      To overcome these limitations, traveling wave methods have been developed. These techniques analyze the propagation of transient signals generated by faults, using their arrival times at different measurement points to triangulate fault location. Traveling wave methods offer improved precision and are less affected by DG penetration [5]. Recent advancements also integrate synchro phasor measurements from Phasor Measurement Units (PMUs), achieving fault location accuracy within ±1% of line length [6].

   2. Fault Detection

      Fault detection involves identifying abnormal conditions in the power system that deviate from normal operation. Early detection is essential to prevent cascading failures and widespread blackouts.

      Classical techniques include Fourier analysis, which decomposes signals into frequency components, and the Wavelet Transform, which captures both time and frequency information, making it effective for transient fault signals [7]. Modern approaches increasingly rely on machine learning models. Artificial Neural Networks (ANNs) can learn fault signatures from historical data, while Long Short-Term

      Memory (LSTM) networks excel at modeling sequential data, making them suitable for detecting evolving fault conditions [8]. Ensemble learning methods further enhance classification accuracy by combining multiple models.

      These advanced techniques improve adaptability in dynamic environments, particularly in DG-rich grids where fault currents vary significantly. They also enable real-time fault detection, supporting smarter and more resilient distribution systems.

   3. Distribution Networks

      Distribution networks form the critical link between transmission systems and end-users. They can be configured as open-loop systems, which are simpler but less reliable, or closed-loop systems, whih provide redundancy and improve fault tolerance [10].

      Reconfiguration techniques are often employed to balance loads, reduce losses, and enhance voltage stability. For example, automated switching can reroute power during faults, minimizing service interruptions. Optimization methods, such as genetic algorithms and particle swarm optimization, have been applied to improve network efficiency and reliability [11].

      Daily load variations, influenced by household and industrial activities, further complicate distribution network management. Advanced optimization ensures that networks remain stable under fluctuating demand, reducing the likelihood of faults and improving resilience.

   4. Smart Grids

   The Smart Grid represents the evolution of traditional distribution networks into intelligent, adaptive systems. By integrating Internet of Things (IoT) devices, advanced communication technologies, and renewable energy sources, smart grids enable real-time monitoring and control [12].

   Key components include smart meters, microgrids, and distributed energy resources (DERs), which collectively enhance efficiency and sustainability. Smart grids also facilitate bi-directional communication, allowing consumers to actively participate in energy management through demand response programs.

   Fault management in smart grids requires adaptive strategies. Advanced signal processing tools, such as Teager Energy Operator (TEO) and Variational Mode Decomposition (VMD), improve fault detection accuracy in noisy environments. Combined with machine learning, these tools enable predictive maintenance and proactive fault management [13].

   Ultimately, smart grids improve resilience by reducing outage durations, enhancing reliability indices such as SAIDI and SAIFI, and supporting the integration of renewable energy into modern distribution systems.

3. REVIEW OF RELATED WORKS

   Traditional Fault Location Methods and Challenges with DG. Traditional techniques rely on impedance estimation or differential protection, assuming radial topologies and unidirectional flows. Impedance-based methods calculate fault

   distance using voltage/current ratios but suffer from errors in branched networks or high-resistance faults (errors >2%). Traveling wave methods, analyzing transient wavefront propagation, offer better precision but require high-sampling devices.

   DG exacerbates issues: it reduces grid fault currents, introduces multi-directional flows, and causes wave reflections, leading to mis localization. For instance, in high-DG penetration (e.g., 5075%), traditional methods fail to detect low-magnitude faults [provided document]. Studies show DG can increase errors by 25% without adaptations.

   1. Advanced Traveling Wave-Based Techniques

      Traveling wave methods detect fault-induced transients, estimating location via time differences of arrival (TDOA). They are robust to DG as they focus on wave timing rather than currents.

      Time Matrix Approach: Uses inherent time matrix (ITM) from network topology and post-fault wave data to form a time determination matrix (TDM). Fault sections are identified by zero elements in TDM, with distance averaged from reference nodes. Advantages: Error <0.5% in IEEE 33-bus with DG (e.g., IBDG, DFIG); handles hybrid overhead-cable lines; economical device placement. Limitations: Sensitive to wave head extraction errors; assumes normalized velocities.

      Teager Energy Operator (TEO) and Variational Mode Decomposition (VMD): Enhances wavefront detection in noisy signals. In the provided document, TEO highlights energy bursts, achieving mean absolute errors (MAE) <16m in IEEE 13/34-bus feeders with DG up to 75%. Advantages: Real-time (0.050.15s execution); resilient to high-resistance faults (up to 100). Limitations: Requires accurate wave speed; hardware-dependent.

      Table 1 compares traveling wave methods:

      Method	DG Handli ng	Accur acy (Error %)	Computat ional Time	Key Limita tion
      Time Matrix	Robust (IBDG /DFIG )	<0.5	Low	Wave extract ion errors
      TEO-VMD [provide d]	Adapti ve to penetr ation	MAE <0.16 %	0.05 0.15s	Wave speed variati ons
      Modal Analysis	Moder ate	0.52	Medium	Noise sensiti vity

   2. AI-Driven and Machine Learning Methods

      AI addresses DG complexities through data-driven pattern recognition.

      Neural Networks and Hybrids: ANNs, SVMs, fuzzy logic, and genetic algorithms classify faults and locate them via feature extraction (e.g., wavelet transforms). Hybrid ANFIS models

      adapt to DG flows, achieving <2% errors. Advantages: High accuracy in noisy/bidirectional scenarios; fast response.

      Graph Convolutional Networks (GCN): Models network as graph with nodes (busbars) and edges (lines), using features like voltage/power. Multi-layer GCNs classify faulty nodes. In IEEE 33-bus with high DG (PV/wind/storage), achieves 98.5% accuracy, 95.7% F1-score, 10ms inference. Advantages: Captures spatial dependencies; robust to noise; outperforms CNN/RF. Limitations: 6-hour training; simulated data reliance.

      Other AI: Particle swarm optimization for multi-objective fault location [web:42 in provided]; VGAE-Graphs AGE for feature extraction [web:49 in provided].

      Advantages: Adaptable to DG variability; no need for exact models. Limitations: Data-intensive; retraining for changes; high compute load.

      Table 2: AI Methods Overview

      Category	Examp les	DG Relev ance	Accuracy	Limita tions
      Neural Nets	ANN, ANFI S, SVM	Bidire ctiona l flow model ing	>95%	Data volum e; overfit ting
      GCN	Multi-layer convol ution	High penetr ation handli ng	98.5%	Traini ng time
      Hybrid [provide d]	Geneti c + multi-source	Varyi ng penetr ation	<1%	Comp utation al compl exity

   3. Hybrid and Emerging Approaches

      Hybrids combine impedance with AI or waves with signal processing for enhanced robustness. For LV grids, wavelet + AI handles limited metering. Emerging: Graph neural nets for meshed DG networks

4. DISCUSSION

   Synthesis reveals a shift from impedance to traveling wave and AI methods, with DG driving innovations like TDM and GCN for <1% errors in simulations.

   Trends: Increased AI adoption (e.g., 20242025 papers [web:4,13]); focus on real-time (sub-second) and LV grids.

   Contradictions: AI excels in complexity but demands data, while waves are efficient but hardware-reliant.

   Gaps: Limited LV research; scarce real-world tests; cybersecurity overlooks; handling EVs/ESS. Future: Adaptive wave speed estimation; AI with GANs for data scarcity; integrated FDIR systems [web:1,4].

5. CONCLUSION

This review underscores the evolution of fault location techniques toward DG-resilient, intelligent methods. Traveling wave and GCN approaches offer promising accuracy and speed, paving the way for smarter grids. Implementing these could reduce outages significantly, aligning with sustainability goals.

REFERENCES

1. A. E. Emanuel, High impedance fault detection in distribution systems, IEEE Trans Power Delivery, vol. 5, no. 2, pp. 676686, 1990.

2. Y. Zhang, Impedance-based fault location in DG-integrated networks, IEEE Trans. Power Delivery, vol. 32, no. 5, pp. 21012109, 2017.

3. J. Qi, Mode-dependent fault detection filters for stochastic delayed systems, Int. J. Control, vol. 91, no. 3, pp. 512526, 2018.

4. IEEE PES, IEEE standard test feeders, IEEE Power & Energy Society, 2013.

5. H. He et al., Machine learning-based fault detection in microgrids, IEEE Trans. Smart Grid, vol. 8, no. 2, pp. 10301038, 2017.

6. P. Crossley and P. Gale, Traveling wave fault location in distribution systems, IEEE Trans. Power Delivery, vol. 3, no. 3, pp. 893901, 1988.

7. A. Phadke and J. Thorp, Synchronized Phasor Measurements and Their Applications. Springer, 2008.

8. N. E. Huang et al., The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis, Proc. R. Soc. Lond. A, vol. 454, no. 1971, pp. 903995, 1998.

9. Dashtdar, Masoud, Arif Hussain, Hassan Z. Al Garni, Abdullahi Abubakar, Waseem Haider, Kareem M. AboRas, and Hossam Kotb. "Fault Location in Distribution Network by Solving the Optimization Problem Based on Power System Status Estimation Using the PMU." Machines11, no. 1 (2022): 109. Accessed April 20, 2025.

   https://doi.org/10.3390/machines11010109.

10. L. Xie and M. Fabin, Optimization of distribution networks using genetic algorithms, Appl. Energy, vol. 250, pp. 120130, 2021.

11. R. Hutchings and Y. Sun, Advanced power distribution controllers for smart grids, Energy Reports, vol. 6, pp. 112120, 2020.

12. G. Andersson et al., Causes of the 2003 major grid blackouts in North America and Europe, and recommended means to improve system dynamic performance, IEEE Trans. Power Syst., vol. 20, no. 4, pp. 19221928, 2005.

13. N. E. Huang et al., The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis, Proc. R. Soc. Lond. A, vol. 454, no. 1971, pp. 903995, 1998.

1. Carpenter, M.; Hoad, R.R.; Bruton, T.D.; Das, R.; Kunsman, S.A.; Peterson, J.M. Staged-fault testing for high impedance fault data collection. In Proceedings of the 58th Annual Conference for Protective Relay Engineers, College Station, TX, USA, 57 April 2005; pp. 917.

2. Wester, C.G. High impedance fault detection on distribution systems. In Proceedings of the 1998 Rural Electric Power Conference Presented at 42nd Annual Conference, St. Louis, MO, USA, 2628 April 1999.

3. Silva, S.; Costa, P.; Gouvea, M.; Lacerda, A.; Alves, F.; Leite, D. High impedance fault detection in power distribution systems using wavelet transform and evolving neural network. Electr. Power Syst. Res. 2018, 154, 474483.

4. Silva, P.R.N.; Carvalho, A.P.A.; Vieira, P.; Costa, C.T. Characterization of failures in insulators to maintenance in transmission lines. In Proceedings of the 2017 IEEE International Conference on Smart Energy Grid Engineering, Oshawa, ON, Canada, 1417 August 2017; pp. 713

5. Redfern, M.A.; Bo, Z.Q.; Montjean, D. Detection of broken conductors using the positional protection technique. Proc. IEEE Power Eng. Soc. Transm. Distrib. Conf. 2001, 2, 11631168.

6. Redfern, M.A.; Bo, Z.Q.; Montjean, D. Detection of broken conductors using the positional protection technique. Proc. IEEE Power Eng. Soc. Transm. Distrib. Conf. 2001, 2, 11631168

7. M. Krenz et al., Power distribution system and method of distributing power, Mar. 28, 2013 Available: https://patents.google.com/patent/US20130257153A1/en.

8. M. H. Seavey, Power Distribution, pp. 135138, Jan. 2025, doi: 10.1201/9781003188377-13.

9. L. Xie and Z. Fabin, Power distribution system for distributed power supplying system, Feb. 26, 2019.

10. S. Meijun, L. Jie, Z. Pengfei, and L. Yanjun, Power distribution system, Nov. 03, 2020

Orozco-Henao, C., Mora-Flórez, J., Marín-Quintero, J., Velez, J. C., da Silva,

M. P. C., & Perez-Londono, S. (2019). Fault Location System for Active Distribution Networks. IEEE PES Innovative Smart Grid Technologies Conference. https://doi.org/10.1109/ISGT-LA.2019.8895375

1. R. Jiang, A review of Network Topology, International Conference on Computer, Mechatronics, Control and Electronic Engineering, pp. 11671170, Nov. 2015, doi: 10.2991/ICCMCEE-15.2015.222. Available:

   https://www.atlantis-press.com/proceedings/iccmcee-15/25839746.

2. A. Hari, Network Topology and Sensitivity Analysis, International Conference on Electrical Engineering and Information Communication Technology, pp. 0106, Apr. 2023, doi: 10.1109/ICEEICT56924.2023.10157035.

3. Y. Dengfeng, M. Qingpeng, L. Rui, W. Jintao, and W. Xiaojun, Network topology and information presentation method thereof, Jun. 29, 2018.

1. A. Varga, Fault Diagnosis, Springer International Publishing, 2023, pp. 2959. doi: 10.1007/978-3-031-35767-1_3

2. D. Galar and U. Kumar, Chapter 5 Diagnosis, 2017, pp. 235310. doi: 10.1016/B978-0-12-811153-6.00005-1. Available:

   https://www.sciencedirect.com/science/article/pii/B9780128111536000051.

3. A. Xuewu et al., Fault diagnosis method and system, Dec. 15, 2017

4. Uchida, Yoshiaki, Koichi Fujiwara, Tatsuki Saito, and Taketsugu Osaka. "Causal Plot: Causal-Based Fault Diagnosis Method Based on Causal Analysis." Processes10, no. 11 (2022): 2269. Accessed April 21, 2025. https://doi.org/10.3390/pr10112269.

5. Smart Grid, pp. 2526, Jan. 2022, doi: 10.1002/9783527803422.ch4

6. S. Misra and S. Bera, Introduction to Smart Grid, pp. 317, Mar. 2018, doi: 10.1017/9781108566506.003. Available: https://www.cambridge.org/core/books/smart-grid-technology/introduction-to-smart-grid/E7EB5897CE916FA993051C32523A8F1B

7. G. T. Basavanna, D. Y. Chavan, and A. Verma, Smart Grid Using IOT, International Journal of Advanced Research in Science, Communication and Technology, pp. 580587, Dec. 2024, doi: 10.48175/ijarsct-22769

8. Hamed, Abdullah, RazzaqulAhshan, Nasser Hosseinzadeh, Reza Ghorbani, and Eklas Hossain. "Survey of Smart Grid Concepts and Technological Demonstrations Worldwide Emphasizing on the Oman Perspective." Applied System Innovation3, no. 1 (2020): 5. Accessed April 21,

   2025. https://doi.org/10.3390/asi3010005.

9. Qin, L., Guo, Z., Huang, W., Chen, Y., & Zhang, B. (2022). A Novel Fault Locating Method of Distribution Network Based on Multiple Objective Particle Swarm Optimization. IEEE International Conference on Power and Renewable Energy, 458464. https://doi.org/10.1109/ICPRE55555.2022.9960661

10. Shang, L., & Li, Z. (2024). Research on fault location method of active distribution network based on improved multiverse algorithm. https://doi.org/10.1088/1742-6596/2814/1/012060

11. Sun, G., , ., Fu, R., He, B., & Shang, J. (2024). A Method for Fault Location in Distribution Networks Based on Graph Neural Networks and Improved Convolutional Neurl Networks. 699704. https://doi.org/10.1109/iceeps62542.2024.10693156

12. Zhao, J., Zhang, W., Cheng, L., & Zhang, X. (2024). Research on Fault Location Method of Distribution Network from Multi-Source Wide Area. 252

    256. https://doi.org/10.1109/itnec60942.2024.10733191

13. Yang, Z., Xie, C., & Yin, C. (2024). Fault Location Method for Distribution Network Using an Additional Inductance Strategy. Electronics. https://doi.org/10.3390/electronics13040712

14. Wang, Y., Shi, Q., Zhou, Y., Du, L., Wu, Y., & Lin, F.

    (2023). Distribution Network Fault Location Method Based on Improved Graph Convolutional Neural Network. https://doi.org/10.1109/cespe60923.2023.00036

15. Fan, M., Xia, J., Zhang, H., & Zhang, Q. (2024). Fault Location Method of Distribution Network Based on VGAE-GraphSAGE. Processes, 12(10), 2179. https://doi.org/10.3390/pr12102179