Journal list menu
Peak load is an important concept in the electric power industry, with applications in demand response, energy trading, system planning, and so forth. While majority of the load forecasting studies in the literature focus on minimizing errors for the entire load profile, not many are devoted to peaks or peak periods. How to evaluate peak load forecasts? Does a better model for the entire load profile offers a better accuracy in peak periods? Are there methods specifically targeting peak loads? Are the models selected based on performance in magnitudes also lead to precise peak timing forecasts? How to best incorporate peak load forecasts to make business decisions? Many questions like these still await a solid answer. This Virtual Collection in IET Smart Grid invites researchers and practitioners to develop and propose novel methods and models on peak load forecasting and applications in Smart Grids.
Edited by :
Tao Hong, University of North Carolina at Charlotte, USA, E: [email protected]
Yi Wang, The University of Hong Kong, China, E: [email protected]
Jingrui Xie, National Grid, USA, E-mail: [email protected]
Pu Wang, Duke Energy Corporation, USA, E-mail: [email protected]
Export Citations
Table of Contents
Privacy-preserving peak time forecasting with Learning to Rank XGBoost and extensive feature engineering
- First Published: 22 October 2023

The authors apply a dedicated Learning to Rank XGBoost algorithm to forecast peak times with only ranks of loads instead of absolute load magnitudes as target feature, thereby offering potential privacy-preserving properties. In addition, the authors introduce an extensive feature engineering process and Bayesian hyperparameter optimisation.
Medium-term forecasting of daily aggregated peak loads from heat pumps using clustering-based load duration curves to calculate the annual impact on medium to low voltage transformers
- First Published: 05 October 2023

The authors calculate the annual impact on medium to low voltage (MV/LV) transformers based on a medium-term forecast of daily aggregated peak loads of heat pumps using the number, their rated power, an estimate of a load duration curve on a cold day, and the daily average ambient temperature. The forecast differentiates between daily aggregated peak loads of heat pumps with different heating demands and allows studying the varying impact due to seasonal weather differences.
Predicting the magnitude and timing of peak electricity demand: A competition case study
- First Published: 21 December 2023

Global forecasting competitions provide a unique opportunity to compare multiple methodologies under common performance criteria and incentives. This paper details the methodology and results from the BigDEAL Challenge 2022 forecasting competition used by the team ‘peaky-finders’ and investigates the suitability of using hourly methods to forecast daily peak magnitude, time, and shape. The resulting approach provides a reproducible ensemble benchmark against which to evaluate more complex methods.
BigDEAL Challenge 2022: Forecasting peak timing of electricity demand
- First Published: 23 March 2024

BigDEAL Challenge 2022, which was devoted to short-term ex-ante peak timing forecasting, attracted 78 teams formed by 121 contestants from 27 countries. The authors introduce the competition in detail, including its precursor competitions held in the 2010s, the framework and setup, and a summary of the methods used by the participants.
Using conditional Invertible Neural Networks to perform mid-term peak load forecasting
- First Published: 26 April 2024

The cINN Forecaster consists of multiple components. First a feature extraction component extracts calendar information, processes weather data, and predicts future statistics of the time series. These features are used by the second component to generate controlled time series samples. Finally, out of these samples, the predicted time series is formed.