Online Frequency Prediction of Renewable Energy Power System Based on DPGMM-LSTM

The rising proportion of renewable energy sources installed and the new grid structure of AC/DC hybrid increase the risk of frequency instability of renewable energy power system after disturbance. It is urgent to carry out online prediction of frequency response curve after disturbance. The traditional model driven frequency prediction method has the contradiction between calculation speed and accuracy, which is difficult to meet the demand of online frequency prediction under complex and changeable operation mode and disturbance conditions. Therefore, in this paper, deep learning is introduced into the study of frequency online prediction after disturbance, and an online prediction method of frequency response curve based on DPGMM-LSTM is proposed. Firstly, DPGMM is used to cluster the control mode of renewable energy power station and add it to the model training feature set, and then LSTM asynchronous sequence to sequence mode is used as the prediction framework to realize the online rolling prediction of frequency curve of renewable energy power system. The performance of the proposed method is tested based on IEEE 10-machine 39-bus system, and the simulation results verify the advantages of the proposed method.