Federated Learning based Energy Demand Prediction with Clustered Aggregation

Tun, Ye Lin; Thar, Kyi; Thwal, Chu Myaet; Hong, Choong Seon

doi:10.1109/bigcomp51126.2021.00039

Cited by 39 publications

(11 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ref. [17] proposed a prediction model based on the analysis of the energy usage patterns of the households. They used a clustering algorithm to group the households with similar energy consumption patterns and then trained a prediction model for each cluster.…”

Section: Related Workmentioning

confidence: 99%

“…FL enables multiple clients to train a machine learning model collaboratively without sharing their raw data. FL has been successfully applied to energy load prediction for smart homes, improving prediction accuracy while preserving data privacy [15][16][17]. Current FL approaches do not take into consideration the nature and additional properties of data.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Data Aging Matters: Federated Learning-Based Consumption Prediction in Smart Homes via Age-Based Model Weighting

et al. 2023

View full text Add to dashboard Cite

Smart homes, powered mostly by Internet of Things (IoT) devices, have become very popular nowadays due to their ability to provide a holistic approach towards effective energy management. This is made feasible via the deployment of multiple sensors, which enables predicting energy consumption via machine learning approaches. In this work, we propose FedTime, a novel federated learning approach for predicting smart home consumption which takes into consideration the age of the time series datasets of each client. The proposed method is based on federated averaging but aggregates local models trained on each smart home device to produce a global prediction model via a novel weighting scheme. Each local model contributes more to the global model when the local data are more recent, or penalized when the data are older upon testing for a specific residence (client). The approach was evaluated on a real-world dataset of smart home energy consumption and compared with other machine learning models. The results demonstrate that the proposed method performs similarly or better than other models in terms of prediction error; FedTime achieved a lower mean absolute error of 0.25 compared to FedAvg. The contributions of this work present a novel federated learning approach that takes into consideration the age of the datasets that belong to the clients, experimenting with a publicly available dataset on grid import consumption prediction, while comparing with centralized and decentralized baselines, without the need for data centralization, which is a privacy concern for many households.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Data Aging Matters: Federated Learning-Based Consumption Prediction in Smart Homes via Age-Based Model Weighting

et al. 2023

View full text Add to dashboard Cite

show abstract

“…The results obtained revealed that federated learning could generate high-performance models with a significantly reduced networking load compared with a centralized model. The study [22] introduced a clustering method using household attributes to increase the convergence rate of the global model in federated learning. In the experiment, the clusterspecific global models exhibited significantly faster convergence than the model trained without clustering.…”

Section: Introductionmentioning

confidence: 99%

Residential Load Forecasting Using Modified Federated Learning Algorithm

Park

Son

2023

IEEE Access

View full text Add to dashboard Cite

The feasibility of utilizing electricity consumption data is increasing with the widespread use of advanced metering infrastructure among consumers. In addition, the need for consumer demand management and services is likely to increase further as distributed resources are activated. Load forecasting technology is essential for customer demand management. Forecasting technology using deep learning has been applied extensively in recent years. Deep-learning-based load forecasting requires data for training a forecasting model, and the available data may be limited by the Personal Information Protection Act. Federated learning has emerged as a solution to this problem. However, in federated learning, the global model in the central server is trained by aggregating the local model sent from the client without filtration. This may result in the global model overfitting or failure to converge by aggregating the model that is not trained effectively in the client. In this study, we introduce a modified federated learning algorithm to solve this problem and propose a method to forecast residential loads. The proposed method is analyzed and evaluated experimentally using smart meter data. The experimental results reveal that the proposed method improves forecasting performance and convergence. For the global model, when the number of clients was two, three, four, and five, the forecasting performances were 4. 891, 5.228, 5.488, and 5.633, respectively, on a mean absolute percentage error (MAPE) basis, showing performance improvements of 28.4%, 10.0%, 16.3%, and 5.8%.INDEX TERMS Residential load forecasting, federated learning, deep learning, LSTM.

show abstract

“…electricity imports and export, population, installed capacity, and gross electricity generation Ramsami and King [11] electricity demand adaptive network-based fuzzy inference system, ANN, RNN historical electricity data Bendaoud et al [12] electrical energy demand CNN load profile Sen et al [13] electricity consumption ANN-SVM population, GDP, inflation rate, and unemployment rate Tun et al [14] energy demand RNN past energy usage data Kolokas et al [15] energy demand and generation multi-step time series forecasting past energy data and weather forecasts Al-Musaylh et al [16] electricity demand online sequential extreme learning machine (OS-ELM) climate variables Moustris et al [17] load demand ANN meteorological data, cooling power index (CP)…”

Section: Introductionmentioning

confidence: 99%

Electricity Demand Forecasting with Use of Artificial Intelligence: The Case of Gokceada Island

2022

View full text Add to dashboard Cite

This study reviews a selection of approaches that have used Artificial Neural Networks (ANN), Particle Swarm Optimization (PSO), and Multi Linear Regression (MLR) to forecast electricity demand for Gokceada Island. Artificial Neural Networks, Particle Swarm Optimization, and Linear Regression methods are frequently used in the literature. Imports, exports, car numbers, and tourist-passenger numbers are used as based on input values from 2014 to 2020 for Gokceada Island, and the electricity energy demands up to 2040 are estimated as an output value. The results obtained were analyzed using statistical error metrics such as R2, MSE, RMSE, and MAE. The confidence interval analysis of the methods was performed. The correlation matrix is used to show the relationship between the actual value and method outputs and the relationship between independent and dependent variables. It was observed that ANN yields the highest confidence interval of 95% among the method utilized, and the statistical error metrics have the highest correlation for ANN methods between electricity demand output and actual data.

show abstract

Federated Learning based Energy Demand Prediction with Clustered Aggregation

Cited by 39 publications

References 9 publications

Data Aging Matters: Federated Learning-Based Consumption Prediction in Smart Homes via Age-Based Model Weighting

Data Aging Matters: Federated Learning-Based Consumption Prediction in Smart Homes via Age-Based Model Weighting

Residential Load Forecasting Using Modified Federated Learning Algorithm

Electricity Demand Forecasting with Use of Artificial Intelligence: The Case of Gokceada Island

Contact Info

Product

Resources

About