Data-driven Fault Location Method for Active Distribution Network with CNN-LSMT

With the increasing penetration of renewable energy sources (RES), the traditional distribution network is developing towards the active distribution network (ADN). Due to the uncertain and intermittent power output, the operation status of the system becomes more complex, which brings new challenges for fault location (FL) of ADN. In this paper, a data-driven FL method with convolutional neural networks-long short-term memory (CNN-LSTM) is proposed. Firstly, the optimal configuration model of phasor measurement unit (PMU) is established to minimize the configuration cost while guaranteeing the enough data for FL. Subsequently, a FL method based on CNN-LSTM is proposed, in which the data set containing power line features is constructed and employed to train the CNN-LSTM. With the trained CNN-LSTM, the location of line can be obtained accurately and automatically in a fast rate. Finally, numerical simulations are conducted on the modified IEEE 33-bus test system to verify the effectiveness and feasibility of the proposed method.