aceOptimizing solar and wind forecasting with iHow optimization algorithm and multi-scale attention networks
Deep learning models often encounter two key challenges in developing intelligent and scalable forecasting frameworks for renewable energy systems: input feature space dimensionality and sensitivity to hyperparameter settings. These limitations increase computational cost and compromise generalization and robustness. This paper presents a hybrid deep learning-optimization framework that leverages cognitively inspired metaheuristics to address these challenges, employing the Binary iHow Optimization Algorithm (bi…
"Wind Turbines ^ Solar Panels at Upper Holton Wind ^ Solar Farm - geograph.org.uk - 4618380" by Geographer is licensed under CC BY-SA 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by-sa/2.0/.
Researchers developed a hybrid deep learning-optimization framework for renewable forecasting that pairs a Multi-Scale Attention Network with cognitively inspired metaheuristics. The Binary iHow algorithm selects features and the continuous iHow algorithm tunes hyperparameters, targeting high-dimensional inputs and sensitivity issues in wind and solar time series.
By the March 2026 publication date, the framework had demonstrated lower forecasting errors and reduced misclassification rates compared to baseline and other metaheuristics, suggesting potential for more adaptive and scalable prediction in smart grids. Remaining questions include performance on diverse real-world grid datasets, operational deployment costs, and robustness under extreme weather variability not detailed in the supplied text.
- Framework combines Binary iHow Optimization Algorithm for feature selection and continuous iHow for hyperparameter tuning of Multi-Scale Attention Network.
- MSAN backbone is designed to capture multi-scale temporal dependencies from short-term fluctuations to long-term seasonal patterns in renewable time series.
- Feature selection reduced average misclassification rate to 0.3925 for wind and 0.4161 for solar while identifying compact interpretable subsets.
- Optimized model outperformed HHO, GWO, PSO, and JAYA metaheuristics on forecasting accuracy.
Hybrid framework using Binary iHow for feature selection and iHow for hyperparameter tuning of a Multi-Scale Attention Network reduced forecasting error for wind and solar generation and improved computational scalability for smart-grid management.
The rundown
The authors address dimensionality of input features and sensitivity to hyperparameters that raise cost and hurt generalization in renewable forecasting.
They implement biHOW to select compact feature subsets and iHOW to tune architectural and training parameters of the MSAN backbone.
Reported results include MSE of 0.0105 for wind and 0.0976 for solar before final tuning, with further MSE reduction after iHOW tuning and lower misclassification rates after feature selection.
Sources
- Peer-reviewedScientific Reports2026-03-10
How should this claim be treated?
The debate