A Lag-Free Adaptive Extraction Method for Wind Power Grid Integration Based on Bayes-CNN-LSTM and Variable Parameter Kalman Filter

To address issues such as power lag and large charging/discharging power in traditional wind power grid connection power extraction, this paper proposes a lag-free adaptive power extraction method for wind power grid connection based on Bayes-CNN-LSTM and variable-parameter Kalman filtering. This method utilizes a Bayes-optimized CNN-LSTM model to achieve ultra-short-term wind power prediction, providing advance information for lag-free extraction; combined with a variable-parameter Kalman filter algorithm to dynamically adjust filter parameters, it outputs grid-connected power without time delay. Compared with the moving average algorithm and fixed-parameter Kalman filter algorithm, the proposed method effectively eliminates the lag issues in traditional grid-connected power extraction, reduces the charging and discharging power of hybrid energy storage systems while meeting grid connection standards.