面向充电场站负荷波动的配电网主动运维方法OA

Ultra-fast charging stations and other high-power,highly volatile loads in distribution networks can easily impose risks such as power shocks and severe overloading on distribution transformers,thereby weakening the efficiency of distribution network operation and maintenance(O&M).To address this issue,a prediction-driven active O&M method for distribution networks based on TBG-Net(transformer bi-dPGRU and lightGBM network)was proposed.Firstly,for the high-fluctuation load scenario of charging stations,a multi-expert collaborative forecasting structure composed of trend forecasting,differential mutation modeling and statistical compensation was constructed.Combined with a ramp-aware gating mechanism,the proposed structure enables adaptive fusion between steady operating conditions and power mutation intervals.Secondly,the load forecasting results were embedded into the distribution O&M analysis process.By dynamically comparing the forecasted load with transformer capacity constraints,overload risk warning and active O&M support were realized at the station or distribution-transformer level,and it was validated based on measured data from a charging station in a city in Hubei Province.The results show that,compared with advanced load forecasting methods,the proposed method improves the root mean square error and mean absolute error by approximately 45％,while the coefficient of determination increases by about 0.40.The proposed method also significantly outperforms the benchmark models in different load ramping intensity scenarios,which effectively support peak-shaving control and improve the operational safety margin of distribution equipment.