Skip to main content
placeholder image

Short-Term Electricity Demand Forecasting Based on Multiple LSTMs

Journal Article


Abstract


  • In recent years, the problem of unbalanced demand and supply

    in electricity power industry has seriously affected the development

    of smart grid, especially in the capacity planning, power dispatching

    and electric power system control. Electricity demand forecasting, as a

    key solution to the problem, has been widely studied. However, electricity

    demand is influenced by many factors and nonlinear dependencies,

    which makes it difficult to forecast accurately. On the other hand, deep

    neural network technologies are developing rapidly and have been tried

    in time series forecasting problems. Hence, this paper proposes a novel

    deep learning model, which is based on the multiple Long Short-Term

    Memory (LSTM) neural networks to solve the problem of short-term

    electricity demand forecasting. Compared with autoregressive integrated

    moving average model (ARIMA) and back propagation neural network

    (BPNN), our model demonstrates competitive forecast accuracy, which

    proves that our model is promising for electricity demand forecasting.

Authors


  •   Yong, Binbin (external author)
  •   Shen, Zebang (external author)
  •   Wei, Yongqiang (external author)
  •   Shen, Jun
  •   Zhou, Qingguo (external author)

Publication Date


  • 2020

Citation


  • Yong, B., Shen, Z., Wei, Y., Shen, J. & Zhou, Q. (2020). Short-Term Electricity Demand Forecasting Based on Multiple LSTMs. Lecture Notes in Computer Science, 11691 192-200. Guangzhou, China The 10th International Conference on Brain-Inspired Cognitive Systems BICS 2019

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers1/3684

Number Of Pages


  • 8

Start Page


  • 192

End Page


  • 200

Volume


  • 11691

Place Of Publication


  • Germany

Abstract


  • In recent years, the problem of unbalanced demand and supply

    in electricity power industry has seriously affected the development

    of smart grid, especially in the capacity planning, power dispatching

    and electric power system control. Electricity demand forecasting, as a

    key solution to the problem, has been widely studied. However, electricity

    demand is influenced by many factors and nonlinear dependencies,

    which makes it difficult to forecast accurately. On the other hand, deep

    neural network technologies are developing rapidly and have been tried

    in time series forecasting problems. Hence, this paper proposes a novel

    deep learning model, which is based on the multiple Long Short-Term

    Memory (LSTM) neural networks to solve the problem of short-term

    electricity demand forecasting. Compared with autoregressive integrated

    moving average model (ARIMA) and back propagation neural network

    (BPNN), our model demonstrates competitive forecast accuracy, which

    proves that our model is promising for electricity demand forecasting.

Authors


  •   Yong, Binbin (external author)
  •   Shen, Zebang (external author)
  •   Wei, Yongqiang (external author)
  •   Shen, Jun
  •   Zhou, Qingguo (external author)

Publication Date


  • 2020

Citation


  • Yong, B., Shen, Z., Wei, Y., Shen, J. & Zhou, Q. (2020). Short-Term Electricity Demand Forecasting Based on Multiple LSTMs. Lecture Notes in Computer Science, 11691 192-200. Guangzhou, China The 10th International Conference on Brain-Inspired Cognitive Systems BICS 2019

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers1/3684

Number Of Pages


  • 8

Start Page


  • 192

End Page


  • 200

Volume


  • 11691

Place Of Publication


  • Germany