Back to Search
Start Over
Deep Learning Algorithm for Solving Interval of Weight Coefficient of Wind–Thermal–Storage System.
- Source :
- Energies (19961073); Mar2024, Vol. 17 Issue 5, p1082, 18p
- Publication Year :
- 2024
-
Abstract
- Under the premise of ensuring the safe and stable operation of a wind–thermal–storage power system, this paper proposes an optimization model aimed at improving its overall economic efficiency and effectively reducing the peak-to-valley load difference. The model transforms the multi-objective optimization problem to solve a feasible interval of weight coefficients. We introduce a novel fusion model, where a Convolutional Neural Network (CNN) is melded with a Long Short-Term Memory Network (LSTM) to form the target network structure. Additionally, for datasets with limited samples, we incorporate a Self-Attention Mechanism (SAM) into the Model-Agnostic Meta-Learning (MAML). Ultimately, we build an MAML-SAM-CNN-LSTM network model to solve the interval of weight coefficients. An arithmetic validation of a modified IEEE 30-node system demonstrates that the MAML-SAM-CNN-LSTM network proposed in this paper can adeptly solve the feasible intervals of weight coefficients in the optimization model of the wind-thermal storage system. This is achieved under the constraints of the specified wind-thermal storage power system operation indexes. The evaluation indexes of the network model, including its accuracy, precision, recall, and F1 score, all exceed 98.72%, 98.57%, 98.30%, and 98.57%, respectively. This denotes a superior performance compared to the other three network models, offering an effective reference for optimizing decision-making and facilitating the enhanced realization of multi-objective, on-demand scheduling in the wind-thermal storage power system. [ABSTRACT FROM AUTHOR]
- Subjects :
- MACHINE learning
CONVOLUTIONAL neural networks
DEEP learning
ECONOMIC efficiency
Subjects
Details
- Language :
- English
- ISSN :
- 19961073
- Volume :
- 17
- Issue :
- 5
- Database :
- Complementary Index
- Journal :
- Energies (19961073)
- Publication Type :
- Academic Journal
- Accession number :
- 175986918
- Full Text :
- https://doi.org/10.3390/en17051082