A CNN-LSTM based carbon price forecasting model and its impact on green grid dispatch cost functions

Accurate carbon price forecasting is crucial for optimizing power system operations with high penetration of renewable energy sources such as photovoltaic and wind power. This paper proposes a novel CNN-LSTM hybrid model that effectively captures the spatiotemporal dependencies in carbon market data while addressing the specific needs of green grid dispatch optimization. Our model integrates convolutional layers for spatial feature extraction from multiple influencing factors and LSTM layers for temporal dependency modeling. Through comprehensive experiments using data from China's national carbon market, we demonstrate that our approach significantly outperforms traditional forecasting methods. More importantly, we develop an application framework that incorporates carbon price predictions into grid dispatch decisions, achieving substantial cost reductions by better balancing thermal generation costs against carbon emission costs. The proposed method provides power system operators with an effective tool for enhancing the economic efficiency of renewable energy integration while maintaining grid reliability.