Open research questions in Energy Load and Power Forecasting

What the literature leaves open

• Further research could explore the application of graph neural networks to other complex systems. Investigating the use of different attention mechanisms and their impact on the model's performance.

  Dynamic cost prediction of power technical transformation projects based on graph neural networks and multi-scale attention mechanisms · 2026 · DOI

• The lack of a model that can capture complex interactions between project elements. The inability of additive models to reproduce coupled cost shifts. The need for a model that can handle nonlinear associations and temporal dynamism.

  Dynamic cost prediction of power technical transformation projects based on graph neural networks and multi-scale attention mechanisms · 2026 · DOI

• This study addressed the enduring challenge of stochastic volatility in residential Short-Term Load Forecasting. We argued that conventional methods for forecasting immediate power consumption are fundamentally limited by the irreducible noise associated with human behavior. Therefore, we proposed a paradigm shift toward Denoised Trend Forecasting, implemented through a novel Robust Residual-Attention Network.

  Reducing Phase Lag and Noise in Residential Load Forecasting with a Hybrid Residual-Attention Model · 2026 · DOI

• Future work will examine the model using geographically diverse datasets and different forecasting horizons, extend the framework to probabilistic load forecasting, and integrate the forecasting results with unit-commitment and reserve-allocation models to quantify their operational and economic benefits. 91%, respectively, confirming that the observed improvement is not limited to a single favorable initialization.

  A Short-Term Electric Load Forecasting Method Integrating Grouped Exogenous Variable Recalibration and Calendar–Causal Dual Correction · 2026 · DOI

• Although the load forecasting method for distribution networks based on power big data technology and Stacking ensemble learning proposed in this paper has achieved high prediction accuracy and good stability in experimental verification, there are still several limitations that need to be further studied and improved. The stability of the model under different data distributions still needs to be further explored.

  An integrated load forecasting learning approach for distribution networks based on electricity big data technology · 2026 · DOI

• Future research may focus on incorporating advanced attention mechanisms or Transformer-based architectures to improve the model’s ability to capture long-range dependencies in time-series 1 3N.

  Hybrid deep learning model BiLSTM–BiGRU for energy prediction using recursive feature elimination and cuckoo search · 2026 · DOI

• Although the independent test results confirm accurate prediction under the studied operating conditions, the validation remains limited to a single LabVolt experimental platform and the investigated wind speed range.

  Neural Network-Based Real-Time Wind Energy Estimation · 2026 · DOI

• Methodological limitations, including the lack of a comprehensive review of AI-based techniques. Data-related constraints, including data scarcity and privacy concerns. Behavioral complexity, including the irregularity and unpredictability of household electricity consumption patterns.

  AI-driven household electricity load forecasting: challenges, methods, and future directions · 2026 · DOI

• First, further validation on a variety of hydroelectric data sets would enhance the external validity of the framework across different operational systems.

  Renewable energy consumption forecasting using the Swordfish movement optimization algorithm (SMOA) for feature selection and hyperparameter tuning · 2026 · DOI

• This paper addresses key challenges in carbon emission time series forecasting, including nonlinearity, non-stationarity, and the difficulty of adaptively determining LSTM hyperparameters. Based on the framework of the stochastic social learning optimization algorithm, an elite-guided collaborative stochastic social learning optimization algorithm (EGC-SSLO) is proposed.

  Elite-Guided Collaborative Stochastic Social Learning Optimization for LSTM-Based Carbon Emission Forecasting · 2026 · DOI

• Future research will focus on extending the multi-scale residual framework to probabilistic load forecasting by integrating deep quantile regression (QR) components to construct reliable prediction intervals and quantify forecasting uncertainties.

  A Multi-Scale Convolutional Neural Network with Residual Blocks and LSTM for Multi-Step Forecasting of Electricity Load · 2026 · DOI

• Additionally, although the authors used HP filtering based on previous research, a significant number of studies have employed alternative decomposition techniques, and it remains to be seen whether these techniques outperform HP filtering.

  An Integrated regARIMA–HP Filter–CNN Framework with Attention Mechanism for Monthly Electricity Demand Forecasting · 2026 · DOI

• In previous experiments, three classic metrics–MAPE, RMSE and RMSPE–were adopted to verify its effectiveness, yet it remains unclear whether the method is only applicable to the aforementioned metrics or possesses generalizability across different evaluation metrics.

  An integrated wind speed prediction and model selection system for multi-scenarios incorporating improved fuzzy comprehensive evaluation · 2026 · DOI

• Our findings align with those of, who echoed that while the LSTM model attains high predictive accuracy, it provides very low R2 values, and posit that its model architecture is capable of effectively forecasting return direction yet is still insufficient in capturing the magnitude and underlying structure of return dynamics.

  Multi-Task Time Series Forecasting of Plant Load Losses: A Comparative Study of a Temporal Fusion Transformer and LSTM · 2026 · DOI

• Future work will focus on additional wind farms and seasonal conditions, multi-step forecasting horizons, systematic sensitivity analysis, temporal cross-validation, and computational efficiency optimization to further evaluate the robustness and practical applicability of the proposed framework.

  Short-Term Wind Power Forecasting Based on ICEEMDAN-EWT Decomposition, Complexity-Adaptive Networks, and Error Correction · 2026 · DOI

• Prior work has focused on forecasting load-rate using various machine learning models, but the contributions of meteorological features are not well understood. There is a need for a more accurate and interpretable forecasting model. The study aims to fill this gap by providing a Random Forest approach with feature attribution analysis.

  Prediction of Transformer Load-Rate and Analysis of Meteorological Feature Contributions for Sustainable Operation of Regional Substations: An Explainable Machine Learning Approach · 2026 · DOI

• Then, wind and photovoltaic power generation are fully utilized, and when the power supply is still insufficient to meet the electrical load, battery and fuel cell generation, subsequent generation from the microturbine, or the purchase of power from the main grid are considered to meet the load.

  Optimization of Microgrid Dispatching by Integrating Photovoltaic Power Generation Forecast · 2025 · DOI

• Compounding these issues is the lack of standardisation in back-testing and real-time evaluation protocols, with inappropriate use of metrics like MAPE limiting the evaluation and comparison of models.

  A Review of Electricity Price Forecasting Models in the Day-Ahead, Intra-Day, and Balancing Markets · 2025 · DOI

• Future research will explore transfer learning and domain adaptation methods to adapt pre-trained models to new climates with minimal data requirements.

  Enhancing Smart and Zero-Carbon Cities Through a Hybrid CNN-LSTM Algorithm for Sustainable AI-Driven Solar Power Forecasting (SAI-SPF) · 2025 · DOI

• Future research will focus on extending the UMAP-CDDA framework to other renewable energy forecasting tasks, such as solar power prediction and hydroelectric generation modeling, where temporal dependencies and dataset variability present similar challenges.

  Cross-Dataset Data Augmentation Using UMAP for Deep Learning-Based Wind Speed Prediction · 2025 · DOI

• As demonstrated in Figure 18, a major limitation of this method is its inability to adequately capture consumption behaviour during operational interruptions caused by holidays.

  Direct Multiple-Step-Ahead Forecasting of Daily Gas Consumption in Non-Residential Buildings Using Wavelet/RNN-Based Models and Data Augmentation— Comparative Evaluation · 2025 · DOI

• For the purpose of producing more accurate forecasts where data are scarce, we leverage XGBoost’s ability to pre-train on a large dataset (often synthesized by SITS) and then boost training using a smaller, less divergent dataset from the target location.

  Utility of Domain Adaptation for Biomass Yield Forecasting · 2025 · DOI

• To address the challenges of model robustness under extreme seasonal conditions and adaptability to climate variability, we propose the following directions for future research. These approaches aim to improve predictive reliability across abnormal years, new climate regimes, and geographically diverse environments.

  Utility of Domain Adaptation for Biomass Yield Forecasting · 2025 · DOI

• Future research should focus on exploring seasonal variations in NDVI-LST relationships, integrating additional environmental variables such as soil moisture and precipitation, and assessing how extreme climate events impact temperature–vegetation interactions.

  Global NDVI-LST Correlation: Temporal and Spatial Patterns from 2000 to 2024 · 2025 · DOI

• Traditional methods do not fully capture complex temporal features and important information in power grid load data. The need for a model that can efficiently and precisely forecast short-term power grid load.

  Power Grid Load Forecasting Using a CNN-LSTM Network Based on a Multi-Modal Attention Mechanism · 2025 · DOI

Most-cited papers in Energy Load and Power Forecasting

• Electricity price forecasting: A review of the state-of-the-art with a look into the future · International Journal of Forecasting · 2014 · 1,438 citations
• Probabilistic electric load forecasting: A tutorial review · International Journal of Forecasting · 2016 · 1,239 citations
• Artificial intelligence in sustainable energy industry: Status Quo, challenges and opportunities · Journal of Cleaner Production · 2021 · 1,148 citations
• A review of wind speed and wind power forecasting with deep neural networks · Applied Energy · 2021 · 932 citations
• Probabilistic energy forecasting: Global Energy Forecasting Competition 2014 and beyond · International Journal of Forecasting · 2016 · 909 citations
• Artificial Neural Networks Based Optimization Techniques: A Review · Electronics · 2021 · 696 citations
• A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting · International Journal of Forecasting · 2019 · 676 citations
• Data-driven probabilistic machine learning in sustainable smart energy/smart energy systems: Key developments, challenges, and future research opportunities in the context of smart grid paradigm · Renewable and Sustainable Energy Reviews · 2022 · 664 citations
• Energy Forecasting: A Review and Outlook · IEEE Open Access Journal of Power and Energy · 2020 · 624 citations
• A survey on deep learning methods for power load and renewable energy forecasting in smart microgrids · Renewable and Sustainable Energy Reviews · 2021 · 613 citations

Most recent work

• PowerMistral: A data-efficient wind power forecasting framework leveraging pre-trained large language models · Applied Energy · 2026
• An integrated TCN-ECA-BiLSTM and TD3 framework for high-accuracy forecasting and optimal load allocation in combined heat and power systems · Energy · 2026
• Development of a physics-guided bidirectional long short-term memory for wind power forecasting · Engineering Applications of Artificial Intelligence · 2026
• TSBMNet-XAI interpretable multi-step offshore wind power forecasting for enhanced energy storage integration · Computers & Electrical Engineering · 2026
• Transfer learning techniques on temporal fusion transformers for short-term building load forecasting under limited data conditions · Energy and Buildings · 2026
• Short-term prediction of wind vector at multi-heights via deep learning techniques based on marine measurements from Light-Detection-and-Ranging device · Physics of Fluids · 2026
• Q-learning driven artificial lemming algorithm with elite pool and dynamic vertical crossover strategy: a case study in wind power forecasting · Expert Systems with Applications · 2026
• EPformer: Unlocking day-ahead electricity price forecasting accuracy using the time–frequency domain feature learning strategy considering renewable energy · Renewable Energy · 2026
• DMoE-LLM: A dual-branch mixture-of-experts framework with large language models for wind power forecasting · Expert Systems with Applications · 2026
• Short-term load forecasting based on the improved bi-directional long short-term memory model · International Journal of Electrical Power & Energy Systems · 2026

Related topics in Engineering