.png){kind=icon}
.png){kind=icon}
Return
Advanced renewable energy forecasting under uncertainty using empirical mode decomposition and machine learning for resilient microgrid optimization
.png){kind=icon}
PRE
AI
F
DOI：10.1007/s13748-026-00425-z.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a novel renewable energy forecasting system which uses advanced Piecewise Cubic Hermite Interpolating Polynomial Empirical Mode Decomposition method with machine learning schemes to improve energy management during uncertain situations. The proposed method shows improved prediction performance because it achieves a 9.51% reduction in Mean Absolute Percentage Error and reaches an R2 value of 0.966 which surpasses the results of standard decomposition methods and baseline models such as decision trees and support vector regression. The system achieves its operation by uniting neural networks with optimization methods to form a hybrid model that predicts power output from wind and photovoltaic sources. Kernel Density Estimation is applied to quantify uncertainty by modeling the distributions of forecast errors, providing a robust framework for dealing with unpredictable renewable energy sources. A comprehensive case study employing actual wind and solar data demonstrates the efficacy of the proposed methodology. Furthermore, a new Mixed Integer Linear Programming based scheduling framework is developed to minimize operational costs, maximize battery energy storage efficiency, and improve grid resilience. The strategy achieved a 26% reduction in operational costs, enabling the efficient management of dispersed energy sources, making it highly effective for microgrid energy management. The research shows that combining modern signal processing with machine learning prediction systems produces optimal renewable energy management results for current microgrid networks which leads to dependable low-cost power systems.
Keywords:
Renewable energy forecasting
Empirical mode decomposition
Machine learning model
Kernel density estimation
Microgrid energy management
Mixed integer linear programming
Journal
P
IF:
2.4
Papers:
44
Citations:
0
Organization
No organization information available
Cited Papers