Adaptive Hosting Capacity Forecasting in Distribution Networks with Distributed Energy Resources

Md Tariqul Islam, M. Jahangir Hossain (), Md. Ahasan Habib and Muhammad Ahsan Zamee ()
Additional contact information
Md Tariqul Islam: School of Electrical and Data Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia
M. Jahangir Hossain: School of Electrical and Data Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia
Md. Ahasan Habib: School of Electrical and Data Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia
Muhammad Ahsan Zamee: School of Electrical and Data Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia

Energies, 2025, vol. 18, issue 2, 1-25

Abstract: The sustainable integration of distributed energy resources (DER) into distribution networks requires accurate forecasting of hosting capacity. The network and DER variables alone do not capture the full range of external influences on DER integration. Traditional models often overlook the dynamic impacts of these exogenous factors, leading to suboptimal predictions. This study introduces a Sensitivity-Enhanced Recurrent Neural Network (SERNN) model, featuring a sensitivity gate within the neural network’s memory cell architecture to enhance responsiveness to time-varying variables. The sensitivity gate dynamically adjusts the model’s response based on external conditions, allowing for improved capture of input variability and temporal characteristics of the distribution network and DER. Additionally, a feedback mechanism within the model provides inputs from previous cell states into the forget gate, allowing for refined control over input selection and enhancing forecasting precision. Through case studies, the model demonstrates superior accuracy in hosting capacity predictions compared to baseline models like LSTM, ConvLSTM, Bidirectional LSTM, Stacked LSTM, and GRU. Study shows that the SERNN achieves a mean absolute error (MAE) of 0.2030, a root mean square error (RMSE) of 0.3884 and an R-squared value of 0.9854, outperforming the best baseline model by 48 per cent in MAE and 71 per cent in RMSE. Additionally, Feature engineering enhances the model’s performance, improving the R-squared value from 0.9145 to 0.9854. The sensitivity gate also impacts the model’s performance, lowering MAE to 0.2030 compared to 0.2283 without the sensitivity gate, and increasing the R-squared value from 0.9152 to 0.9854. Incorporating exogenous factors such as the time of day as a sensitivity gate input, further improves responsiveness, making the model more adaptable to real-world conditions. This advanced SERNN model offers a reliable framework for distribution network operators, supporting intelligent planning and proactive DER management. Ultimately, it provides a significant step forward in hosting capacity analysis, enabling more efficient and sustainable DER integration within next-generation distribution networks.

Keywords: artificial intelligence; hosting capacity; distributed energy resources; PV; EVs; battery energy storage systems; sensitivity gate; contextual input; sensitivity-enhanced recurrent neural network (SERNN) (search for similar items in EconPapers)
JEL-codes: Q Q0 Q4 Q40 Q41 Q42 Q43 Q47 Q48 Q49 (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/1996-1073/18/2/263/pdf (application/pdf)
https://www.mdpi.com/1996-1073/18/2/263/ (text/html)

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:gam:jeners:v:18:y:2025:i:2:p:263-:d:1563239

Access Statistics for this article

Energies is currently edited by Ms. Agatha Cao

More articles in Energies from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().