Search papers, labs, and topics across Lattice.
This study introduces a delta-based target reformulation approach for short-term electricity load forecasting, shifting the focus from predicting absolute load values to forecasting the change in load between consecutive time steps. By leveraging multi-year, real-world electricity load data from India and incorporating meteorological and calendar features, the authors demonstrate that this method significantly enhances forecasting accuracy, particularly for hour-ahead predictions. The results reveal over 50% reductions in Mean Absolute Percentage Error (MAPE) compared to traditional absolute formulations, highlighting the model- and horizon-dependent effectiveness of this approach.
Delta-based reformulation boosts hour-ahead load forecasting accuracy by over 50%, reshaping how we approach electricity demand prediction.
Accurate short-term electricity load forecasting is critical for the reliable and economic operation of modern power systems, under non-stationarity arising from weather variability, calendar effects, and evolving consumption patterns. While deep learning models such as LSTMs and Transformers show promising performance, most existing studies focus on direct absolute load prediction without explicitly addressing target non-stationarity. Motivated by classical time-series differencing techniques in ARIMA models, this paper investigates a delta-based target reformulation for short-term electricity load forecasting using deep learning. Instead of directly predicting absolute load values, the proposed formulation trains models to predict the change in load between consecutive time steps, with final forecasts reconstructed using the last observed load. This aims to stabilize the learning target and reduce forecasting difficulty. Using multi-year, hourly real-world electricity load data from India, augmented with meteorological variables from the NASA POWER project and calendar features, this study evaluates LSTM and Transformer models under both formulations, benchmarking them against LightGBM. Experiments are conducted for hour-ahead and day-ahead horizons, assessing performance via Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). Results show that delta-based reformulation consistently improves forecasting accuracy for hour-ahead prediction across all evaluated models, yielding MAPE reductions of over 50% compared to absolute formulations. For day-ahead forecasting, delta targets specifically benefit deep sequence models (LSTM and Transformer), while LightGBM remains competitive under the absolute formulation. These findings indicate that while delta reformulation is a powerful inductive bias for neural networks, its efficacy is model- and horizon-dependent.
