cs.LGFeb 21, 2026

From ARIMA to Attention: Power Load Forecasting Using Temporal Deep Learning

Authors: Suhasnadh Reddy Veluru, Sai Teja Erukude, Viswa Chaitanya Marella

Organizations: Department of Computer Science and Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India · Department of Electronics and Communication Engineering, Bharat Institute of Engineering and Technology, Hyderabad, Telangana, India

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Accurate short-term power load forecasting is important to effectively manage, optimize, and ensure the robustness of modern power systems. This paper performs an empirical evaluation of a traditional statistical model and deep learning approaches for predicting short-term energy load. Four models, namely ARIMA, LSTM, BiLSTM, and Transformer, were leveraged on the PJM Hourly Energy Consumption data. The data processing involved interpolation, normalization, and a sliding-window sequence method. Each model's forecasting performance was evaluated for the 24-hour horizon using MAE, RMSE, and MAPE. Of the models tested, the Transformer model, which relies on self-attention algorithms, produced the best results with 3.8 percent of MAPE, with performance above any model in both accuracy and robustness. These findings underscore the growing potential of attention-based architectures in accurately capturing complex temporal patterns in power consumption data.

Figures & tables

ModelMAERMSEMAPE (%)Notes
ARIMA2303008.2Traditional baseline
LSTM1452104.5Baseline DL model
BiLSTM1321954.2Slight improvement
Transformer1201803.8Best overall performance
TABLE I: Forecasting Performance of Each Model on PJM Dataset
Fig. 1: Energy Consumption Forecasting: ARIMA vs True Values for One Week. The plot compares ARIMA model predictions (red dashed line) with the actual energy consumption (black line) over 7 days.
Fig. 2: Energy Consumption Forecasting: LSTM vs. True Values for One Week. The plot shows LSTM model predictions (blue dashed line) compared to actual energy usage (black line) over a representative week.
Fig. 3: Energy Consumption Forecasting: BiLSTM vs True Values for One Week. The plot depicts BiLSTM model predictions (green dashed line) against the true energy consumption (black line) over 168 hours.
Fig. 4: Energy Consumption Forecasting: Transformer vs True Values for One Week. The Transformer model prediction (purple dashed line) closely follows actual energy consumption (black line), showing improved accuracy and robustness.
Fig. 5: Comparison of MAE and RMSE across Models
Fig. 6: Comparison of MAPE (%) across Models
More in this topic

Electricity Load Forecasting

Electricity load forecasting predicts future power demand to support grid operation, planning, and energy management.

See all 46 papers

Explore similar work

CardsList

  1. Delta-Based Target Reformulation for Short-Term Electricity Load Forecasting Using LSTM and Transformer Models

    Jun 16, 2026Vansh BansalTime Series ForecastingNon-Stationary Time Series Forecasting

  2. SolarTformer: A Transformer Based Deep Learning Approach for Short Term Solar Power Forecasting

    Apr 27, 2026Ankan Basu, Jyotiraditya Roy, Aditya Datta +2Multivariate Time Series ForecastingTime Series Forecasting

    Positional Astronomy Centre, India Meteorological Department, Kolkata, India · Department of Computer Science and Engineering, Jadavpur University, Kolkata, India · Department of Computer Science and Engineering (Artificial Intelligence and Machine Learning), Heritage Institute of Technology, Kolkata, India +2

  3. A Benchmark for Electrical Load Forecasting Across Grid Levels: Time-Series Transformers Outperform Established Methods

    Jul 17, 2026Matthias Hertel, Sebastian Pütz, Jonathan Kolar +3Time Series ForecastingForecasting Benchmarks

    Karlsruhe Institute of Technology, Germany · Helmholtz AI, Germany