Abstract:

In multi-energy systems, the complexity of interconnections and the profound impact of system dynamics and uncertainties often lead to conflicts and trade-offs between components. These challenges make it difficult to maintain a dynamic balance between energy supply and demand. To address this challenge, a coupled optimization and control method for source network load storage multi-energy networks based on integrated neural networks is proposed. In addition, the proposed method considers the carbon emissions of multiple links in multi-energy systems, including source, grid, load, and storage, enabling comprehensive and accurate emission assessment. By combining a bidirectional weighted gated recurrent unit neural network with the Bagging ensemble algorithm, an ensemble neural network model can be constructed to capture contextual information about the time series data. By combining multiple weak learners, the proposed model significantly reduced the prediction error, achieving accurate carbon emission predictions in multi-energy networks. The regulation process focuses on three core optimization objectives: minimizing carbon emissions, and power generation costs, and maximizing the consumption of new energy. To address these challenges in complex multi-energy systems, the NSGA-II algorithm was applied to achieve comprehensive optimization regulation. The experimental results demonstrate that the proposed method accurately predicts carbon emissions. Furthermore, a regulation test demonstrated that the proposed method significantly improved the energy utilization efficiency, optimized the energy consumption, and enhanced the unit output stability. This research achievement is of great significance for sustainable development and efficient operation of multi-energy systems.

Key words: integrated neural network, multi energy system, source network load storage, optimizing regulation