Nodal Marginal Electricity Price Prediction for Renewable Energy Scenarios

Nodal marginal price prediction is crucial for resource allocation and market decision making. In this paper, a hybrid prediction framework integrating system simulation method and deep learning is proposed for the difficult problem of electricity price prediction in renewable energy scenarios. Firstly, a coordinated operation model of wind power and storage is constructed based on a six-node system, and the optimal power flow calculation is used to obtain the node marginal price data; secondly, this paper utilizes the advantages of convolutional neural network (CNN) and bi-directional gated recurrent unit (BiGRU) and introduces the attention mechanism to construct a CNN-BiGRU-Attention model. Experiments show that the model has an $\mathbf{R}^{2}$ of $\mathbf{9 5. 1 9 1 \%}$ in predicting node $\mathbf{4}$ price, which is significantly better than BiGRU, CNN-LSTM and other models. The study verifies the influence mechanism of wind power and transmission congestion on electricity price, which provides a reference for the electricity price prediction of power systems with high renewable penetration.