ktye-UF / PowerSystemTopology

This repo is the code for the 2024 IEEE PES GM paper. It proposes a novel topology embedding method for handling topology problem in power system.

PowerSystemTopology

Under Development

Matlab code for the paper: Power System Overloading Risk Assessment Considering Topology and Renewable Uncertainties, IEEE PES GM 2024.

This repo is currently under development as more detailed introduction and better-quality plots will be included.

Requriements

Matpower

UQLab

Matlab C++ Interface

Introduction

The code consists of two parts:

1. Data Generation: Code comments can be found in "dataset_generation/main.m". This part is for generating power flow simulation data.

2. Proposed PPGP method and topology embedding approach: Code comments can be found in "main_ppgp_topo.m". This part is for constructing PPGP model and embedding topology information into input.

Topology Embedding

• Topology embedding while preserving input structure:

• Topology embedding via Laplacian matrix and SVD:
  • Laplacian matrix (degree matrix minus adjacency matrix): $L = D - A$
  • SVD decomposition: $L = U \Sigma V^T$
  • New input: $\hat{X} = X \times V^T$

FolderTree

.
└── PPGP-topology/
    ├── dataset_generation/        # This folder is for generating dataset
    │   └── ...
    ├── plot_topo/
    │   └── ...
    ├── ppgp_iso/                  # This folder is for PPGP with isotropical kernel
    │   └── ...
    ├── utility/                   # This folder contains utility functions
    │   └── ...
    ├── main_ppgp_topo.m           # main function
    └── ReadMe.txt

References

Parallel partial Gaussian process by Mengyang Gu: RobustGaSP-in-Matlab or RobustGaSP-in-R

Note: An isotropic kernel is highly recommended for efficiency. Its implementation can be found in the R package but not in Matlab currently (will be included in this repo soon).