The growing ubiquity of electrical load data sourced from utility companies has led to a growing interest in utilizing this data for long-term and short-term load forecasting. More specifically, accurate and efficient forecasts are valued greatly for their ability to reduce overhead and ongoing costs as well as harmful environmental effects such as CO2 emissions and other forms of pollution. However, the uncertainty exhibited in prediction often proves a barrier to practical adoption. To this end, a lightweight deep learning model is proffered for short-term load forecasting to address this issue. Kalman filter-based smoothing is used to restructure the data which is fed into the model along with the original sample, sequenced in the form of 24-hour lagged observations to more accurately simulate, day to day data entry. The model is then used to forecast load demand over a 168-hour period (i.e.,1 week). The proposed approach has been assessed using real utility data, more specifically utility load data from GEFCom2014-E(Extended) dataset, and the results establish that the variations and uncertainties in utility load data may be better comprehended via use of the proposed approach and that competitive performance may be obtained.
Load Forecasting Using Deep Learning
02.12.2021
537468 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
A deep learning short-term load forecasting method for extreme scenarios
British Library Conference Proceedings | 2023
|Electric vehicle charging demand forecasting using deep learning model
Taylor & Francis Verlag | 2022
|Deep Learning for Flight Demand and Delays Forecasting
AIAA | 2021
|A Causal Deep Learning Framework for Traffic Forecasting
IEEE | 2023
|