Learning Day-Ahead Electricity Price Dynamics in Renewable Power Systems: A Time-Adaptive Segmented Multi-Model Ensemble Forecasting Approach

Climate change and energy security concerns are accelerating the shift from conventional generation to renewable power systems, where the variable output of wind and photovoltaic generation amplifies the volatility and complexity of day-ahead electricity prices, making accurate price forecasting essential for energy saving, renewable energy integration and sustainable power system operation. This paper proposes an adaptive ensemble electricity price forecasting framework that integrates Random Forest, XGBoost and CNN-LSTM through a validation-based mixture selection strategy. A closed candidate space includes equal weighting, individual experts and non-negative least-squares combinations, with shrinkage towards equal weighting calibrated through forward-chaining cross-validation. State identification, gating, expert attention, similar-period correction and temporal consistency are implemented as optional modules and retained only when they improve validation performance. The framework is evaluated using seasonal rolling forecasts for the Shaanxi electricity spot market, with similarity retrieval restricted to historical observations to prevent future information leakage. The proposed ensemble achieves an overall mean absolute percentage error (MAPE) of 27.16%, against 30.14% for Random Forest, 32.80% for XGBoost and 32.02% for CNN-LSTM, and also better than six additional baseline methods. Comparative and ablation analyses indicate that the framework combines complementary expert predictions while allowing the contribution of each component to be assessed. These results support adaptive ensemble forecasting under varying electricity market conditions.

Authors

Institutions

Publication Details

Journal
Sustainability
Published
2026-11-18
DOI
https://doi.org/10.3390/su182211720
Primary Topic
Energy Load and Power Forecasting
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Learning Day-Ahead Electricity Price Dynamics in Renewable Power Systems: A Time-Adaptive Segmented Multi-Model Ensemble Forecasting Approach

Tao Ding, Daen Bao, Linlin Wang, Zhuo Wang et al.
2 links
Sustainability
Energy Load and Power Forecasting
article

Learning Day-Ahead Electricity Price Dynamics in Renewable Power Systems: A Time-Adaptive Segmented Multi-Model Ensemble Forecasting Approach

Tao Ding, Daen Bao, Linlin Wang, Zhuo Wang, Shiwei He, Tong Yu
article en

Abstract

Climate change and energy security concerns are accelerating the shift from conventional generation to renewable power systems, where the variable output of wind and photovoltaic generation amplifies the volatility and complexity of day-ahead electricity prices, making accurate price forecasting essential for energy saving, renewable energy integration and sustainable power system operation. This paper proposes an adaptive ensemble electricity price forecasting framework that integrates Random Forest, XGBoost and CNN-LSTM through a validation-based mixture selection strategy. A closed candidate space includes equal weighting, individual experts and non-negative least-squares combinations, with shrinkage towards equal weighting calibrated through forward-chaining cross-validation. State identification, gating, expert attention, similar-period correction and temporal consistency are implemented as optional modules and retained only when they improve validation performance. The framework is evaluated using seasonal rolling forecasts for the Shaanxi electricity spot market, with similarity retrieval restricted to historical observations to prevent future information leakage. The proposed ensemble achieves an overall mean absolute percentage error (MAPE) of 27.16%, against 30.14% for Random Forest, 32.80% for XGBoost and 32.02% for CNN-LSTM, and also better than six additional baseline methods. Comparative and ablation analyses indicate that the framework combines complementary expert predictions while allowing the contribution of each component to be assessed. These results support adaptive ensemble forecasting under varying electricity market conditions.

SustainabilityVol. 18(22)
CHN Energy (China) (CN), Xi'an Jiaotong University (CN)
Climate action, Affordable and clean energy, Industry, innovation and infrastructure
Energy Load and Power Forecasting
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon