Graph Attention Networks with Physical Constraints for Anomaly Detection

Andres Caro; Iman Khazrak; Mar Avila; Mohammadhossein Homaei; Ruben Molano

REVIEW 5 major objections 6 minor 15 references

Graph Attention Networks with Physical Constraints for Anomaly Detection

T0 review · 5 major / 6 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Physics-GAT detects attacks on water-distribution networks with F1=0.979, a 3.3-point gain over the best model-based baseline, by converting normalized mass and energy balance residuals into graph attention features.

desk verdict Solid applied paper with a genuine result on BATADAL, but the robustness story is thinner than the abstract suggests; the core method and ablations justify a round of peer review. read the letter →

arxiv 2601.12426 v1 pith:GDSW243S submitted 2026-01-18 cs.LG cs.CR

Mohammadhossein Homaei , Iman Khazrak , Ruben Molano , Andres Caro , Mar Avila This is my paper · ORCID

classification cs.LGcs.CR

keywords graphattentionnetworksphysics-informedlearninganomalydetectioncyber-physicalsystemswaterdistributionconservationlawsmulti-scaleBATADAL

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

The paper tries to show that explicitly feeding hydraulic conservation-law violations into a graph neural network gives better, more interpretable, and more robust anomaly detection in water distribution systems than either pure data-driven models or pure model-based residual checks. On the BATADAL benchmark, the proposed Physics-GAT reaches F1=0.979 versus 0.946 for the best model-based baseline, with non-overlapping confidence intervals and a 10.6% faster median time to detection. Ablations show the physics-informed features are the single most important component, and robustness tests indicate the normalized residuals keep performance above 0.95 under ±15% pipe-roughness error, where the baseline falls below 0.83. If these claims hold, physics-informed deep learning could close the gap between accuracy and interpretability for critical water infrastructure monitoring.

What carries the argument

The key machinery is the normalized conservation-law violation feature: the mass balance residual φ_mass(t) = |ΣQ_in − ΣQ_out − D_i| / (ΣQ + ε) at each node, and the energy gradient residual φ_energy(t) = |(p_i+z_i) − (p_j+z_j) − h_L(Q)| / max(p_i+z_i, p_j+z_j) along each edge. These fractions map the physical laws of continuity and energy conservation into scale-free inputs that a graph attention network can learn to reweight through learned attention coefficients, combined with a bidirectional LSTM for temporal context and a multi-scale (node/cluster/network) adaptive fusion for localization. The attention mechanism is what allows the model to adapt to parameter shifts—systematic biases in

What would settle it

Run the same detector on a network where demand estimates are deliberately biased in a pattern spatially correlated with the attack (for example, overestimating demand in the attacked zone) while keeping the uniform ±15% roughness error the paper tested; if the F1 margin over the model-based baseline shrinks to near zero or reverses, the normalization claim is falsified. More directly, compute the Spearman correlation between the mass-balance residual and the ground-truth attack label under such correlated bias—if it drops while raw SCADA still carries signal, the physics features are encoding

Watch

Extended reading notes

Core claim

The central claim is that normalized residuals of mass and energy conservation—scaled by local inflow and piezometric head—are stable, transferable, and attack-specific signals when processed by a graph attention network with bidirectional LSTM temporal fusion. On BATADAL, this combination yields F1=0.979 with a 95% confidence interval of [0.971, 0.986], a 3.3 percentage point improvement over the leading model-based detector (B1), a large effect size (Cohen's d=1.24), and a reduced time-to-detection of 1.44 hours. The residuals generalize zero-shot across three other network topologies with F1 above 0.91, and the GAT attention weights correlate with hydraulic shortest paths (Spearman ρ=0.81

Load-bearing premise

The physics features rely on estimated demand D_i(t) and assumed Hazen-Williams roughness coefficients, and the whole robustness story rests on the claim that normalizing the residuals removes enough of that parametric error so that the remaining signal reflects the attack rather than model error; the paper itself concedes in Section 5 that epistemic uncertainty from estimated demand remains.

Editorial extensions

If this is right

A water utility could deploy a single model trained on one network and transfer it to another topology with minimal fine-tuning, since physics residuals transfer zero-shot with F1 above 0.91 in the paper's multi-network tests.
Operators could trust the detector under realistic hydraulic uncertainty: ±15% error in pipe roughness—typical of aging infrastructure—causes less than a 3-point drop in F1, compared to a 12.7-point drop for hard-threshold residuals.
Detection speed matters for limiting contamination or service loss: the median time-to-detection drops from 1.61 to 1.44 hours, a 10-minute speed-up in responding to attacks.
The model's explanations are usable: attention weights trace hydraulic propagation paths (ρ=0.81) and physics-attribution identifies whether mass or energy inconsistency triggers the alarm, supporting operator diagnosis.
Scaling to larger networks (beyond a few hundred nodes) requires mini-batch inference, since the per-graph cost grows with the number of edges and attention heads, but the paper estimates roughly 300 ms for 500-node networks on a mid-range GPU.

Reading between the lines

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

The normalization trick is the load-bearing idea: if it holds, the same physics-residual formulation could be adapted to other utility networks governed by conservation laws—gas pipelines, power grids, or district heating—where model-based residual checks currently suffer from parameter sensitivity.
The paper's robustness to uniform parameter error does not guarantee robustness to spatially biased error; a targeted test where roughness or demand is misestimated only in the attacked zone would separate model error from attack signal and is a natural next experiment.
Because the mass-balance residual also flags physical losses, the approach might detect leaks and equipment degradation in addition to cyber-attacks, though the paper only evaluates attack scenarios.
Combining the physics residual with a demand-forecasting module that provides a tighter prior on D_i(t) would reduce the epistemic uncertainty the paper concedes in Section 5, potentially extending the gains to poorly calibrated networks.

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.

Desk Editor's Note

read the letter

Short version: a genuine, mostly well-executed application of physics-informed GNNs to water-distribution anomaly detection, with a new result on BATADAL and an honest ablation. The main claims hold up internally, but the robustness/transfer story is oversold at the edges — the zero-shot label is not really zero-shot, and the uncertainty tests only cover zero-centered noise. Still worth refereeing.

The actual new thing here is the specific construction: normalized conservation-law residuals (mass and energy) as node features in a GAT+BiLSTM with multi-scale fusion. Each piece is standard, but the combination is not in the cited literature, and the ablation suggests the physics features are what buy the F1 gain (removing both drops F1 from 0.979 to 0.930, a solid 5%). The central result, F1=0.979 vs B1's 0.946, is reported with bootstrap intervals and a Wilcoxon test, and the intervals don't overlap. The attention-alignment check (Spearman 0.81 with hydraulic paths) and the case study are nice touches. They also state a real limitation in §5 about demand estimation, which counts in their favor.

The soft spots are real but not fatal. The robustness tables (4 and 5) have no error bars, and the demand ±20% result in §5 is asserted without a table. RF and SVM are listed as baselines in §4.1 but never appear in the results. The 'zero-shot transfer' label overstates the protocol: PI features consume target topology and approximate parameters, so it's not fully blind. The stress-test concern about correlated parameter error is legitimate — the robustness tests are zero-centered, so if demand or roughness bias tracks attack signatures, the residual features could encode model error rather than anomaly. But the authors acknowledge the epistemic uncertainty, and this concern attacks the robustness generalization paragraph, not the core benchmark comparison. On the main BATADAL result, the paper is internally coherent.

Who's this for: applied ML researchers and water-utility cyber-physical security folks. It won't change how we think about GNNs, but it is a careful, reproducible (code linked, though no commit hash) contribution to a real problem. I'd send it to peer review with requests to add error bars to the robustness tables, actually include RF/SVM results, provide the demand-bias table, and soften 'zero-shot' to 'topology-aware transfer.'

Referee Report

5 major / 6 minor

Summary. The paper proposes Physics-GAT, a graph attention network with bidirectional LSTM for anomaly detection in water distribution systems. It augments raw SCADA data with physics-informed features—mass-balance and energy-gradient residuals—and uses a multi-scale fusion of node, cluster, and global scores. On the BATADAL benchmark, the method achieves F1=0.979, a 3.3 percentage-point gain over the model-based baseline B1, with non-overlapping bootstrap confidence intervals, lower time-to-detection (1.44 h vs 1.61 h), and claimed robustness to ±15% Hazen–Williams roughness perturbations. The paper also reports ablations, cross-network transfer experiments, and explainability analyses.

Significance. If the reported results are fully supported, the contribution is valuable: it demonstrates that explicit conservation-law violations, combined with topology-aware deep learning, can outperform both model-based and purely data-driven baselines on a standard benchmark, while providing interpretable attention and physics-based diagnostics. The availability of code, bootstrap CIs, and ablation controls strengthens reproducibility. However, the significance is conditional on the robustness and generalization claims, which currently rest on incomplete evidence—most notably the absence of uncertainty quantification in the parameter-perturbation experiments and the lack of a structured demand-bias test.

major comments (5)

[§4.4, Table 5] The central robustness claim (F1 degradation <3% under ±15% C_ij error) is reported only as point estimates, with no bootstrap CIs or significance tests, unlike Table 1. The §5 sentence about ±20% demand deviation ('drop is only 2.1 pp vs 9.9 for B1') has no corresponding table or experimental description at all. Since robustness is a headline contribution, the authors should provide full tables with CIs and significance tests for both roughness and demand perturbations, and specify how the demand perturbation was generated.
[Eq. (1)–(2), §5] The PI features in Eq. (1) depend on estimated demand D_i(t), and Eq. (2) assumes known C_ij. The robustness tests in Table 5 and the §5 demand statement use zero-centered, independent perturbations. The manuscript itself acknowledges in §5 that 'phi_mass depends on estimated demand, so some epistemic uncertainty remains,' but it never tests the more dangerous case where demand- or roughness-error is biased in a way that correlates with attack signatures. Without such a test, the claim that normalized residuals encode true attacks rather than model error is not established. Please add a structured-bias experiment (e.g., demand underestimation during high-consumption events) or provide an argument with quantitative bounds.
[§4.1, Table 1] The setup lists RF and SVM as graph-free baselines, but they never appear in Table 1 or anywhere else in the results. As the paper claims to outperform 'all baselines,' this omission is significant. Either include their results or explicitly state that they were omitted and why. Relatedly, Table 4 reports fusion-configuration F1 values without any error bars; the 'adaptive fusion' gain of ~1.6 pp should be accompanied by CIs.
[§3.5, Table 2] The multi-network validation uses D-Town, L-Town, and Modena from the 2024 Multi-Network dataset, but the manuscript gives no details on the attack scenarios, label definitions, sensor layouts, or how 'approximate parameters (C_ij, D_i from averages)' were derived. The zero-shot transfer claim is further supported only by a sentence that 'when using only raw SCADA (no PI), F1 falls to 0.83–0.87'—no table or CI is provided. Please add dataset descriptions and include the no-PI comparison in Table 2, ideally with the same bootstrap CIs as the other rows.
[Eq. (4), §3.4, Table 3] The same phi_mass and phi_energy values appear both as input features and inside the physics regularizer L_physics. The ablation 'W/o phi_mass' removes the feature, but if L_physics still computes max(phi_mass, phi_energy) on normal samples, the physics signal is not fully removed. This confounds the conclusion that 'PI features are the most critical component.' Please clarify exactly what is removed in each ablation row and, ideally, report an ablation where both the input feature and the regularizer term are removed simultaneously.

minor comments (6)

[Abstract, §3.2] The term 'normalized conservation law violations' is used, but Eq. (1) divides by total inflow plus epsilon; Eq. (2) divides by max pressure head. This is a scaling, not a normalization that fully removes parameter error. Please define 'normalized' precisely to avoid overstating the invariance.
[Eq. (2)] The Hazen–Williams head loss h_L(Q_ij) is never defined in the text. Please provide the formula and the units/parameters L_ij, D_ij, C_ij.
[§4.2.1] The sentence 'This explains a 9.6% F1 gap under extreme uncertainty (0.954 vs 0.828)' appears inconsistent with Table 5, which shows 0.951 vs 0.828 at δ=-15% (a 12.3 pp gap). Please reconcile the numbers.
[Table 3] The row 'W/o normalization' shows a larger drop (−6.3%) than 'W/o both phi' (−5.0%). Since normalization presumably applies to phi features, the ordering is counterintuitive. Please clarify what 'normalization' removes and how it interacts with the phi features.
[§4.5] The Spearman correlation ρ=0.81±0.09 is reported with 'p<0.001', but the number of samples and the construction of the confidence interval are not given. Please report the sample size and the CI method.
[General] Minor typos: 'practica' at the end of §3.5; 'HV AC' in the Related Work; the reference list has inconsistent formatting. Also, Figures 2–4 are referenced in the appendix but not present in the main text; ensure they are properly embedded.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: empirical supervised evaluation with internal ablations; the shared physics residuals are redundancy, not definitional collapse.

full rationale

The paper does not derive its headline numbers from the physics features by construction. Physics-GAT is a supervised model trained on BATADAL labels; F1=0.979, TTD=1.44h, and the multi-network transfer numbers are test-set measurements, not outputs entailed by the input equations. The PI features in Eqs. (1)-(2) are hand-crafted physical residuals; the model learns to map them (plus raw SCADA) to labels. The ablation in Table 3 provides an internal control: removing both phi features drops F1 by 5.0%, so the physics signal is empirically load-bearing rather than definitionally identical to the target. The L_physics regularizer in Eq. (4) does reuse the same phi residuals on normal samples, but this is redundancy between two training signals, not a prediction that reduces to its input; the W/o L_physics ablation quantifies its small marginal contribution. The self-citation [3] (Homaei et al., digital twins) is a background reference for SCADA-based protection and is not load-bearing; no uniqueness theorem and no ansatz is imported from the authors' prior work. The robustness claim is vulnerable to the acknowledged limitation in Section 5 ('the phi_mass term depends on estimated demand, so some epistemic uncertainty remains even after normalization') if demand/roughness errors are biased in an attack-correlated way, but that is an external-validity concern, not a circular derivation. The deferred 'additional tuning' appendix is absent from the provided manuscript, a reporting gap, but not circular. Therefore no circular step is present.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The empirical pipeline assumes that mass/energy residual features carry genuine anomaly signal, that labeled BATADAL attacks generalize, and that unmeasured-node interpolation preserves hydraulic meaning. These are domain assumptions rather than derived results; the only hand-set parameters are the regularization weights and interpolation bandwidth.

free parameters (2)

regularization weights lambda_p, lambda_c = lambda_p=0.1, lambda_c=0.05
Tuned in experiments (Section 3.4 and 4.1); the composite loss L = L_BCE + lambda_p * L_physics + lambda_c * L_consist depends on these values, and the ablation for L_physics shows a 1.5% F1 impact.
interpolation bandwidth sigma = sigma=2
Ad hoc choice in pressure-driven interpolation for unmeasured nodes (Section 3.2). The feature vectors of unmeasured nodes are built from inverse-distance weighting with this bandwidth, so full-graph performance depends on it.

assumptions (5)

domain assumption Hazen-Williams head-loss formula h_L(Q_ij) with known pipe length, diameter, and roughness C_ij describes the true hydraulics.
Used in Eq. (2) to define the energy-gradient violation and in all residual computations; the robustness experiments vary C_ij but still assume the functional form is correct.
domain assumption Estimated demand D_i(t) and roughness C_ij are available or approximable, and normalization removes enough parameter error to keep residuals informative.
The mass-balance residual in Eq. (1) contains D_i(t); the authors' stated limitation in Section 5 acknowledges epistemic uncertainty in this quantity.
domain assumption BATADAL and the Multi-Network dataset provide representative, correctly labeled attacks, and similar attack types across networks justify transfer.
Section 3.5 and 4.2.2 rely on these datasets for the F1 claims and zero-shot transfer; no external real-world validation is provided.
ad hoc to paper Inverse-distance interpolation with sigma=2 preserves hydraulic continuity for unmeasured nodes.
Section 3.2 states 'This maintains hydraulic continuity' but no independent evidence or sensitivity analysis for sigma is supplied.
domain assumption Graph attention coefficients learned from labeled attacks correspond to hydraulic propagation paths.
The explainability validation (Spearman rho=0.81, Section 4.5) assumes that high attention weights on physical pipe edges are meaningful; the authors note 2 of 14 attacks violate this.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Graph Attention Networks with Physical Constraints for Anomaly Detection." pith.science (2026). https://pith.science/paper/GDSW243S

@misc{pith2026260112426,
  author       = {Pith},
  title        = {Pith review of: Graph Attention Networks with Physical Constraints for Anomaly Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GDSW243S}},
  note         = {Machine review of arXiv:2601.12426}
}

abstract

Water distribution systems (WDSs) face increasing cyber-physical risks, which make reliable anomaly detection essential. Many data-driven models ignore network topology and are hard to interpret, while model-based ones depend strongly on parameter accuracy. This work proposes a hydraulic-aware graph attention network using normalized conservation law violations as features. It combines mass and energy balance residuals with graph attention and bidirectional LSTM to learn spatio-temporal patterns. A multi-scale module aggregates detection scores from node to network level. On the BATADAL dataset, it reaches $F1=0.979$, showing $3.3$pp gain and high robustness under $15\%$ parameter noise.

Figures

Figures reproduced from arXiv: 2601.12426 by the authors.

**Figure 1.** Physics-GAT workflow: physics-informed features feed a GAT-BiLSTM core and an adaptive multi-scale fusion that produces [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Physics violation heatmap for Attack #3. J42 (red circle) [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** GAT attention flow during Attack #3. High attention [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Temporal profile at J42. Mass violation detected [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 1 linked inside Pith

[1]

Battle of the Attack Detection Algorithms: Disclosing Cyber Attacks on Wa- ter Distribution Networks,

R. Taormina, S. Galelli, N. O. Tippenhauer, et al., “Battle of the Attack Detection Algorithms: Disclosing Cyber Attacks on Wa- ter Distribution Networks,”Journal of Water Resources Plan- ning and Management, vol. 144, no. 8, p. 04018048, Aug. 2018

2018
[2]

At- tack detection in water distribution systems using machine learning,

D.T. Ramotsoela, G.P. Hancke, and A.M. Abu-Mahfouz, “At- tack detection in water distribution systems using machine learning,”Human-centric Computing and Information Sciences, vol. 9, no. 1, 2019

2019
[3]

arXiv:2412.06694 (2024)

Homaei, M.H., Di Bartolo, A.J., ´Avila, M., Mogoll´on-Guti´errez, ´O., Caro, A.: Digital Transformation in the Water Distribution System based on the Digital Twins Concept. arXiv:2412.06694 (2024)

arXiv 2024
[4]

Deep-learning approach to the de- tection and localization of cyber-physical attacks on water dis- tribution systems,

R. Taormina and S. Galelli, “Deep-learning approach to the de- tection and localization of cyber-physical attacks on water dis- tribution systems,”Journal of Water Resources Planning and Management, vol. 144, no. 10, p. 04018065, Oct. 2018

2018
[5]

A CV AE-Based Anomaly Detection Algorithm for Cyber Physical Attacks for Water Dis- tribution Systems,

H. H. Addeen, Y . Xiao, and T. Li, “A CV AE-Based Anomaly Detection Algorithm for Cyber Physical Attacks for Water Dis- tribution Systems,”IEEE Access, vol. 12, pp. 48321–48334, 2024

2024
[6]

Ex- plainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI,

A. Barredo Arrieta, N. D ´ıaz-Rodr´ıguez, J. Del Ser, et al., “Ex- plainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI,”Informa- tion Fusion, vol. 58, pp. 82-115, 2020

2020
[7]

Model-based approach for cyber- physical attack detection in water distribution systems,

M. Housh and Z. Ohar, “Model-based approach for cyber- physical attack detection in water distribution systems,”Water Research, vol. 139, pp. 132-143, Jun. 2018

2018
[8]

Physics and Chemistry of the Earth, Parts A/B/C134, 103558 (2024)

Kanyama, M.N., Bhunu Shava, F., Gamundani, A.M., Hart- mann, A.: Machine learning applications for anomaly detec- tion in Smart Water Metering Networks: A systematic review. Physics and Chemistry of the Earth, Parts A/B/C134, 103558 (2024)

2024

Show all 15 references

[9]

Deep H2O: Cyber attacks detection in water distribution systems using deep learn- ing,

M.N. Sikder, S. Islam, and M.A. Rahman, “Deep H2O: Cyber attacks detection in water distribution systems using deep learn- ing,”Journal of Water Process Engineering, vol. 52, p. 103568, Apr. 2023

2023
[10]

Evolution of Com- plex Network Topologies in Urban Water Infrastructure,

J. Zischg, C. Klinkhamer, X. Zhan, et al., “Evolution of Com- plex Network Topologies in Urban Water Infrastructure,” in Proceedings of the World Environmental and Water Resources Congress 2017, May 18, 2017

2017
[11]

N. M. Nejad, “Building Resilient, Adaptive, and Integrated Electric Vehicle (EV) Supply Chains for Sustainable Perfor- mance: Geopolitical Disruption, Digital Transformation, and Circular Supply Chain Practices,” Ph.D. dissertation, The Uni- versity of Toledo, 2025

2025
[12]

Large-Scale Multipurpose Benchmark Datasets for Assessing Data-Driven Deep Learning Approaches for Water Distribution Networks,

A. Tello, H. Truong, A. Lazovik, and V . Degeler, “Large-Scale Multipurpose Benchmark Datasets for Assessing Data-Driven Deep Learning Approaches for Water Distribution Networks,” in *The 3rd International Joint Conference on Water Distribu- tion Systems Analysis and Computing...

2024
[13]

Graph attention networks,

P. Veli ˇckovi´c, G. Cucurull, A. Casanova, A. Romero, P. Li `o, and Y . Bengio, “Graph attention networks,” inProc. 6th Inter- national Conference on Learning Representations (ICLR), Van- couver, BC, Canada, Apr. 2018

2018
[14]

Fast unfolding of communities in large networks,

V .D. Blondel, J.-L. Guillaume, R. Lambiotte, and E. Lefebvre, “Fast unfolding of communities in large networks,”Journal of Statistical Mechanics, vol. 2008, p. P10008, 2008

2008
[15]

Explain- ing anomalies detected by autoencoders using Shapley Additive Explanations,

L. Antwarg, R.M. Miller, B. Shapira, and L. Rokach, “Explain- ing anomalies detected by autoencoders using Shapley Additive Explanations,”Expert Systems with Applications, vol. 186, p. 115736, 2021. Appendix Figure 2. Physics violation heatmap for Attack #3. J42 (red circle) s...

2021

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [1]

Battle of the Attack Detection Algorithms: Disclosing Cyber Attacks on Wa- ter Distribution Networks,

R. Taormina, S. Galelli, N. O. Tippenhauer, et al., “Battle of the Attack Detection Algorithms: Disclosing Cyber Attacks on Wa- ter Distribution Networks,”Journal of Water Resources Plan- ning and Management, vol. 144, no. 8, p. 04018048, Aug. 2018

2018

[2] [2]

At- tack detection in water distribution systems using machine learning,

D.T. Ramotsoela, G.P. Hancke, and A.M. Abu-Mahfouz, “At- tack detection in water distribution systems using machine learning,”Human-centric Computing and Information Sciences, vol. 9, no. 1, 2019

2019

[3] [3]

arXiv:2412.06694 (2024)

Homaei, M.H., Di Bartolo, A.J., ´Avila, M., Mogoll´on-Guti´errez, ´O., Caro, A.: Digital Transformation in the Water Distribution System based on the Digital Twins Concept. arXiv:2412.06694 (2024)

arXiv 2024

[4] [4]

Deep-learning approach to the de- tection and localization of cyber-physical attacks on water dis- tribution systems,

R. Taormina and S. Galelli, “Deep-learning approach to the de- tection and localization of cyber-physical attacks on water dis- tribution systems,”Journal of Water Resources Planning and Management, vol. 144, no. 10, p. 04018065, Oct. 2018

2018

[5] [5]

A CV AE-Based Anomaly Detection Algorithm for Cyber Physical Attacks for Water Dis- tribution Systems,

H. H. Addeen, Y . Xiao, and T. Li, “A CV AE-Based Anomaly Detection Algorithm for Cyber Physical Attacks for Water Dis- tribution Systems,”IEEE Access, vol. 12, pp. 48321–48334, 2024

2024

[6] [6]

Ex- plainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI,

A. Barredo Arrieta, N. D ´ıaz-Rodr´ıguez, J. Del Ser, et al., “Ex- plainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI,”Informa- tion Fusion, vol. 58, pp. 82-115, 2020

2020

[7] [7]

Model-based approach for cyber- physical attack detection in water distribution systems,

M. Housh and Z. Ohar, “Model-based approach for cyber- physical attack detection in water distribution systems,”Water Research, vol. 139, pp. 132-143, Jun. 2018

2018

[8] [8]

Physics and Chemistry of the Earth, Parts A/B/C134, 103558 (2024)

Kanyama, M.N., Bhunu Shava, F., Gamundani, A.M., Hart- mann, A.: Machine learning applications for anomaly detec- tion in Smart Water Metering Networks: A systematic review. Physics and Chemistry of the Earth, Parts A/B/C134, 103558 (2024)

2024