A Wind Power Prediction Framework for Distributed Power Grids

To reduce carbon emissions,clean energy is being integrated into the power system.Wind power is connected to the grid in a distributed form,but its high variability poses a challenge to grid stability.This article combines wind turbine monitoring data with numerical weather prediction(NWP)data to create a suitable wind power prediction framework for distributed grids.First,high-precision NWP of the turbine range is achieved using weather research and forecasting models(WRF),and Kriging interpolation locates predicted meteorological data at the turbine site.Then,a preliminary predicted power series is obtained based on the fan’s wind speed-power conversion curve,and historical power is reconstructed using variational mode decomposition(VMD)filtering to form input variables in chronological order.Finally,input variables of a single turbine enter the temporal convolutional network(TCN)to complete initial feature extraction,and then integrate the outputs of all TCN layers using Long Short Term Memory Networks(LSTM)to obtain power prediction sequences for all turbine positions.The proposed method was tested on a wind farm connected to a distributed power grid,and the results showed it to be superior to existing typical methods.

Automatic Generation Control in a Distributed Power Grid Based on Multi-step Reinforcement Learning

The increasing use of renewable energy in the power system results in strong stochastic disturbances and degrades the control performance of the distributed power grids.In this paper,a novel multi-agent collaborative reinforcement learning algorithm is proposed with automatic optimization,namely,Dyna-DQL,to quickly achieve an optimal coordination solution for the multi-area distributed power grids.The proposed Dyna framework is combined with double Q-learning to collect and store the environmental samples.This can iteratively update the agents through buffer replay and real-time data.Thus the environmental data can be fully used to enhance the learning speed of the agents.This mitigates the negative impact of heavy stochastic disturbances caused by the integration of renewable energy on the control performance.Simulations are conducted on two different models to validate the effectiveness of the proposed algorithm.The results demonstrate that the proposed Dyna-DQL algorithm exhibits superior stability and robustness compared to other reinforcement learning algorithms.