Deep-Learning Forecasting Method for Electric Power Load via Attention-Based Encoder-Decoder with Bayesian Optimization

Xue-Bo Jin,Yu-Ting Bai,Xiao-Yi Wang,Ting-Li Su,Seng Lin,Jian-Lei Kong,Wei-Zhen Zheng

doi:10.3390/en14061596

Abstract

Short-term electrical load forecasting plays an important role in the safety, stability, and sustainability of the power production and scheduling process. An accurate prediction of power load can provide a reliable decision for power system management. To solve the limitation of the existing load forecasting methods in dealing with time-series data, causing the poor stability and non-ideal forecasting accuracy, this paper proposed an attention-based encoder-decoder network with Bayesian optimization to do the accurate short-term power load forecasting. Proposed model is based on an encoder-decoder architecture with a gated recurrent units (GRU) recurrent neural network with high robustness on time-series data modeling. The temporal attention layer focuses on the key features of input data that play a vital role in promoting the prediction accuracy for load forecasting. Finally, the Bayesian optimization method is used to confirm the model’s hyperparameters to achieve optimal predictions. The verification experiments of 24 h load forecasting with real power load data from American Electric Power (AEP) show that the proposed model outperforms other models in terms of prediction accuracy and algorithm stability, providing an effective approach for migrating time-serial power load prediction by deep-learning technology.

Highlights

With the rapid economic and social development, electric power plays an increasingly important role in all aspects of humanity’s domestic life and industrial and commercial practical applications
With the temporal attention layer, this paper proposes an attention-based codec prediction model, which mainly consists of three parts: An encoder composed of a multilayer gated recurrent units (GRU) network, an attention layer, and a decoder composed on basis of multilayer GRU network
The electric power load data is from American Electric Power Company (AEP), which includes 26,280 data from 1 January 2017–1 January 2020, in which a sampling frequency is 1 h

Summary

Introduction

With the rapid economic and social development, electric power plays an increasingly important role in all aspects of humanity’s domestic life and industrial and commercial practical applications. A large number of a production order for epidemic vaccines and prevention materials, coupled with drastic cooling in cold weather this winter, have caused insufficient electric power supply in partial areas of Chinese southern, even though China is the country with the largest electricity production and the largest increase in the world [3]. To deal with these constant changes between electricity generation and consumption, it is expected that implementing efficient management operations in the electronic power systems to balance the electricity supply and demand as much as possible. Constructing accurate, robust, and fast models forecasting electric power load became the fundamental approach to achieve reliable and high-efficiency operational management of abundant power utilities, such as electricity production planning, high-voltage transmission decision-making, power load dispatch, and so on [4,5]

Methods

Results

Conclusion