Improved wind power and storage system smoothing control strategy based on RE reinforcement learning and low pass filtering algorithms

Under the background of large-scale distributed wind power connected to the user side, an improved wind power and storage system smoothing control strategy based on Roth Erev reinforcement learning(RERL) and low pass filtering algorithms(LPFA) are presented in this paper. The randomness and fluctuations of wind power and the states of charge(SOC) of the battery energy storage system (BESS) are considered simultaneously. According to the smoothing effects of wind power, the proposed approach modifies the probability distributions of the filtering time constant under different SOC of BESS by sensing the SOC of the batteries and calculating the critical and the limit moments of the SOC, so as to change the proportion of the high and low frequency power components, and control the BESS to absorb the fluctuation components of the wind power effectively, and to avoid energy storage batteries overcharge and overdischarge. Simulation analyses demonstrate that this strategy can guarantee the smoothness of wind power outputs and can further reduce the fluctuations of the SOC at the same time, and can ensure the batteries in safe operation and long service life.