[Paper Review] Model-less Robust Voltage Control in Active Distribution Networks using Sensitivity Coefficients Estimated from Measurements

Rahul Gupta, Fabrizio Sossan+1 more|arXiv (Cornell University)|Jan 11, 2022

Optimal Power Flow Distribution22 references24 citations

TL;DR

This paper proposes a model-less robust voltage control framework for active distribution networks that uses recursively estimated voltage sensitivity coefficients and their uncertainties from real-time measurements to ensure reliable voltage regulation. By incorporating uncertainty from instrument transformer noise via recursive least squares with directional forgetting, the method outperforms non-robust approaches, maintaining voltage within limits even under high-accuracy measurement noise (IT class 1.0), with performance matching model-based controls.

ABSTRACT

Measurement-rich power distribution networks may enable distribution system operators (DSOs) to adopt model-less and measurement-based monitoring and control of distributed energy resources (DERs) for mitigating grid issues such as over/under voltages and lines congestions. However, measurement-based monitoring and control applications may lead to inaccurate control decisions due to measurement errors. In particular, estimation models relying on regression-based schemes result in significant errors in the estimates (e.g., nodal voltages) especially for measurement devices with high Instrument Transformer (IT) classes. The consequences are detrimental to control performance since this may lead to infeasible decisions. This work proposes a model-less robust voltage control accounting for the uncertainties of measurement-based estimated voltage sensitivity coefficients. The coefficients and their uncertainties are obtained using a recursive least squares (RLS)-based online estimation, updated whenever new measurements are available. This formulation is applied to control distributed controllable photovoltaic (PV) generation in a distribution network to restrict the voltage within prescribed limits. The proposed scheme is validated by simulating a CIGRE low-voltage system interfacing multiple controllable PV plants.

Motivation & Objective

• . The paper investigates the impact of uncertainty in measurement-based voltage sensitivity coefficient estimates on control performance in distribution networks.
• It aims to develop a robust voltage control scheme that accounts for estimation uncertainty in model-less control frameworks.
• The objective is to improve reliability and safety of voltage control in active distribution networks where network models are unavailable or outdated.
• The study evaluates and compares different estimation techniques for sensitivity coefficients and their uncertainties to identify the most effective approach.

Proposed method

• . Voltage sensitivity coefficients and their uncertainties are estimated online using recursive least squares (RLS) with multiple forgetting schemes (e.g., RLS-SF, RLS-DF).
• Initial estimates are obtained via offline least squares (LS) using historical measurements.
• Uncertainties are derived from the inverse of the Fisher information matrix to quantify estimation reliability.
• A two-stage framework is employed: first, sensitivity coefficients and their uncertainties are estimated from real-time measurements; second, a robust optimization problem is solved using these estimates.
• The robust control formulation accounts for uncertainty in sensitivity coefficients to prevent infeasible or unsafe control actions.
• The method is validated on the CIGRE LV distribution network with multiple controllable PV plants.

Experimental results

Research questions

• RQ1 . How do uncertainties in measurement-based voltage sensitivity coefficient estimates affect the performance of model-less voltage control in distribution networks?
• RQ2 Which recursive estimation technique (e.g., RLS with different forgetting schemes) provides the most accurate and reliable sensitivity coefficient estimates under varying instrument transformer (IT) class noise?
• RQ3 Can a model-less robust voltage control scheme, accounting for estimation uncertainty, achieve performance comparable to model-based control?
• RQ4 How does the proposed robust control perform under high-accuracy measurement noise (e.g., IT class 1.0), where estimation errors are most detrimental?
• RQ5 What is the impact of ignoring estimation uncertainty on voltage control performance, particularly in terms of voltage violations?

Key findings

• . Non-robust voltage control, even when using the best-performing estimation method (RLS-DF), results in significant voltage violations, with maximum nodal voltage reaching 1.059 pu under IT class 1.0.
• The proposed robust control using RLS-DF achieves a maximum nodal voltage of 1.034 pu under IT class 1.0, closely matching the model-based control (1.031 pu).
• RLS-DF and RLS-SF estimation schemes consistently outperform other methods, achieving the lowest RMSE (0.05–0.07) and highest PICP (1.00) across all IT classes.
• Robust control using RLS-DF reduces voltage violations to negligible levels, even under high IT class noise, while non-robust control fails to respect voltage limits regardless of estimation quality.
• The model-less robust control with RLS-DF achieves performance comparable to model-based control, with only a 3% increase in curtailed PV energy (104 kWh vs. 86.5 kWh) while ensuring voltage constraints are met.
• The study confirms that ignoring estimation uncertainty leads to infeasible control decisions, even with high-quality measurements, highlighting the necessity of uncertainty-aware control in model-less frameworks.