A Novel Distribution Network Reconfiguration Method with the Power Forecasting of DGs

As the development and utilization of renewable energy increase, more distributed generators (DGs) are employed in distribution systems. When a line fault occurs, the inherent intermittency and randomness of DGs’ power are a challenge for distribution network reconfiguration. In this paper, to reduce the influence of power uncertainty, a novel fault reconfiguration method is proposed with the power forecasting of DGs. First, a hybrid power forecasting method is proposed to predict the short-term power of DGs. The history power data of DGs are decomposed by the VMD method, and the decomposition signals are inputted into the TCN-LSTM method. The learning factor and penalty coefficient of TCN-LSTM are optimized by the sparrow search algorithm (SSA) to acquire better prediction results. Then, according to the prediction values of DGs, a fault reconfiguration method is constructed with maximum power restoration and minimum network losses, where the switch operation counts and island operation are taken into account. Simulation results verify the effectiveness of the proposed method. The VMD-SSA-TCN-LSTM method has a better performance than traditional forecasting methods in short-term prediction. The proposed fault reconfiguration method can restore the non-fault line load effectively and is suitable for multi-line fault conditions.