Federated Learning for Long-term Forecasting of Electricity Consumption towards a Carbon-neutral Future

Shen, Zhiheng; Wu, Qiaofeng; Qian, Jiajia; Gu, Chenlin; Sun, Fuchun; Jia, Tan

doi:10.1109/icsp54964.2022.9778813

Cited by 7 publications

(6 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Demand forecasting [8], [9], [19], [25], [26], [27], [30], [31], [32], [33], [34], [40], [51], [55], [63], [65], [66], [67], [69], [70], [73], [74], [77], [79], [81], [90], [91], [92] Achieving Generation forecasting [7], [35], [38],…”

Section: Forecastingmentioning

confidence: 99%

“…In [114], different machine learning algorithms for building load prediction are analyzed. However, these centralized machine techniques cannot overcome challenges such as data privacy and security [19], [31], [32], [51], [74], communication overhead [25], [31], [32], [34], computational ability [63], data insufficiency [66], [77]; hence, the federated learning based model has been introduced, which can also be used to improve the model's scalability [74], [92] and the model generation ability [30], and to mitigate the problems of data heterogeneity [81], [92].…”

Section: ) Demand Forecastingmentioning

confidence: 99%

“…Typically, the historical load dataset, along with some relevant information such as temperature and housing attributes, serves as the primary input data for shortterm forecasting [32], [33]. For long-term forecasting at the regional level, more influential factors need to be considered, such as urbanization rate, resident population, carbon emission, load profile, and weather conditions [30]. These additional considerations significantly increase the complexity of long-term predictions compared to short-term predictions at the individual level.…”

Section: ) Demand Forecastingmentioning

confidence: 99%

“…• Model generalization ability: model generalization ability refers to the ability of a machine learning model to deal with unfamiliar datasets. FL can improve model generalization by training on diverse and distributed datasets[30].• Data heterogeneity: data heterogeneity refers to the data with diverse or varied formats, contents, or characteristics within one dataset or multiple datasets. FL is well-suited for dealing with heterogeneity, as it can train models on data with varying characteristics and distributions across different locations or entities within the power system[20].FIGURE 1.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Advancing Power System Services With Privacy-Preserving Federated Learning Techniques: A Review

Zheng,

Sumper,

Aragüés-Peñalba

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Digitalization has enabled the potential for artificial intelligence techniques to lead the power system to a sustainable transition by extracting the data generated by widely deployed edge devices, including advanced sensing and metering. Due to the increasing concerns about data privacy, federated learning has attracted much attention and is emerging as an innovative application for machine learning solutions in the power and energy sector. This paper presents a holistic analysis of federated learning applications in the energy sector, ranging from applications in generation, microgrids, and distribution systems to the energy market and cyber security. The following federated learning-based services for energy sectors are analyzed: non-intrusive load monitoring, fault detection, energy theft detection, demand forecasting, generation forecasting, energy management systems, voltage control, anomaly detection, and energy trading. The identification and classification of the data-driven methods are conducted in collaboration with federated learning implemented in these services. Furthermore, the interrelation is mapped between the categories of machine learning, data-driven techniques, the application domain, and application services. Finally, the future opportunities and challenges of applying federated learning in the energy sector will be discussed.

show abstract

Section: Forecastingmentioning

confidence: 99%

Section: ) Demand Forecastingmentioning

confidence: 99%

Section: ) Demand Forecastingmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Advancing Power System Services With Privacy-Preserving Federated Learning Techniques: A Review

Zheng,

Sumper,

Aragüés-Peñalba

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…These models possess the ability to capture long-term dependencies and are suitable for handling nonlinear and non-stationary time series data. However, due to their complexity, computational expenses, and the often substantial amount of data required, their application in certain carbon neutrality research contexts can be challenging (Shen et al, 2022). On another front, the Transformer model is emerging as a notable contender in the field of time series forecasting.…”

Section: Introductionmentioning

confidence: 99%

ISSA-enhanced GRU-Transformer: integrating sports wisdom into the frontier exploration of carbon emission prediction

Jiang,

Liu,

et al. 2024

Front. Ecol. Evol.

View full text Add to dashboard Cite

IntroductionCarbon neutrality has become a key strategy to combat global climate change. However, current methods for predicting carbon emissions are limited and require the development of more effective strategies to meet this challenge. This is especially true in the field of sports and competitions, where the energy intensity of major events and activities means that time series data is crucial for predicting related carbon emissions, as it can detail the emission patterns over a period of time.MethodIn this study, we introduce an artificial intelligence-based method aimed at improving the accuracy and reliability of carbon emission predictions. Specifically, our model integrates an Improved Mahjong Search Algorithm (ISSA) and GRU-Transformer technology, designed to efficiently process and analyze the complex time series data generated by sporting events. These technological components help to capture and parse carbon emission data more accurately.ResultsExperimental results have demonstrated the efficiency of our model, which underwent a comprehensive evaluation involving multiple datasets and was benchmarked against competing models. Our model outperformed others across various performance metrics, including lower RMSE and MAE values and higher R2 scores. This underscores the significant potential of our model in enhancing the accuracy of carbon emission predictions.DiscussionBy introducing this new AI-based method for predicting carbon emissions, this study not only provides more accurate data support for optimizing and implementing carbon neutrality measures in the sports field but also improves the accuracy of time series data predictions. This enables a deeper understanding of carbon emission trends associated with sports activities. It contributes to the development of more effective mitigation strategies, making a significant contribution to global efforts to reduce carbon emissions.

show abstract

A Survey of Integrating Federated Learning with Smart Grids: Application Prospect, Privacy Preserving and Challenges Analysis

Tang

Yan

Wang

et al. 2023

Communications in Computer and Information Science

View full text Add to dashboard Cite

Federated Learning for Long-term Forecasting of Electricity Consumption towards a Carbon-neutral Future

Cited by 7 publications

References 10 publications

Advancing Power System Services With Privacy-Preserving Federated Learning Techniques: A Review

Advancing Power System Services With Privacy-Preserving Federated Learning Techniques: A Review

ISSA-enhanced GRU-Transformer: integrating sports wisdom into the frontier exploration of carbon emission prediction

A Survey of Integrating Federated Learning with Smart Grids: Application Prospect, Privacy Preserving and Challenges Analysis

Contact Info

Product

Resources

About