Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU

Given the inherent fluctuation of photovoltaic (PV) generation, accurately forecasting solar power output and grid feed-in is crucial for optimizing grid operations. Data-driven methods facilitate efficient supply and demand management in smart grids, but predicting solar power remains challenging d...

Full description

Saved in:
Bibliographic Details
Main Authors: Pascal Riedel, Kaouther Belkilani, Manfred Reichert, Gerd Heilscher, Reinhold von Schwerin
Format: Article
Language:English
Published: Elsevier 2024-12-01
Series:Energy and AI
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2666546824001186
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846117670055837696
author Pascal Riedel
Kaouther Belkilani
Manfred Reichert
Gerd Heilscher
Reinhold von Schwerin
author_facet Pascal Riedel
Kaouther Belkilani
Manfred Reichert
Gerd Heilscher
Reinhold von Schwerin
author_sort Pascal Riedel
collection DOAJ
description Given the inherent fluctuation of photovoltaic (PV) generation, accurately forecasting solar power output and grid feed-in is crucial for optimizing grid operations. Data-driven methods facilitate efficient supply and demand management in smart grids, but predicting solar power remains challenging due to weather dependence and data privacy restrictions. Traditional deep learning (DL) approaches require access to centralized training data, leading to security and privacy risks. To navigate these challenges, this study utilizes federated learning (FL) to forecast feed-in power for the low-voltage grid. We propose a bottom-up, privacy-preserving prediction method using differential privacy (DP) to enhance data privacy for energy analytics on the customer side. This study aims at proving the viability of an enhanced FL approach by employing three years of meter data from three residential PV systems installed in a southern city of Germany, incorporating irradiance weather data for accurate PV power generation predictions. For the experiments, the DL models long short-term memory (LSTM) and gated recurrent unit (GRU) are federated and integrated with DP. Consequently, federated LSTM and GRU models are compared with centralized and local baseline models using rolling 5-fold cross-validation to evaluate their respective performances. By leveraging advanced FL algorithms such as FedYogi and FedAdam, we propose a method that not only predicts sequential energy data with high accuracy, achieving an R2 of 97.68%, but also adheres to stringent privacy standards, offering a scalable solution for the challenges of smart grids analytics, thus clearly showing that the proposed approach is promising and worth being pursued further.
format Article
id doaj-art-da65713426ff4f95accfb9fefdd053c1
institution Kabale University
issn 2666-5468
language English
publishDate 2024-12-01
publisher Elsevier
record_format Article
series Energy and AI
spelling doaj-art-da65713426ff4f95accfb9fefdd053c12024-12-18T08:53:08ZengElsevierEnergy and AI2666-54682024-12-0118100452Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRUPascal Riedel0Kaouther Belkilani1Manfred Reichert2Gerd Heilscher3Reinhold von Schwerin4Institute of Databases and Information Systems, Ulm University, Ulm, 89081, Germany; Corresponding author.Smart Grids Research Group, Ulm University of Applied Sciences, Ulm, 89081, GermanyInstitute of Databases and Information Systems, Ulm University, Ulm, 89081, GermanySmart Grids Research Group, Ulm University of Applied Sciences, Ulm, 89081, GermanyInstitute of Databases and Information Systems, Ulm University, Ulm, 89081, GermanyGiven the inherent fluctuation of photovoltaic (PV) generation, accurately forecasting solar power output and grid feed-in is crucial for optimizing grid operations. Data-driven methods facilitate efficient supply and demand management in smart grids, but predicting solar power remains challenging due to weather dependence and data privacy restrictions. Traditional deep learning (DL) approaches require access to centralized training data, leading to security and privacy risks. To navigate these challenges, this study utilizes federated learning (FL) to forecast feed-in power for the low-voltage grid. We propose a bottom-up, privacy-preserving prediction method using differential privacy (DP) to enhance data privacy for energy analytics on the customer side. This study aims at proving the viability of an enhanced FL approach by employing three years of meter data from three residential PV systems installed in a southern city of Germany, incorporating irradiance weather data for accurate PV power generation predictions. For the experiments, the DL models long short-term memory (LSTM) and gated recurrent unit (GRU) are federated and integrated with DP. Consequently, federated LSTM and GRU models are compared with centralized and local baseline models using rolling 5-fold cross-validation to evaluate their respective performances. By leveraging advanced FL algorithms such as FedYogi and FedAdam, we propose a method that not only predicts sequential energy data with high accuracy, achieving an R2 of 97.68%, but also adheres to stringent privacy standards, offering a scalable solution for the challenges of smart grids analytics, thus clearly showing that the proposed approach is promising and worth being pursued further.http://www.sciencedirect.com/science/article/pii/S2666546824001186Federated learningDeep learningRecurrent neural networksData privacySolar power forecastingSmart grid
spellingShingle Pascal Riedel
Kaouther Belkilani
Manfred Reichert
Gerd Heilscher
Reinhold von Schwerin
Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
Energy and AI
Federated learning
Deep learning
Recurrent neural networks
Data privacy
Solar power forecasting
Smart grid
title Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
title_full Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
title_fullStr Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
title_full_unstemmed Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
title_short Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU
title_sort enhancing pv feed in power forecasting through federated learning with differential privacy using lstm and gru
topic Federated learning
Deep learning
Recurrent neural networks
Data privacy
Solar power forecasting
Smart grid
url http://www.sciencedirect.com/science/article/pii/S2666546824001186
work_keys_str_mv AT pascalriedel enhancingpvfeedinpowerforecastingthroughfederatedlearningwithdifferentialprivacyusinglstmandgru
AT kaoutherbelkilani enhancingpvfeedinpowerforecastingthroughfederatedlearningwithdifferentialprivacyusinglstmandgru
AT manfredreichert enhancingpvfeedinpowerforecastingthroughfederatedlearningwithdifferentialprivacyusinglstmandgru
AT gerdheilscher enhancingpvfeedinpowerforecastingthroughfederatedlearningwithdifferentialprivacyusinglstmandgru
AT reinholdvonschwerin enhancingpvfeedinpowerforecastingthroughfederatedlearningwithdifferentialprivacyusinglstmandgru