PowerModelsDistributionStateEstimation.jl

PowerModelsDistributionStateEstimation.jl is an extension package of PowerModelsDistribution.jl for Static Power Distribution Network State Estimation. The package is a flexible design tool, enabling benchmarks between different state estimation models. Different state estimation models can be built by using different power flow formulations, state estimation criteria, (in)equality constraints, etc. The package has documentation, which we try to keep up to date.

A state estimator determines the most-likely state of power distribution networks given a set of uncertainties, e.g., measurement errors, pseudo-measurements, etc. These uncertainties may pertain to any quantity of any network component, e.g., voltage magnitude (vm) of a bus, power demand (pd) of a load, etc.

Estimation Criteria:

• (Weighted) Least Squares ((W)LS)
• (Weighted) Least Absolute Values ((W)LAV)
• Maximum Likelihood Estimation (MLE)

Measurement Uncertainties:

• a deterministic value Float64, or
• a continuous univariate distribution ContinuousUnivariateDistribution
  • normal distribution, included through (W)LS or (W)LAV
  • non-normal distributions, included through MLE

• Exact Formulations

  • (reduced) ACP
  • (reduced) ACR
  • (reduced) IVR

• Linear Approximation

  • LinDist3Flow

• Other formulations might be added in the future, little preliminary work exists for a SDP relaxation, but it is currently suspended for lack of research interest. If you would like to contribute (to add formulations or any other addition/improvement), you are welcome to get in touch.

To use the package, two type of data inputs are required:

• Network data: support exists for OpenDSS “.dss”, matpower ".m" and some specific JSON files
• Measurement data: CSV “.csv” files

See the relative section of the docs for more info.

As of version 0.4.0, PMDSE supports the following bad data detection and identification functionalities:

• Chi-square analysis
• Largest normalized residuals
• Analysis of residuals from robust LAV estimation

Examples on how to use PowerModelsDistributionStateEstimation can be found in Pluto Notebooks inside the /examples directory.

This code has been developed at KU Leuven (University of Leuven). The primary developers are Marta Vanin (@MartaVanin) and Tom Van Acker (@timmyfaraday) with support from the following contributors:

• Frederik Geth (@frederikgeth), CSIRO, General PowerModelsDistribution.jl Advice.
• Sander Claeys (@sanderclaeys), KU Leuven, General PowerModelsDistribution.jl Advice, ENWL data parser.

If you find PowerModelsDistributionStateEstimation.jl useful for your work, we kindly invite you to cite our paper:

@ARTICLE{Vanin2022,
  author={Vanin, Marta and Van Acker, Tom and D'hulst, Reinhilde and Van Hertem, Dirk},
  journal={IEEE Transactions on Power Systems}, 
  title={A Framework for Constrained Static State Estimation in Unbalanced Distribution Networks}, 
  year={2022},
  volume={37},
  number={3},
  pages={2075-2085},
  doi={10.1109/TPWRS.2021.3116291}}

If you are particularly interested in the non-Gaussian state estimation capabilities, you can refer to this other paper:

@ARTICLE{Vanin2023,
  author={Vanin, Marta and Van Acker, Tom and D’hulst, Reinhilde and Van Hertem, Dirk},
  journal={IEEE Transactions on Instrumentation and Measurement}, 
  title={Exact Modeling of Non-Gaussian Measurement Uncertainty in Distribution System State Estimation}, 
  year={2023},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TIM.2023.3287253}}

It is possible to use the fast prototyping features of PMDSE to augment the state variables, e.g., with network parameters. This is discussed in the following two references where, jointly to the conventional state, we derive:

1. impedance matrices or cable lengths: https://www.sciencedirect.com/science/article/pii/S0142061523002120
2. customer phase connectivity: https://arxiv.org/abs/2206.08436

This code is provided under a BSD license.

• The intention of this package is not to provide the fastest SE algorithms, but a framework to facilitate the distribution SE design process. If faster solution times are crucial, a customized algorithm can be developed afterwards, once the optimal design is chosen.
• Accurate load and transformer models are available on PowerModelsDistribution and can be easily included in this package for state estimation purposes, e.g., to perform a multi-level MV/LV state estimation. Extending the package to include these models in a more automatic and intuitive manner is scheduled for future releases.