Deep-Learning Based Reactive Voltage Control of Regional Power Grids Integrated with Renewable Energy Resources

With the increasing penetration rate of distributed renewable energy and more complexity of grid, it is essential to keep a secure voltage profile for the safety, stability and economy of the power system. A reactive voltage control approach in regional power grid is proposed in this paper, which can provide online decisions without the requirement of the model and parameters. The proposed approach enhances the ability to rapidly restore the voltage back to normal after severe system disturbances. Furthermore, to regulate voltage profiles and reduce shunt operations, an LSTM based reactive voltage control model is also proposed by considering transformers and capacitors. A case of a regional power grid in Jiangsu is studied to verify the proposed scheme. The experimental results show the capability and efficiency of the proposed scheme. In contrast to traditional methods, our approach improves the accuracy of the control strategy by prediction model meanwhile limiting the action times of control devices.