A Transformer-Based Method of Multienergy Load Forecasting in Integrated Energy System

Wang, Chen; Wang, Ying; Ding, Zhetong; Zheng, Tao; Hu, Jiangyi; Zhang, Kaifeng

doi:10.1109/tsg.2022.3166600

Cited by 117 publications

(29 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Voß et al [19] compared two forecasting strategies, i.e., one local model for all consumers and one global model for each consumer, for individual consumer load forecasting and reported that using a single model for all consumers yielded superior performance. Wang et al [20] proposed a transformer-based model for forecasting over different types of loads simultaneously and explored the impact of attention mechanisms for this task. Han et al [21] introduced a shortterm forecasting model for individual residential loads based on deep learning and K-means clustering.…”

Section: Related Workmentioning

confidence: 99%

A Global Modeling Framework for Load Forecasting in Distribution Networks

Grabner

Wang

Wen

et al. 2023

IEEE Trans. Smart Grid

View full text Add to dashboard Cite

With the increasing numbers of smart meter installations, scalable and efficient load forecasting techniques are critically needed to ensure sustainable situation awareness within the distribution networks. Distribution networks include a large amount of different loads at various aggregation levels, such as individual consumers, low-voltage feeders, and transformer stations. It is impractical to develop individual (or so-called local) forecasting models for each load separately. Additionally, such local models also (i) (largely) ignore the strong dependencies between different loads that might be present due to their spatial proximity and the characteristics of the distribution network, (ii) require historical data for each load to be able to make forecasts, and (iii) are incapable of adjusting to changes in the load behavior without retraining. To address these issues, we propose a global modeling framework for load forecasting in distribution networks that, unlike its local competitors, relies on a single global model to generate forecasts for a large number of loads. The global nature of the framework, significantly reduces the computational burden typically required when training multiple local forecasting models, efficiently exploits the cross-series information shared among different loads, and facilitates forecasts even when historical data for a load is missing or the behavior of a load evolves over time. To further improve on the performance of the proposed framework, an unsupervised localization mechanism and optimal ensemble construction strategy are also proposed to localize/personalize the global forecasting model to different load characteristics. Our experimental results show that the proposed framework outperforms naive benchmarks by more than 25% (in terms of Mean Absolute Error) on real-world dataset while exhibiting highly desirable characteristics when compared to the local models that are predominantly used in the literature. All source code and data are made publicly available to enable reproducibility: https://github.com/mihagrabner/GlobalModelingFramework.

show abstract

Section: Related Workmentioning

confidence: 99%

A Global Modeling Framework for Load Forecasting in Distribution Networks

Grabner

Wang

Wen

et al. 2023

IEEE Trans. Smart Grid

View full text Add to dashboard Cite

show abstract

“…Since statistical methods require high stability of time series, so it is difficult to reflect the non-linear influence of weather, events and other factors. In comparison, the machine learning algorithms such as fuzzy inference (Wang et al, 2022), support vector machines, artificial neural networks (Fan et al, 2022) and cluster analysis (Shen et al, 2016) have demonstrated their capability of improving the non-linear fitting ability of the models. It is believed that the computational model-based load capacity prediction methods have a good prospect.…”

Section: Computational Model-based Predictionmentioning

confidence: 99%

Load capability assessment and enhancement for transformers with integration of large-scale renewable energy: A brief review

Kang

et al. 2022

Front. Energy Res.

View full text Add to dashboard Cite

With the rapid development of large-scale distributed renewable energy in China, the load factor of distribution transformer increases year by year, this directly become a new challenge to the load capacity of the transformer. Once subjected to this condition for a long time will have an impact on the hotspot temperature and consequently, the useful life of the transformer. In this regard, load capacity assessment of the distribution transformer becomes an essential work for unit upgrading or even replacement. This article systematically summarized the evaluation methods of transformer load capacity, as well as the load capacity prediction and dynamic capacity enhancement strategies of transformer. Discussions for the future work regarding this topic are also carried out.

show abstract

“…Wang C [8] used Transformer for multi-energy load prediction. Li W [9] fused Ensemble Empirical Mode Decomposition (EEMD), regression neural network and particle swarm optimization of SVR algorithm to propose EEMD-SCGRNN-PSVR hybrid power prediction model. Deng T Y [10] used EEMD to decompose load series and proposed EEMD-GRU-MLR combined forecasting model for load forecasting.…”

Section: Introductionmentioning

confidence: 99%

Research on power purchase forecasting technology based on EEMD-GRU-RF

Li,

Sun,

et al. 2024

Ninth International Conference on Energy Materials and Electrical Engineering (ICEMEE 2023)

View full text Add to dashboard Cite

The power purchase forecast can help the power system make reasonable power pre-purchase plans and achieve scientific coordination of grid planning and construction. This paper proposes a forecast model of power purchase based on EEMD-GRU-RF. The model first decomposes the original power purchase series into high and low frequency components using EEMD, and then selects the random forest algorithm to predict the low frequency components of power purchase using meteorological, fault work order, historical power consumption series and holiday information as features, and uses GRU network to achieve the prediction of high frequency eigenmodes, and finally reconstructs the superposition of both to achieve power purchase prediction. The experimental results show that the evaluation indexes RMSE, MAE and R2 of the proposed algorithm are 0.019, 0.011 and 0.976, respectively, which are better than other mainstream power purchase prediction algorithms and can achieve higher accuracy of power purchase prediction.

show abstract

A Transformer-Based Method of Multienergy Load Forecasting in Integrated Energy System

Cited by 117 publications

References 26 publications

A Global Modeling Framework for Load Forecasting in Distribution Networks

A Global Modeling Framework for Load Forecasting in Distribution Networks

Load capability assessment and enhancement for transformers with integration of large-scale renewable energy: A brief review

Research on power purchase forecasting technology based on EEMD-GRU-RF

Contact Info

Product

Resources

About