skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


This content will become publicly available on April 10, 2026

Title: Initializing PV-PQ Switching in Power Flow Problems Using Neural Networks
To model reactive power limited generators within power flow problems, PV-PQ switching fixes generator voltages when reactive power outputs are within limits but allows the voltages to vary with a constant reactive power injection when limits are reached. Power flow algorithms often use heuristics that iteratively modify the generators’ PV versus PQ representation as the algorithm executes. The convergence behavior and speed of power flow algorithms with these heuristics significantly depend on their initialization. To improve computational performance, we propose an approach for using neural networks to initialize PV-PQ switching heuristics. After offline training where the neural networks learn the power flow solution’s PV vs. PQ generator statuses across varying load demands, the neural networks are deployed to initialize power flow algorithms in online applications. Numerical results demonstrate the effectiveness of this approach with speedup factors of 1.55× to 4.32× over the nominal generator PV-PQ status initialization.  more » « less
Award ID(s):
2145564
PAR ID:
10649625
Author(s) / Creator(s):
 ;  
Publisher / Repository:
IEEE
Date Published:
Page Range / eLocation ID:
1 to 6
Subject(s) / Keyword(s):
Neural networks, Power flow, PV-PQ switching, Initialization
Format(s):
Medium: X
Location:
Urbana, IL
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; (, MSE-Optimal Neural Network Initialization via Layer Fusion)
    null (Ed.)
    Deep neural networks achieve state-of-the-art performance for a range of classification and inference tasks. However, the use of stochastic gradient descent combined with the nonconvexity of the underlying optimization problems renders parameter learning susceptible to initialization. To address this issue, a variety of methods that rely on random parameter initialization or knowledge distillation have been proposed in the past. In this paper, we propose FuseInit, a novel method to initialize shallower networks by fusing neighboring layers of deeper networks that are trained with random initialization. We develop theoretical results and efficient algorithms for mean-square error (MSE)- optimal fusion of neighboring dense-dense, convolutional-dense, and convolutional-convolutional layers. We show experiments for a range of classification and regression datasets, which suggest that deeper neural networks are less sensitive to initialization and shallower networks can perform better (sometimes as well as their deeper counterparts) if initialized with FuseInit. 
    more » « less
  2. ; ; ; (, ArXivorg)
    Deep neural networks achieve state-of-the-art performance for a range of classification and inference tasks. However, the use of stochastic gradient descent combined with the nonconvexity of the underlying optimization problems renders parameter learning susceptible to initialization. To address this issue, a variety of methods that rely on random parameter initialization or knowledge distillation have been proposed in the past. In this paper, we propose FuseInit, a novel method to initialize shallower networks by fusing neighboring layers of deeper networks that are trained with random initialization. We develop theoretical results and efficient algorithms for mean-square error (MSE)-optimal fusion of neighboring dense-dense, convolutional-dense, and convolutional-convolutional layers. We show experiments for a range of classification and regression datasets, which suggest that deeper neural networks are less sensitive to initialization and shallower networks can perform better (sometimes as well as their deeper counterparts) if initialized with FuseInit. 
    more » « less
  3. ; ; (, IEEE Power & Energy Society General Meeting)
    Increasing emphasis on reliability and resiliency call for advanced distribution system restoration (DSR). The integration of grid sensors, remote controls, and distributed generators (DG) brings about exciting opportunities in DSR. In this context, this work considers the task of single-step restoration of a single-phase power distribution system. Different from existing works, the devised restoration scheme achieves optimal formation of islands without heuristically pre-identifying reference buses. It further facilitates multiple DGs running within the same island, and establishes a coordination hierarchy in terms of their PV/PQ operation modes. Generators without black-start capability are guaranteed to remain connected to a black-start DG or a substation. The proposed scheme models remotely-controlled voltage regulators exactly, and integrates them in the restoration process. Numerical tests on a modified IEEE 37-bus feeder demonstrate that the proposed mixed-integer linear program (MILP) takes less than four seconds to handle random outages of 1-5 lines. The scalability of this novel MILP formulation can be attributed to the unique use of cycles and paths on the grid infrastructure graph; the McCormick linearization technique; and an approximate power flow model. 
    more » « less
  4. (, 2020 IEEE Kansas Power and Energy Conference (KPEC))
    null (Ed.)
    With more and more renewable energy resources integrated into the power grid, the system is losing inertia because power electronics-based generators do not provide natural inertia. The low inertia will cause the microgrid to be more sensitive to disturbance and thus a small load change may result in a severe deviation in frequency. Based on the basic VSG algorithm, which is to mimic the characteristic of the traditional synchronous generator, the frequency can be controlled to a stable value faster and more smoothly when there is a fluctuation in the PV power generation and/or load change. However, characteristic of the VSG depends on the system structure in consideration of multiple generations, such as Synchronous Generator (SG), PV and Battery Energy Storage System (BESS), which greatly increases the complexity of applying VSG in practical power system. Furthermore, with BESS-VSG, Maximum Power Point (MPP) operation of PV is guaranteed. In addition, an adaptive VSG method is developed for a microgrid system, and the corresponding simulation in Matlab/Simulink shows the effectiveness of the adaptive VSG method. 
    more » « less
  5. ; (, IEEE Access)
    Power systems with utility-scale solar photovoltaic (PV) can significantly influence the operating points (OPs) of synchronous generators, particularly during periods of high solar PV generation. A sudden drop in solar PV output due to cloud cover or other transient conditions will alter the generation of synchronous generators shifting their OPs. These shifted OPs can become a challenge for stability as the system may operate closer to its stability limits. If a disturbance occurs while the system is operating at the shifted OP, with reduced stability margins, it will be more vulnerable to increased oscillations, loss of synchronism of its generator(s) and system instability. This study introduces a scalable delta-automatic generation control (delta-AGC) logic method designed to address stability challenges arising from shifts in the OPs of synchronous generators during abrupt drops in PV generation. By temporarily adjusting the OPs of synchronous generators through modification of their participation factors (PFs) in the AGC logic dispatch, the proposed method enhances power system stability. The proposed delta-AGC logic method focuses on the optimal determination of delta-PFs in power systems with large number of generators, using the concept of coherency and employing a hierarchical optimization strategy that includes both inter-coherent and intra-coherent group optimization. Additionally, a new electromechanical oscillation index (EMOI), integrating both time response analysis (TRA) and frequency response analysis (FRA), is utilized as an online situational awareness tool (SAT) for optimizing the system’s stability under various conditions. This online SAT has been implemented in a decentralized manner at the area level, limiting wide-area communication overheads and any cybersecurity concerns. The delta-AGC logic method is illustrated on a modified IEEE 68 bus system, incorporating large utility-scale solar PV plants, and is validated through real-time simulation. Various cases, including high-loading conditions with and without power system stabilizers, conventional AGC logic, and delta-AGC logic, are carried out to evaluate the effectiveness of the proposed delta-AGC logic method. The results illustrate the performance and benefits of the delta-AGC logic method, highlighting its potential to significantly enhance power system stability. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email