基于白箱树突网络的电压灵敏度实时计算方法OA北大核心CSTPCD

Since the traditional voltage sensitivity calculation requires the knowledge of network parameters,various data-driven voltage sensitivity fitting methods have been proposed recently.However,existing data-driven methods necessitate the repetition of all regression operations for new scenarios in order to achieve accurate values,thereby rendering them unsuitable for real-time calculations.This paper proposes a Dendritic network(DN)based voltage sensitivity calculation method,which establishes an end-to-end mapping of state parameters to node voltages for any scenario based on the white-box property that DN fitting is equivalent to Taylor expansion.By extracting and deriving the voltage explicit expressions from DN,the explicit expression of voltage sensitivity containing only an extremely small amount of computation is ultimately formed.In addition,in the explicit expression of DN based voltage sensitivity,the first-order component coefficient of each input keeps constant under any scenario,which can be directly used as an evaluation index to judge the correlation between node voltage and state parameters.The calculation results based on the three different scale systems verify the correctness and superiority of the methods proposed in this paper.