arxiv: 2605.13390 · v1 · submitted 2026-05-13 · 📡 eess.SY · cs.SY

Recognition: no theorem link

Sensitivity Quantification for Distribution System State Estimation

Bet\"ul Mamudi , Jochen Stiasny , Jochen Cremer

Authors on Pith no claims yet

Pith reviewed 2026-05-14 18:29 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords distribution system state estimationpseudo-measurementsuncertainty quantificationCramér-Rao boundweighted least squaresFisher information matrixsensitivity analysispower system monitoring

0 comments

The pith

The choice of pseudo-measurement distribution directly distorts the confidence limits produced by weighted least squares in distribution system state estimation.

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

This paper tests whether the uncertainty bounds reported by standard weighted least squares distribution system state estimation change when the probability distribution assigned to pseudo-measurements is altered. It introduces a diagnostic that computes the ratio between the true Cramér-Rao Bound and the bound implied by the WLS solution at convergence. Experiments on the CIGRE medium-voltage network across 100 operating points show that heavy-tailed and skewed distributions cause WLS to report overly wide bounds, with the size of the mismatch differing by bus and scenario. The same diagnostic remains insensitive to shifts in the mean of the pseudo-measurements. The results establish that any uncertainty-aware DSSE method must treat the specific distributional form of pseudo-measurements as an explicit input rather than a fixed assumption.

Core claim

Varying pseudo-measurement distributions across 22 shape variants while holding spread constant produces systematic differences between the true Cramér-Rao Bound and the bound assumed under WLS; heavy-tailed and skewed cases cause WLS to overstate its uncertainty, the degree of overstatement is bus- and scenario-dependent, and the ratio diagnostic fails to detect mean-shift bias.

What carries the argument

A per-bus, per-scenario ratio of the true Cramér-Rao Bound to the WLS-assumed bound, obtained from the Fisher Information Matrix evaluated at the converged WLS solution.

If this is right

Uncertainty bounds reported by existing WLS DSSE implementations must be corrected once the actual pseudo-measurement distribution is known.
Diagnostic ratios should be computed routinely to flag locations where WLS confidence limits are miscalibrated.
Variance-only uncertainty diagnostics are insufficient when pseudo-measurements carry mean bias or heavy tails.
Uncertainty-aware DSSE methods require an explicit step that accounts for distribution shape rather than treating it as a fixed input.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same shape sensitivity may appear in other state estimators that rely on second-order approximations of the measurement model.
Real-time control applications that use DSSE confidence intervals for constraint enforcement could become unreliable under realistic pseudo-measurement distributions.
Extending the diagnostic to time-series or multi-time-step formulations would test whether the observed mismatches accumulate or average out over time.

Load-bearing premise

The true uncertainty bound equals the Cramér-Rao Bound computed directly from the converged weighted least squares solution, and comparing distributions at equal spread isolates shape effects from variance effects.

What would settle it

Generate synthetic measurements from each tested distribution, run WLS state estimation, and check whether the empirical coverage of the reported confidence intervals matches the CRB ratios predicted by the diagnostic.

Figures

Figures reproduced from arXiv: 2605.13390 by Bet\"ul Mamudi, Jochen Cremer, Jochen Stiasny.

**Figure 2.** Figure 2: CRB Ratio ρk for voltage magnitude at all 14 non-slack buses (xaxis). 100 operating scenarios (dots). Results for all distribution variants (see Table I). ρk < 1 indicates that WLS overstates the uncertainty bounds. increasing severity as α grows. For the Biased Gaussian, assumed and true coverage coincide, consistent with ρk = 1, but both fall significantly below the nominal level at large bias magnitude… view at source ↗

**Figure 3.** Figure 3: RMSE of WLS-based DSSE under distributional assumptions [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

read the original abstract

Pseudo-measurements are the dominant source of uncertainty in distribution system state estimation (DSSE), yet their distributional assumptions are treated as fixed inputs by existing uncertainty quantification methods. This paper investigates whether the uncertainty bounds assumed by weighted least squares (WLS)-based DSSE are sensitive to these distributional assumptions, and whether this sensitivity is quantifiable using the Fisher Information Matrix (FIM). We propose a diagnostic framework that compares the true Cram\'er-Rao Bound (CRB) against the WLS-assumed CRB via a per-bus, per-scenario ratio, computed directly from the converged WLS solution. Pseudo-measurement distributions are varied across five types in 22 variants matched at equal spread to isolate shape effects from variance. Experiments on the CIGRE MV network across 100 operating scenarios yield three findings. First, heavy-tailed and skewed distributions show consistently that WLS systematically overstates its uncertainty bounds. Second, the degree of miscalibration varies across buses and operating scenarios, confirming that distributional sensitivity is not uniform. Third, the CRB ratio is structurally blind to mean-shift bias, exposing a fundamental limitation of variance-based uncertainty diagnostics. Together, these results confirm the hypothesis and show that the choice of pseudo-measurement distribution directly distorts the confidence limits under WLS-based assumptions, which must be explicitly accounted for in any uncertainty-aware DSSE method.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

WLS uncertainty bounds in DSSE shift with pseudo-measurement distribution shape, shown via per-bus CRB ratios, but evaluating both bounds at the WLS estimate rather than the true state mixes in estimation error.

read the letter

The main point is that WLS uncertainty bounds for distribution state estimation depend on the shape of the pseudo-measurement distributions in ways that current methods do not account for. The authors use a ratio of Cramér-Rao bounds to make this visible on a per-bus basis across scenarios. What the paper does is introduce a diagnostic that runs WLS on data drawn from varied distributions while holding spread constant. They test five types across 22 variants on the CIGRE MV network over 100 scenarios. This produces clear evidence that heavy-tailed and skewed cases lead WLS to overstate its own uncertainty, and that the degree of this varies by bus and scenario. The third finding, that the ratio misses mean-shift bias, follows directly from the variance focus of the bound. The experiments are a strength here. They use the converged solution without cherry-picking and match the variants carefully. This gives reproducible support for the claim that distributional assumptions distort the confidence limits. The softer part is the way the true CRB is obtained. Both the true and assumed bounds come from the WLS estimate rather than the known true state. When the pseudo-measurements are non-Gaussian, the estimator is biased, so the Fisher matrix sits at the wrong point. The ratio then mixes the shape effect with this offset. The abstract states the computation is from the converged solution, which confirms the approach. It does not break the result that sensitivity exists, but it does mean the diagnostic is not as isolated as it first appears. This paper is for people who work on or apply state estimation in distribution systems, particularly those interested in uncertainty. It would be useful in a reading group to walk through the CRB step and see if the concern holds in the full methods. The work shows clear engagement with the problem and the simulations are direct, so it deserves a serious referee even if the interpretation of the ratios needs more care.

Referee Report

2 major / 2 minor

Summary. The paper claims that pseudo-measurement distributional assumptions in WLS-based DSSE distort the assumed uncertainty bounds, and proposes a per-bus CRB ratio diagnostic (true CRB under generating distribution vs. WLS-assumed CRB) computed from the converged WLS solution. Experiments on the CIGRE MV network using 22 matched distribution variants across 100 scenarios show that heavy-tailed and skewed distributions cause systematic overstatement of bounds by WLS, with non-uniform variation across buses and scenarios, while the ratio remains blind to mean-shift bias.

Significance. If the central comparison holds, the work provides concrete empirical evidence that standard WLS uncertainty quantification in DSSE is sensitive to pseudo-measurement shape in ways not captured by variance-only diagnostics. The matched-variant design and scale of the simulation (100 scenarios, 22 variants) strengthen the case that distributional assumptions must be explicitly handled in uncertainty-aware DSSE methods, with potential impact on practical confidence-interval usage in distribution networks.

major comments (2)

[§3] §3 (method): Both the true CRB (under the generating distribution) and the WLS-assumed CRB are evaluated at the converged WLS solution rather than the known true state. Under non-Gaussian pseudo-measurements the WLS estimator is generally inconsistent, so the reported ratio may partly reflect the offset between WLS estimate and true state rather than isolating pure shape effects on the bound. Please justify this evaluation point and consider an additional ablation that computes the true CRB at the known true state.
[§4] §4 (experiments): The claim that the 22 variants are matched at equal spread to isolate shape from variance requires explicit verification metrics (e.g., how variance, kurtosis, or other moments were equalized across the five distribution types). Without tabulated confirmation that residual spread differences are negligible, the attribution of CRB-ratio differences solely to distributional shape remains open to confounding.

minor comments (2)

[Abstract] Abstract: briefly name the five distribution types when stating 'five types in 22 variants' to improve immediate readability.
[Figures and §4] Figure captions and §4: ensure all bus indices and scenario identifiers are defined before first use in plots and tables.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments, which help clarify key aspects of the diagnostic framework. We address each major comment below and will revise the manuscript accordingly to strengthen the presentation.

read point-by-point responses

Referee: [§3] §3 (method): Both the true CRB (under the generating distribution) and the WLS-assumed CRB are evaluated at the converged WLS solution rather than the known true state. Under non-Gaussian pseudo-measurements the WLS estimator is generally inconsistent, so the reported ratio may partly reflect the offset between WLS estimate and true state rather than isolating pure shape effects on the bound. Please justify this evaluation point and consider an additional ablation that computes the true CRB at the known true state.

Authors: We agree that the choice of evaluation point merits explicit justification. The diagnostic is intentionally computed from the converged WLS solution because, in operational DSSE, the true state is unavailable; the ratio is therefore meant to reflect what an analyst would observe in practice. Nevertheless, we acknowledge that estimator inconsistency under non-Gaussian pseudo-measurements could contribute to the observed ratios. In the revised manuscript we will add a dedicated ablation subsection that recomputes the true CRB at the known true state for representative scenarios and compares the resulting ratios to those obtained at the WLS estimate. This will quantify any contribution from bias and isolate the pure distributional-shape effects more clearly. revision: yes
Referee: [§4] §4 (experiments): The claim that the 22 variants are matched at equal spread to isolate shape from variance requires explicit verification metrics (e.g., how variance, kurtosis, or other moments were equalized across the five distribution types). Without tabulated confirmation that residual spread differences are negligible, the attribution of CRB-ratio differences solely to distributional shape remains open to confounding.

Authors: We concur that tabulated moment verification is required to substantiate the matching procedure. The original text states that the 22 variants were constructed to have equal spread, but does not include explicit numerical confirmation. In the revision we will insert a new table (or appendix table) reporting the first four moments (mean, variance, skewness, kurtosis) for each of the five distribution families across all 22 variants. This will allow readers to verify that residual spread differences are negligible and that the observed CRB-ratio variations can be attributed primarily to shape rather than to unintended variance mismatches. revision: yes

Circularity Check

0 steps flagged

No significant circularity; diagnostic ratio derived from independent distributions and standard FIM

full rationale

The paper defines a per-bus CRB ratio by comparing the true Cramér-Rao Bound (under explicitly varied generating distributions) against the WLS-assumed bound evaluated at the converged WLS solution. This construction does not reduce to a fitted parameter, self-definition, or self-citation chain; the 22 distribution variants are chosen independently and matched only on spread, while the FIM evaluations follow standard formulas without importing uniqueness theorems or ansatzes from prior author work. The central claim—that distributional shape distorts WLS confidence limits—remains falsifiable against external benchmarks and does not loop back to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The framework rests on standard estimation theory and the experimental design choice to match spreads across distribution shapes.

axioms (2)

standard math The Fisher Information Matrix from the converged WLS solution yields the true Cramér-Rao Bound for the given measurement model.
Invoked in the diagnostic framework description.
domain assumption Varying pseudo-measurement distributions across five types in 22 variants while matching spread isolates shape effects from variance.
Used to design the 22 variants in the experiments.

pith-pipeline@v0.9.0 · 5543 in / 1245 out tokens · 52264 ms · 2026-05-14T18:29:12.282878+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

17 extracted references · 8 canonical work pages

[1]

Power System Static-State Estimation, Part I: Exact Model,

F. Schweppe and J. Wildes, “Power System Static-State Estimation, Part I: Exact Model,”IEEE Transactions on Power Apparatus and Systems, vol. PAS-89, no. 1, pp. 120–125, Jan. 1970. [Online]. Available: http://ieeexplore.ieee.org/document/4074022/

work page arXiv 1970
[2]

A review on distribution system state estimation,

A. Primadianto and C.-N. Lu, “A review on distribution system state estimation,”IEEE Transactions on Power Systems, vol. 32, no. 5, pp. 3875–3883, 2017

2017
[3]

A Comprehensive Review on Smart Grids: Challenges and Opportunities,

J. J. Moreno Escobar, O. Morales Matamoros, R. Tejeida Padilla, I. Lina Reyes, and H. Quintana Espinosa, “A Comprehensive Review on Smart Grids: Challenges and Opportunities,”Sensors, vol. 21, no. 21, p. 6978, Oct. 2021. [Online]. Available: https://www.mdpi.com/1424- 8220/21/21/6978

2021
[4]

Recent advancement in smart grid technology: Future prospects in the electrical power network,

O. Majeed Butt, M. Zulqarnain, and T. Majeed Butt, “Recent advancement in smart grid technology: Future prospects in the electrical power network,”Ain Shams Engineering Journal, vol. 12, no. 1, pp. 687–695, Mar. 2021. [Online]. Available: https://linkinghub.elsevier.com/retrieve/pii/S2090447920301064

2021
[5]

Electrical distribution system state estimation: measurement issues and challenges,

D. Della Giustina, M. Pau, P. A. Pegoraro, F. Ponci, and S. Sulis, “Electrical distribution system state estimation: measurement issues and challenges,”IEEE Instrumentation & Measurement Magazine, vol. 17, no. 6, pp. 36–42, Dec. 2014. [Online]. Available: https://ieeexplore.ieee.org/document/6968929/

work page arXiv 2014
[6]

Bayesian approach for distribution system state estimation,

M. Pau, P. A. Pegoraro, F. Ponci, and S. Sulis, “Bayesian approach for distribution system state estimation,” inPower Distribution System State Estimation, ser. IET Energy Engineering, E. M. Lourenc ¸o and J. B. A. London Jr., Eds. London: The Institution of Engineering and Technology, 2022, no. 183, pp. 209–237

2022
[7]

Pseudo-Measurement Models in Distribution Networks: A Review,

S. Afrasiabi, S. Allahmoradi, and X. Liang, “Pseudo-Measurement Models in Distribution Networks: A Review,”IET Smart Energy Systems, vol. 1, no. 1, pp. 56–72, Jun. 2025. [Online]. Available: https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/ses2.70001

work page doi:10.1049/ses2.70001 2025
[8]

Review of dataset and algorithms for distribution network pseudo measurement,

Y . Ju, X. Jia, and B. Wang, “Review of dataset and algorithms for distribution network pseudo measurement,”Energy Internet, vol. 2, no. 1, pp. 1–12, Apr. 2025. [Online]. Available: https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/ein2.12020

work page doi:10.1049/ein2.12020 2025
[9]

The impact of pseudo-measurements on state estimator accuracy,

K. A. Clements, “The impact of pseudo-measurements on state estimator accuracy,” in2011 IEEE Power and Energy Society General Meeting. San Diego, CA: IEEE, Jul. 2011, pp. 1–4. [Online]. Available: https://ieeexplore.ieee.org/document/6039370/

work page arXiv 2011
[10]

Abur and A

A. Abur and A. G. Exposito,Power system state estimation: theory and implementation. CRC press, 2004

2004
[11]

Bayesian Approach for Distribution System State Estimation With Non-Gaussian Uncertainty Models,

P. A. Pegoraro, A. Angioni, M. Pau, A. Monti, C. Muscas, F. Ponci, and S. Sulis, “Bayesian Approach for Distribution System State Estimation With Non-Gaussian Uncertainty Models,”IEEE Transactions on Instrumentation and Measurement, vol. 66, no. 11, pp. 2957–2966, Nov
[12]

Available: http://ieeexplore.ieee.org/document/8008784/

[Online]. Available: http://ieeexplore.ieee.org/document/8008784/

work page arXiv
[13]

Impact of the non-Gaussian measurement noise on the performance of state-of-the-art state estimators for distribution systems,

S. Cubonovic, D. Cetenovic, and A. Rankovic, “Impact of the non-Gaussian measurement noise on the performance of state-of-the-art state estimators for distribution systems,”Serbian Journal of Electrical Engineering, vol. 21, no. 1, pp. 113–133, 2024. [Online]. Available: https://doiserbia.nb.rs/Article.aspx?ID=1451-48692401113C

2024
[14]

Exact Modeling of Non-Gaussian Measurement Uncertainty in Distribution System State Estimation,

M. Vanin, T. Van Acker, R. D’hulst, and D. Van Hertem, “Exact Modeling of Non-Gaussian Measurement Uncertainty in Distribution System State Estimation,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–11, 2023. [Online]. Available: https://ieeexplore.ieee.org/document/10155202/

work page arXiv 2023
[15]

Impact of Current and Power Measurements on Distribution System State Estimation Uncertainty,

M. Pau, P. A. Pegoraro, A. Monti, C. Muscas, F. Ponci, and S. Sulis, “Impact of Current and Power Measurements on Distribution System State Estimation Uncertainty,”IEEE Transactions on Instrumentation and Measurement, vol. 68, no. 10, pp. 3992–4002, Oct. 2019. [Online]. Available: https://ieeexplore.ieee.org/document/8574940/

work page arXiv 2019
[16]

S. M. Kay,Fundamentals of Statistical Signal Processing, Volume˜I: Estimation Theory. Englewood Cliffs, NJ: PTR Prentice-Hall, 1993

1993
[17]

C. M. Bishop,Pattern recognition and machine learning, ser. Informa- tion science and statistics. New York: Springer, 2006

2006