Enhancing PV hosting capacity and mitigating congestion in distribution networks with deep learning based PV forecasting and battery management

Noman Shabbir, Lauri Kütt, Victor Astapov, Kamran Daniel, Muhammad Jawad, Oleksandr Husev, Argo Rosin and João Martins

Applied Energy, 2024, vol. 372, issue C, No S030626192401153X

Abstract: The extensive deployment of domestic photovoltaic (PV) systems may result in exceeding the limits of the network's PV hosting capacity (HC), which leads to energy delivery congestion and overvoltage problems in low voltage (LV) networks. These problems amplify at times when PV energy generation is near its peak and the domestic load is at its lowest. The convenient solution is to upgrade powerlines or utilize external mitigation devices, both are costly and reconstructing all the networks is a challenging task. Therefore, in this study, a machine learning-based control (MLC) solution is proposed to relieve the overloading of the lines and overvoltage problems caused by the high penetration of PV installations without costly line replacements or additional devices. This MLC technique employs a long short-term memory (LSTM) algorithm for a day ahead PV forecast incorporated to decide the charging/discharging of the residential battery energy storage systems (BESS). The study is conducted based on a real-life LV distribution system with 15 households in the network segment by considering the measured data of actual residential loads and installed PV energy generation for the entire year. Four different rated PV system installation scenarios are investigated through power flow analysis. Moreover, three different BESS utilization techniques are compared along with traditional peak power curtailment (PPC) and reactive power control (RPC) techniques. The MLC approach shows a 30% and 4% lower number of overvoltage hours compared to analytics-driven control (ADC) and trivial battery control (BTC) techniques, respectively for excessive PV installations (case 1). The MLC methods address the HC in a more dedicated manner, eliminating congestion problems in PV installation according to the peak load (Case 2 & 3) and net zero energy (Case 4). The techno-economic analysis indicates that the MLC technique with local grid-supporting capabilities increases the HC and reduces overloading and overvoltage problems in the LV network in all cases. The customers receive 8% to 22% more in savings as compared to TBC control technique and could also get further lower energy costs from the utilities for grid service.

Keywords: Renewable energy; Congestion control; Hosting capacity; Deep learning; LSTM (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S030626192401153X
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:372:y:2024:i:c:s030626192401153x

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2024.123770

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().