Source-linked AI summary
A Survey on State Estimation Techniques and Challenges in Smart Distribution Systems
Kaveh Dehghanpour, Zhaoyu Wang, Jianhui Wang, Yuxuan Yuan, Fankun Bu
TL;DR
DSSE remains challenging because distribution networks have limited observability, uncertain topology, renewable-driven variability, and cyber-security concerns. This paper reviews conventional, data-driven, and probabilistic solutions across these issues, concluding that recent work increasingly modifies conventional DSSE with data-driven and machine-learning methods. It also identifies open directions involving demand response, storage, microgrids, and decision-making under limited observability.
Problem
DSSE must estimate distribution-system states despite limited observability, topology uncertainty, renewable variability, and cyber-security concerns.
Method
The paper conducts a literature review covering DSSE formulation, pseudo-measurements, topology, meter placement, renewable integration, cyber-security, and data-driven and probabilistic techniques.
Results
The survey finds that recent works increasingly use data-driven and machine-learning modifications of conventional DSSE, alongside substantial interest in probabilistic modeling.
Takeaways & Limitations
Future DSSE research includes demand-response effects, retail-market signals, and power management under limited observability with storage and networked microgrids.
Abstract
from arXiv · showhide
This paper presents a review of the literature on State Estimation (SE) in power systems. While covering some works related to SE in transmission systems, the main focus of this paper is Distribution System State Estimation (DSSE). The paper discusses a few critical topics of DSSE, including mathematical problem formulation, application of pseudo-measurements, metering instrument placement, network topology issues, impacts of renewable penetration, and cyber-security. Both conventional and modern data-driven and probabilistic techniques have been reviewed. This paper can provide researchers and utility engineers with insights into the technical achievements, barriers, and future research directions of DSSE.
I. INTRODUCTION
DSSE infers distribution-system state variables from limited measurements and is becoming important for smart-grid monitoring and power management. The review covers formulation, observability, topology, metering, renewables, and cyber-security challenges.
- DSSE maps limited measurements at selected network locations to estimates of system state variables.
- Distribution-level state estimation remains an active research area despite being well developed in transmission systems.
- Distribution systems face limited metering, low x/r values, unbalanced operation, communication constraints, and incomplete topology data.
- Renewable penetration increases uncertainty in distribution-system operation and DSSE.
- The review emphasizes DSSE problem formulation, pseudo-measurements, topology, meter placement, renewable integration, cyber-security, and data-driven approaches.
II. FUNDAMENTALS OF SE
The conventional DSSE formulation uses weighted least squares solved iteratively, while alternative methods address bad data, nonconvexity, and sensitivity to initialization. Virtual measurements can be enforced as equality constraints.
- Conventional Approach: WLS models measurements as z = h(x) + e and estimates the state by minimizing weighted measurement residuals.
- Conventional Approach: The weight matrix represents confidence in measurements; standard WLS assumes independent, zero-mean Gaussian measurement errors.
- Robustification: Residual-based weight updates reduce unreliable measurements’ influence when data quality falls beyond specified thresholds.
- Conventional Approach: Gauss-Newton iteratively solves the WLS stationarity equation, while backtracking, trust-region, quasi-Newton, and SDP methods address convergence or initialization issues.
- Robustification: Virtual measurements encode perfect operator information as equality constraints, enforced through Lagrange multipliers in modified WLS.
H TWH
The paper surveys formulations that improve robustness, relax Gaussian-error assumptions, or reduce computational burden in DSSE. These approaches include robust estimators, probabilistic methods, matrix completion, and quasi-symmetric impedance constraints.
- LMS, LTS, LAV, and GM estimators are reviewed as alternatives to WLS for handling outliers and bad data.
- Field tests indicate that WLS’s Gaussian uncertainty assumption can be inaccurate, motivating non-Gaussian and probabilistic formulations.
- MSE-based estimation avoids Gaussian assumptions, while matrix-completion formulations address limitations involving accurate second-order statistics.
- A quasi-symmetric impedance constraint is introduced to reduce optimization size and accelerate WLS convergence for large-scale feeders.
III. DSSE PROBLEM FORMULATION
DSSE formulation differs substantially from transmission-system estimation because its measurement function reflects distribution power-flow characteristics. Choices of states, AC/DC modeling, and phase representation shape the formulation.
- Distribution-system measurement functions reflect power-flow equations, creating major formulation differences from transmission-system state estimation.
- Voltage-Based DSSE: Voltage magnitude and phase angle are traditional DSSE state variables inherited from transmission-system estimation.
- Branch-Current-Based SE: Branch-current-based estimation uses branch currents as state variables, a formulation described as natural for distribution systems.
IV. DISTRIBUTION SYSTEM OBSERVABILITY
Distribution-system observability is constrained by sparse, imperfect measurements and communication failures. The reviewed literature addresses these constraints through uncertainty assessment and pseudo-measurement generation, including statistical and learning-based approaches.
- Observability depends on the number and location of meters and on the availability and quality of real-time measurement data.
- Distribution systems have poor observability because they are underdetermined and can become unobservable during communication failures or delays.
- Bad data arise from meter malfunction and communication noise, while missing data are treated as a special case of bad data.
- Insufficient measurement redundancy and multiple bad-data sources complicate conventional normalized-residual detection methods.
- Pseudo-measurements augment or correct input measurements using artificially generated data based on distribution-system history.
- Reviewed pseudo-measurement methods include probabilistic, statistical, and machine-learning techniques for generation and uncertainty assessment.
- Many studies rely on standard load profiles rather than real AMI histories, while large practical datasets can make some learning methods computationally expensive.
V. NETWORK TOPOLOGY AND CONFIGURATION
Network-topology research addresses both updating known configurations after local events and discovering largely unknown distribution-network structures. The reviewed approaches use measurements and graph-based or statistical learning methods to improve topology knowledge.
- Topology identification comprises system-configuration identification and topology learning as two related subproblems.
- System configuration identification: System-configuration identification updates known network topology after faults, disconnections, or switching events using system-wide measurements.
- Topology learning: Topology learning discovers network structure from nodal and branch measurements when operators have limited or no prior topology knowledge.
- System configuration identification: A data-driven configuration-detection topic is identified as part of the reviewed topology literature.
- Topology learning: A bottom-to-top structure-learning method can operate under no prior topology knowledge and missing measurement data.
A. Metering Instrument Placement
Meter placement is studied as a constrained optimization problem because distribution systems are large and financial resources may be limited. Proposed objectives span observability, cost, bad-data detection, and DSSE accuracy.
- Metering-instrument placement is significant because distribution systems are large and available financial resources may be limited.
- Placement objectives include improving observability and DSSE accuracy, minimizing installation and maintenance costs, and strengthening bad-data detection.
- Reviewed placement algorithms include Genetic Algorithm, Mixed Integer Linear Programming, Mixed Integer Semi-Definite Programming, and Multi-Objective Evolutionary methods.
B. PMU Applications and Impacts on DSSE
PMUs provide synchronized, high-frequency measurements that support detailed distribution-system monitoring and several identification tasks. Their restricted deployment means many DSSE methods use a small-phase-angle-difference assumption, which introduces bounded inaccuracies while avoiding PMU dependence.
- PMUs provide synchronized voltage, power, and current measurements for state tracking and distribution-system control and management.
- PMU sampling can reach 30 kHz, compared with 0.277 mHz–16.7 mHz for smart meters, providing finer temporal observability.
- Distribution PMUs support high-resolution profiling, oscillation detection, topology identification, and event detection.
- Without PMU phase angles, many methods assume nearly equal nodal voltage phase angles; this enables monitoring but introduces bounded inaccuracies.
- Adding voltage-phase or flow measurements can improve estimation and topology-identification routines.
VII. PENETRATION OF RENEWABLE RESOURCES
Renewable penetration makes DSSE harder because uncertain generation affects operation and voltage profiles. The literature addresses these challenges with advanced monitoring, probabilistic modeling, forecasting, and alternative solution approaches.
- Challenges: Uncertain renewable output is the main challenge for DSSE under high renewable penetration.Deep renewable penetration also affects distribution-system voltage profiles.
- Challenges: Non-Gaussian renewable power distributions can adversely affect conventional WLS-based DSSE methods.Fast state changes can trap WLS-based DSSE in local minima, with errors as high as 105 times the underlying global solution.
- Solution approaches: Conventional Gauss–Newton DSSE depends strongly on initial conditions, making initialization difficult with deep renewable penetration.Several studies therefore adopt alternative approaches for SE and DSSE with renewable-based distributed generation.
- Solution approaches: Probabilistic methods model renewable uncertainty, including forecasting-aided SE that uses temporal and spatial correlations for pseudo-measurements.Reviewed approaches include linear autoregressive forecasting and Gaussian mixture model techniques for load and renewable uncertainty.
VIII. CYBER-SECURITY
The survey reviews cyber-security threats to SE and places them alongside broader DSSE research directions. It identifies false data injection, topology attacks, and eavesdropping as modeled attack categories, while highlighting topology learning after extreme weather as an open direction.
- Threats: SE cyber-attacks studied in the literature include false data injection, topology attacks, and eavesdropping.These attacks target metered data, topology-model data, or communications infrastructure, respectively.
- Threats: False data injection alters selected metering data, whereas topology attacks maliciously modify the system’s topology model.The attacker’s knowledge of system parameters and states can vary in false data injection scenarios.
- Survey synthesis: The survey covers DSSE cyber-security among critical topics and reports growing use of data-driven and machine-learning modifications to conventional DSSE.The stated purposes include improving accuracy, robustness, and system observability as smart meters and micro-PMUs increase.
- Future research directions: Topology discovery after extreme weather remains an open research direction when communication and device failures reduce meter availability and observability.The paper also identifies data-driven methods during different stages of extreme weather as a future research area.