A Modularity-Enhanced Echo State Network for Nonlinear Wind Energy Predicting DOI Creative Commons

Shengqin Yue,

Zhili Zhao,

Taavi Lai

et al.

Energies, Journal Year: 2025, Volume and Issue: 18(7), P. 1858 - 1858

Published: April 7, 2025

With the rapid growth of wind power generation, accurate energy prediction has emerged as a critical challenge, particularly due to highly nonlinear nature speed data. This paper proposes modularized Echo State Network (MESN) model improve forecasting. To enhance generalization, data is first decomposed into time series components, and Modes-cluster employed extract trend patterns pre-train ESN output layer. Furthermore, Turbines-cluster groups turbines based on their characteristics, enabling within same category share matrix for prediction. An integration module then introduced aggregate predicted results, while modular design ensures efficient task allocation across different modules. Comparative experiments with other neural network models demonstrate effectiveness proposed approach, showing that statistical RMSE parameter error reduced by an average factor 2.08 compared traditional models.

Language: Английский

A Modularity-Enhanced Echo State Network for Nonlinear Wind Energy Predicting DOI Creative Commons

Shengqin Yue,

Zhili Zhao,

Taavi Lai

et al.

Energies, Journal Year: 2025, Volume and Issue: 18(7), P. 1858 - 1858

Published: April 7, 2025

With the rapid growth of wind power generation, accurate energy prediction has emerged as a critical challenge, particularly due to highly nonlinear nature speed data. This paper proposes modularized Echo State Network (MESN) model improve forecasting. To enhance generalization, data is first decomposed into time series components, and Modes-cluster employed extract trend patterns pre-train ESN output layer. Furthermore, Turbines-cluster groups turbines based on their characteristics, enabling within same category share matrix for prediction. An integration module then introduced aggregate predicted results, while modular design ensures efficient task allocation across different modules. Comparative experiments with other neural network models demonstrate effectiveness proposed approach, showing that statistical RMSE parameter error reduced by an average factor 2.08 compared traditional models.

Language: Английский

Citations

0