Pith. sign in

REVIEW 3 major objections 6 minor 37 references

Enhanced Rapid Detection of High-impedance Arc Faults in Medium Voltage Electrical Distribution Networks

T0 review · 3 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a Koopman-based forcing signature can flag high-impedance arc faults in 12 kV distribution systems in 0.45 ms, with 100% simulation accuracy.

desk verdict A genuinely new HAVOK application to MV arc fault detection, undermined by a detection-time claim that contradicts the method's own window and causality requirements. read the letter →

arxiv 2502.05846 v2 pith:XGYRQOVC submitted 2025-02-09 eess.SY cs.SY

Kriti Thakur , Divyanshi Dwivedi , K. Victor Sam Moses Babu , Alivelu Manga Parimi , Prasanta K. Panigrahi , Pradeep Kumar Yemula , Pratyush Chakraborty , Mayukha Pal This is my paper · ORCID
classification eess.SYcs.SY
keywords ArcfaultdetectionHigh-impedancefaultsMediumvoltagedistributionHankelalternativeviewoftheKoopmanoperatorForcingPowersystemsafetyPSCADsimulation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that high-impedance arc faults in medium-voltage (12 kV) distribution networks can be detected much faster and more transparently than existing methods allow. It claims that the HAVOK forcing operator, a quantity extracted from the nonlinear current waveform by Koopman analysis, produces a clean, interpretable fault signature within 0.45 ms of fault onset. That would matter because arc faults are hard to see: their currents are weak, noisy, and easily confused with normal load changes or other faults. A detector that is both fast and explainable could let protection systems respond in real time and give engineers a physical reason to trust each alarm.

What carries the argument

The central object is HAVOK (Hankel alternative view of the Koopman), a data-driven method that converts a nonlinear time series into a linear model with an intermittent forcing term. Its machinery: build a Hankel matrix of 40 time-shifted current samples, take its singular value decomposition, keep the leading $r$ eigen-time-delay coordinates, and fit $\frac{d}{dt}v(t)=A v(t)+B v_r(t)$ to the first $r-1$ coordinates, where the leftover coordinate $v_r$ is the forcing operator. The forcing operator carries the nonlinear signature of the arc; bursts that push it into the arc threshold band are the detection signal, and the threshold bands are what separate arc faults, non-arcing disturbances, and other faults.

What would settle it

Re-run the HAVOK pipeline on the same 20 kHz current traces with a causal constraint: the Hankel matrix may contain only samples up to the candidate detection time, and the embedding window must span at least one 50 Hz cycle (20 ms) as the paper prescribes. If the forcing burst first crosses the arc threshold at 0.45 ms under that constraint the latency claim is confirmed; if the earliest compliant detection occurs only after the window has filled, the headline latency is not supported.

Watch

Extended reading notes

Core claim

The central claim is that the forcing operator of HAVOK distinguishes arcing from non-arcing events in a simulated 12 kV, 50 Hz distribution system. On the paper's own terms: by stacking time-shifted current samples into a Hankel matrix, applying SVD, and fitting a linear model to the leading eigen-time-delay coordinates, the last coordinate acts as an intermittent forcing term; arc faults appear as forcing bursts whose magnitude lies between ±0.06 and ±0.18, while load switching stays below ±0.045 and line-to-ground faults exceed ±0.2. The paper reports detection latencies from 0.05 ms (high-current arc) to 0.65 ms (arc with noise) across grounding surfaces and load types, with the headline low-current case detected at 0.45 ms, 99.36% faster than the 70 ms harmonic-randomness and waveform-distortion baseline, and 100% accuracy with 0% false positives over 50 simulations per scenario.

Load-bearing premise

The headline 0.45 ms detection speed assumes the fault can be read off the first nine post-fault current samples, while the paper's own embedding rule says the method needs a window spanning at least one full 20 ms power cycle; if that window is truly required, the reported latency cannot be produced by the method as described.

Editorial extensions

If this is right

  • A protection relay equipped with a HAVOK forcing-operator threshold could, in principle, trip within a fraction of a power cycle, far inside the tens-of-milliseconds windows of conventional harmonic and wavelet methods.
  • Because the fault signature is a dynamical quantity rather than a trained classifier's decision boundary, the method offers a physical explanation for each detection: the arc is an intermittent forcing event in an otherwise linear current response.
  • The same forcing-magnitude bands allow the method to reject load switching and line-to-ground faults without retraining, provided the bands hold for the network in question.
  • Operation under white Gaussian noise down to 60 dB SNR suggests the approach can tolerate realistic measurement noise, with detection delayed but still reliable.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An editorial extension: the reported 0.45 ms latency implies a decision from roughly 9 post-fault samples at 20 kHz, yet the paper's own embedding rule calls for a window of at least one 20 ms cycle; a causal, streaming implementation would need to show whether the forcing signature can be estimated reliably before the full embedding window has elapsed.
  • An editorial extension: the threshold bands (±0.06 to ±0.18 for arcs) are calibrated on simulated cases, so transferring to real feeders would likely require per-network calibration; one testable path is to derive bands from a healthy-signal baseline and then measure separation under staged faults.
  • An editorial extension: because HAVOK bursts flag any strong intermittent nonlinearity, the method may also detect incipient cable faults, transformer inrush, or other transient events; the paper's separation of arc from non-arc depends on the chosen bands holding across those event classes.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes a high-impedance arc fault (HIAF) detector for 12 kV, 50 Hz distribution systems based on Hankel alternative view of Koopman (HAVOK) analysis. A Hankel matrix is built from 40 time-delayed copies of the measured current, decomposed by SVD, and a linear regression model is fit to the leading eigen-time-delay coordinates; the last coordinate is interpreted as a forcing operator whose magnitude is thresholded to distinguish arc faults from non-arcing disturbances and other faults. The authors report detection times as low as 0.45 ms, a 99.36% improvement over the harmonic-randomness and waveform-distortion method of [37], and 100% detection accuracy with 0% false positives across 50 simulations per scenario covering different grounding surfaces, load types, and noise levels. Validation is entirely by PSCAD simulation, and the paper acknowledges the need for future hardware-in-the-loop and field testing.

Significance. If the reported performance were supported by the described algorithm, the paper would offer a practically relevant, interpretable alternative to black-box AI fault detectors, and the comparative evaluation against [37] on the same simulated data is a useful feature. The paper also honestly acknowledges its simulation-only scope and the need for real-world validation. However, the central latency claim is internally contradicted by the method's own embedding-window requirements and by the non-causal derivative computation in Algorithm 1, and the accuracy claims rest on thresholds calibrated on the same simulations used for evaluation. As presented, the main quantitative contributions are not supported, and the significance of the work is therefore substantially reduced.

major comments (3)
  1. [§3.1, §4.1, Algorithm 1] The claimed 0.45 ms detection time in Case A is not causally realizable with the described algorithm. Section 3.1 states that the total embedding window (embedding dimension times sampling interval) should span at least one 50 Hz cycle, i.e., 20 ms, and the paper uses q=40 at a 20 kHz sampling rate, which already gives a 2 ms Hankel window. Even the minimal 2 ms window exceeds the reported 0.45 ms detection latency, and the paper's own one-cycle rule implies a 20 ms window. Furthermore, Algorithm 1 step 4 computes derivatives of V using a fourth-order central difference, which requires future samples, so the forcing-coordinate value reported at t=0.20045 s depends on current data after the claimed detection time. The reported crossing is therefore an offline, post-hoc property of the full simulation record, not a causal real-time alarm. This invalidates the headline latency and the 99.36% speed comparison in Section 5 and Table 3.
  2. [§4 and Algorithm 1] The detection thresholds in Algorithm 1 (0.06–0.18 for arc faults, >0.2 for other faults, <0.045 for non-arcing disturbances) were selected after "numerous experiments" on the same simulations that are then used to report 100% accuracy and 0% false positives across 50 simulations per scenario. No held-out dataset, cross-validation procedure, or separate calibration/test split is described, and the 50 simulations per scenario are not characterized in terms of parameter distributions, seeds, or confidence intervals. The reported accuracy is therefore in-sample calibration performance, not an estimate of detection performance on unseen conditions, so the robustness claim in Section 4 is not supported.
  3. [§4.1 and Algorithm 1] The algorithm does not describe a causal, sliding-window implementation that would allow detection at a well-defined time after fault onset. The Hankel matrix in Eq. (15) is formed from the entire simulated record, the SVD is global, and the forcing operator is extracted from the full-data SVD. There is no procedure for updating the SVD or the forcing coordinate as new samples arrive, and no computational-cost analysis for real-time operation. Without such a causal formulation, the reported detection times in Table 2 are not meaningful as alarm latencies.
minor comments (6)
  1. [Eq. (4)]
  2. [Eq. (12)]
  3. [Algorithm 1, step 6]
  4. [Table 3]
  5. [Section 4.7]
  6. [References]

Circularity Check

2 steps flagged · score 6.0 of 10

Performance claims reduce to in-sample threshold calibration: 100% accuracy, 0% false positives, and the 0.45 ms / 99.36% speed headline all use Algorithm 1 thresholds fitted to the same PSCAD simulations they are then evaluated on.

  1. fitted input called prediction [Section 4 (Results and Discussion), threshold-calibration paragraph before Section 4.1; Algorithm 1 lines 8-13; accuracy statement after Table 2.]
    ""Numerous experiments were conducted, and a threshold was applied to differentiate between various fault types. For arc faults, the range of the forcing operator was measured as +0.06 to + 0.18 and -0.06 to -0.18. ... To assess detection robustness, we conducted 50 simulations for each scenario. The proposed technique attained a detection accuracy of 100% for arc faults, with 0% false positives for non-arcing disturbances.""

    The thresholds encoded in Algorithm 1 (0.06-0.18 arc, >0.2 other, <0.045 non-arcing) are not derived from theory or from an independent training/validation split; the paper states they were measured after 'numerous experiments' on the same simulations. The subsequent 100% accuracy and 0% false-positive rates are therefore a resubstitution report on the calibration data, not an out-of-sample prediction. The classifier is defined by the data it then claims to detect, making the accuracy figures forced by construction.

  2. fitted input called prediction [Section 4.1 (Case A), Table 2, and Section 5 (Comparison, Table 3).]
    ""Upon executing a simulation for 0.5 seconds with the introduction of an arc fault at 0.2 seconds, the resultant forcing operator deviation is noticed at 0.20045 seconds. This suggests that the system could identify the arc fault within a surprisingly short interval of 0.45 milliseconds." ... "the suggested method attains a detection time of 0.45 ms, which is 99.36% enhanced than the Harmonic Randomness and Waveform Distortion approach (70 ms)"."

    The 0.45 ms latency is the time at which Case A's forcing operator first crosses the arc threshold, but that threshold range was itself obtained by measuring the forcing operator in these same experiments (e.g., Case A's value -0.1035 lies inside the calibrated +/-0.06-0.18 band). Thus the headline detection speed and the 99.36% speed comparison are properties of a detector whose decision boundary was fitted to the exact trace being timed; no independent fault or held-out scenario establishes the latency. The comparison to [37] inherits this in-sample calibration.

full rationale

The HAVOK feature construction itself is not circular: it follows Brunton et al.'s externally published method, and the paper cites independent HAVOK literature for delay-embedding choices. Self-citations to [13] are not load-bearing for the main derivation. The circularity is concentrated in the decision rule and the performance metrics. Algorithm 1's thresholds (0.06-0.18 etc.) are stated as measured from the same 'numerous experiments,' and the 100% accuracy / 0% false-positive results and all detection-time values in Table 2 are computed against those fitted thresholds on the same data. The 0.45 ms speed claim and the 99.36% comparison inherit this in-sample calibration. There is also a separate causal-realizability concern (q=40 at 20 kHz implies a 2 ms embedding window, and the one-cycle rule implies 20 ms, both exceeding 0.45 ms; central-difference derivatives require future samples), but that is a correctness/timing defect rather than a circularity defect. Overall: one or more 'predictions' reduce by construction to calibrated thresholds; score 6.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central detection result depends on four practical tuning choices: the forcing thresholds, embedding dimension, model rank, and window length. Beyond those, the validity of the whole approach rests on the Mayr arc model, the HAVOK modeling assumptions, the scalar-threshold classification rule, and the fidelity of PSCAD simulations. The arc model scenario parameters (DURATION, EXTENT, OFFSET, RT) are inputs from the literature and are not treated as tuned detection parameters, but they shape the data and therefore indirectly affect all reported numbers.

free parameters (4)
  • Forcing operator thresholds = Arc: 0.06 to 0.18 and -0.06 to -0.18; non-arcing: 0.045; other faults: 0.2
    Chosen after numerous simulations (Section 4) and encoded in Algorithm 1; all reported accuracy and detection times use these thresholds on the same data that were used to set them.
  • Embedding dimension q = 40
    Hand-selected based on prior HAVOK studies and a trade-off between dynamics capture and complexity (Section 3.1). It directly controls the Hankel matrix size and the minimum embedding window.
  • Model rank r = Not specified
    Determined by an optimal hard threshold on singular values (Algorithm 1, step 3), but the threshold coefficient is not reported. The forcing coordinate and detection results depend on this choice.
  • Window length p = Dynamic residual length
    Described as dynamically established as the residual signal length after delay embedding (Section 3.1). It changes the SVD and the forcing signal, and no fixed reproducible value is given.
assumptions (4)
  • domain assumption The Mayr-type arc model with a segmented residual power function reproduces real HIAF current distortion in MV networks.
    Equations 4-12 and Section 2 rely on this model, borrowed from references [24,28], with parameters tuned to match published waveforms. No field measurements validate the model for the claimed scenarios.
  • domain assumption HAVOK analysis of a Hankel matrix yields Koopman-invariant coordinates whose last coordinate acts as an intermittent forcing term representing arc faults.
    Section 3 invokes delay embedding, SVD, and the linear-with-forcing model of [11,31,32]. Whether these assumptions hold for power-system fault currents is asserted but not demonstrated on real data.
  • ad hoc to paper The magnitude of the forcing operator alone is sufficient to distinguish arc faults from non-arcing disturbances and other fault types.
    Algorithm 1 encodes fixed thresholds on |v_r| with no physical derivation connecting forcing magnitude to fault class. The classification rule is tailored to the simulated cases.
  • domain assumption PSCAD simulation of a 12 kV distribution network with the described arc model faithfully represents real MV system behavior, including grounding surfaces and noise.
    Section 2 and Section 4 use PSCAD as ground truth. The conclusion acknowledges the absence of field testing and calls for future hardware-in-the-loop validation.
invented entities (1)
  • Koopman forcing operator as an arc fault signature
    purpose: A scalar SVD coordinate used to detect and classify arc faults by thresholding its magnitude.
    The forcing operator is a derived statistical feature from simulated current data, not an independently measured physical quantity. No field recordings or separate physical mechanism are provided to validate it outside the simulation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhanced Rapid Detection of High-impedance Arc Faults in Medium Voltage Electrical Distribution Networks." pith.science (2026). https://pith.science/paper/XGYRQOVC

@misc{pith2026250205846,
  author       = {Pith},
  title        = {Pith review of: Enhanced Rapid Detection of High-impedance Arc Faults in Medium Voltage Electrical Distribution Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XGYRQOVC}},
  note         = {Machine review of arXiv:2502.05846}
}
read the original abstract

High-impedance arc faults in AC power systems have the potential to lead to catastrophic accidents. However, significant challenges exist in identifying these faults because of the much weaker characteristics and variety when grounded with different surfaces. Previous research has concentrated predominantly on arc fault detection in low-voltage systems, leaving a significant gap in medium-voltage applications. In this work, a novel approach has been developed that enables rapid arc fault detection for medium-voltage distribution lines. In contrast to existing black-box feature-based approaches, the Hankel alternative view of the Koopman (HAVOK) analysis developed from nonlinear dynamics has been applied, which not only offers interpretable features but also opens up new application options in the area of arc fault detection. The method achieves a much faster detection speed in 0.45 ms, 99.36\% enhanced compared to harmonic randomness and waveform distortion method, thus making it suitable for real-time applications. It demonstrates the ability to detect arc faults across various scenarios, including different grounding surfaces and levels of system noise, boosting its practical importance for stakeholders in safety-critical industries.

Figures

Figures reproduced from arXiv: 2502.05846 by the authors.

Figure 1
Figure 1. Model for the MV distribution system that includes an arc fault. [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Standard structure of a HIAF model. arc current and voltage, respectively. Equation (1) could be subsequently rewritten as: dQ dg . dg dt = P  u.i P − 1  (2) 1 g dg dt = 1 g.P −1 .( dQ dg )  u.i P − 1  (3) where the arc conductance is denoted by g ( Ω −1 ). Using the expression g.P −1 .( dQ dg ) = T, the Mayr model, a widely recognized type of black-box model, could be defined as follows: lng dt = 1 T  u.i P − … view at source ↗
Figure 3
Figure 3. The waveforms of arc current observed in the 12 kV distribution system field. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Modelling of an arc fault. 3. Method The suggested approach is well-suited for examining the dynamic response and showing the nonlinear properties inherent in the electrical distribution system. To illustrate this, take a continuous-time dynamical system with a state v…
Figure 5
Figure 5. Figure 5: An overview of the suggested arc fault detection technique; HAVOK analysis [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Low current Arcing fault with RT = 1000 Ω [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 8
Figure 8. Figure 8: Arcing Fault Present when grounded in wet cement [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: 4.5. Case E: Nonarcing disturbance (Load Switching) Load switching was employed to distinguish arc faults from non-arcing disturbances. The load was altered between 0.2 and 0.3 seconds, with changes in both location and load type. A variation in the forcing operator wa…
Figure 10
Figure 10. Figure 10: Non-arcing disturbance (load switching) [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]
Figure 11
Figure 11. Figure 11: Figure11 [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]
Figure 12
Figure 12. Figure 12: Low current Arcing fault with RT = 1000 Ω with IM load [PITH_FULL_IMAGE:figures/full_fig_p023_12.png]
Figure 14
Figure 14. Figure 14: Low current Arcing fault with RT = 1000 Ω with noise [PITH_FULL_IMAGE:figures/full_fig_p024_14.png]
Figure 15
Figure 15. Figure 15: Arcing fault current if wave￾forms for varying resistance values: (a) RT = 1000 Ω ( high resistance), (b) RT = 500 Ω (moderate resistance), (c) RT = 200 Ω (lower resistance), and (d) RT = 50 Ω (wet soil, low resistance). 24 [PITH_FULL_IMAGE:figures/full_fig_p024_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 35 canonical work pages

  1. [37]

    M. Wei, F. Shi, H. Zhang, Z. Jin, V. Terzija, J. Zhou, H. Bao, High impedance arc fault detection based on the harmonic randomness and waveform distortion in the distribution system, IEEE Transactions on Power Delivery 35 (2) (2020) 837–850.doi:10.1109/TPWRD.2019. 2929329. 31

    work page doi:10.1109/tpwrd.2019 2020

  2. [1]

    IEA, Electricity market report, Electron Marketpl Rep (2022)

    I. IEA, Electricity market report, Electron Marketpl Rep (2022)

    work page 2022

  3. [2]

    M. Pal, A. K. Bharati, S. S. Belkhode, Nuisance trip decision manage- ment using data analytics in electrical protection system (05 2022)

    work page 2022

  4. [3]

    M. Pal, A. K. Bharati, S. S. Belkhode, Data analytics for smart electrical protection systems (11 2021)

    work page 2021

  5. [4]

    X. Wang, J. Gao, X. Wei, G. Song, L. Wu, J. Liu, Z. Zeng, M. Kheshti, High impedance fault detection method based on variational mode de- compositionandteager–kaiserenergyoperatorsfordistributionnetwork, IEEE Transactions on Smart Grid 10 (6) (2019) 6041–6054

    work page 2019

  6. [5]

    X. Zeng, W. Gao, G. Yang, High impedance fault detection in distribu- tion network based on s-transform and average singular entropy, Global Energy Interconnection 6 (1) (2023) 64–80

    work page 2023

  7. [6]

    Hyun, J.-S

    S.-Y. Hyun, J.-S. Hong, S.-Y. Yun, C.-H. Kim, Y. Lee, Arc modeling and kurtosis detection of fault with arc in power distribution networks, Applied Sciences 12 (6) (2022) 2777

    work page 2022

  8. [7]

    Zhang, Y

    W. Zhang, Y. Jing, X. Xiao, Model-based general arcing fault detec- tion in medium-voltage distribution lines, IEEE Transactions on Power Delivery 31 (5) (2016) 2231–2241

    work page 2016

Show all 37 references
  1. [8]

    Kulkarni, S

    S. Kulkarni, S. Santoso, T. A. Short, Incipient fault location algorithm for underground cables, IEEE transactions on smart grid 5 (3) (2014) 1165–1174

    2014

  2. [9]

    Sedighizadeh, A

    M. Sedighizadeh, A. Rezazadeh, N. I. Elkalashy, et al., Approaches in high impedance fault detection a chronological review, Advances in Elec- trical and Computer Engineering 10 (3) (2010) 114–128. 28

    2010

  3. [10]

    Korda, I

    M. Korda, I. Mezić, Linear predictors for nonlinear dynamical sys- tems: Koopman operator meets model predictive control, Automatica 93 (2018) 149–160

    2018

  4. [11]

    S. L. Brunton, B. W. Brunton, J. L. Proctor, E. Kaiser, J. N. Kutz, Chaos as an intermittently forced linear system, Nature communications 8 (1) (2017) 19

    2017

  5. [12]

    J. Tu, C. Rowley, D. Luchtenburg, S. Brunton, J. Kutz, On dynamic mode decomposition: Theory and applications, Journal of Computa- tional Dynamics 1 (11 2013).doi:10.3934/jcd.2014.1.391

    2013 doi

  6. [13]

    Dwivedi, P

    D. Dwivedi, P. K. Yemula, M. Pal, Dynamopmu: A physics informed anomaly detection, clustering and prediction method using non-linear dynamics onµpmu measurements, IEEE Transactions on Instrumenta- tion and Measurement (2023)

    2023

  7. [14]

    J. Yang, J. Zhao, J. Song, J. Wu, C. Zhao, H. Leng, A hybrid method using havok analysis and machine learning for predicting chaotic time series, Entropy 24 (3) (2022) 408

    2022

  8. [15]

    R. G. Moulder, E. Martynova, S. M. Boker, Extracting nonlinear dy- namics from psychological and behavioral time series through havok analysis, Multivariate Behavioral Research 58 (2) (2023) 441–465

    2023

  9. [16]

    J. Li, G. Zou, W. Wang, N. Shao, B. Han, L. Wei, Low-voltage series arc fault detection based on ecmc and vb-scn, Electric Power Systems Research 218 (2023) 109222

    2023

  10. [17]

    B. Wang, J. Geng, X. Dong, High-impedance fault detection based on nonlinear voltage–current characteristic profile identification, IEEE Transactions on Smart Grid 9 (4) (2016) 3783–3791

    2016

  11. [18]

    Modified fft based high impedance fault detection technique considering distribution non-linear loads: Simulation and experimental data anal- ysis, International Journal of Electrical Power and Energy Systems 94 (2018) 124–140

    2018

  12. [19]

    Y. Wang, D. Sheng, H. Hu, K. Han, J. Zhou, L. Hou, A novel series arc fault detection method based on mel-frequency cepstral coefficients and fully connected neural network, IEEE Access 10 (2022) 97983–97994. 29

    2022

  13. [20]

    N. I. Elkalashy, M. Lehtonen, H. A. Darwish, M. A. Izzularab, A.-m. I. Taalab, Modeling and experimental verification of high impedance arc- ing fault in medium voltage networks, IEEE Transactions on Dielectrics and Electrical Insulation 14 (2) (2007) 375–383.doi:10.1109/TDEI....

    2007

  14. [21]

    Ghaderi, H

    A. Ghaderi, H. A. Mohammadpour, H. L. Ginn, Y.-J. Shin, High- impedance fault detection in the distribution network using the time- frequency-based algorithm, IEEE Transactions on Power Delivery 30 (3) (2014) 1260–1268

    2014

  15. [22]

    H. Lala, S. Karmakar, Detection and experimental validation of high impedance arc fault in distribution system using empirical mode de- composition, IEEE Systems Journal 14 (3) (2020) 3494–3505

    2020

  16. [23]

    L. Wang, H. Qiu, P. Yang, L. Mu, Arc fault detection algorithm based on variational mode decomposition and improved multi-scale fuzzy entropy, Energies 14 (14) (2021) 4137

    2021

  17. [24]

    M. Wei, W. Liu, F. Shi, H. Zhang, Z. Jin, W. Chen, Distortion- controllablearcmodelingforhighimpedancearcfaultinthedistribution network, IEEE Transactions on Power Delivery 36 (1) (2020) 52–63

    2020

  18. [25]

    Y. Wang, F. Zhang, X. Zhang, S. Zhang, Series ac arc fault detection method based on hybrid time and frequency analysis and fully connected neural network, IEEE Transactions on Industrial Informatics 15 (12) (2018) 6210–6219

    2018

  19. [26]

    N. Qu, J. Wang, J. Liu, An arc fault detection method based on current amplitude spectrum and sparse representation, IEEE Transactions on Instrumentation and Measurement 68 (10) (2019) 3785–3792.doi:10. 1109/TIM.2018.2880939

    2019

  20. [27]

    N. Qu, J. Wang, J. Liu, An arc fault detection method based on current amplitude spectrum and sparse representation, IEEE Transactions on Instrumentation and Measurement 68 (10) (2018) 3785–3792

    2018

  21. [28]

    survey of calculation methods and application guide, Electra 118 (1988) 63–79

    W.CIGRE,13.01.practicalapplicationofarcphysicsincircuitbreakers. survey of calculation methods and application guide, Electra 118 (1988) 63–79. 30

    1988

  22. [29]

    P. H. Schavemaker, L. Van der Slui, An improved mayr-type arc model based on current-zero measurements [circuit breakers], IEEE Transac- tions on Power delivery 15 (2) (2000) 580–584

    2000

  23. [30]

    Guardado, S

    J. Guardado, S. Maximov, E. Melgoza, J. Naredo, P. Moreno, An im- proved arc model before current zero based on the combined mayr and cassie arc models, IEEE Transactions on Power Delivery 20 (1) (2005) 138–142

    2005

  24. [31]

    S. Qian, C. A. Chou, A koopman-operator-theoretical approach for anomaly recognition and detection of multi-variate eeg system, Biomed- ical Signal Processing and Control 69 (2021) 102911

    2021

  25. [32]

    P. K. Huynh, A. R. Setty, T. B. Le, T. Q. Le, A noise-robust koop- man spectral analysis of an intermittent dynamics method for complex systems: a case study in pathophysiological processes of obstructive sleep apnea, IISE Transactions on Healthcare Systems Engineering 0 (0) (...

    2022

  26. [33]

    B. SL, B. BW, P. JL, Chaos as an intermittently forced linear system, Nat Commun 8 (2017)

    2017

  27. [34]

    V. V. Terzija, H.-J. Koglin, On the modeling of long arc in still air and arc resistance calculation, IEEE Transactions on Power Delivery 19 (3) (2004) 1012–1017

    2004

  28. [35]

    Gaudreau, B

    A. Gaudreau, B. Koch, Evaluation of lv and mv arc parameters, IEEE Transactions on Power Delivery 23 (1) (2007) 487–492

    2007

  29. [36]

    Ghaderi, H

    A. Ghaderi, H. L. Ginn III, H. A. Mohammadpour, High impedance fault detection: A review, Electric power systems research 143 (2017) 376–388

    2017

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.