Deep-Learning-Aided Voltage-Stability-Enhancing Stochastic Distribution Network Reconfiguration

Power distribution networks are approaching their voltage stability boundaries due to the severe violations and inadequate reactive power reserves caused by increasing renewable generations dynamic loads. In broad endeavor resolve this concern, we focus on enhancing through stochastic network reconfiguration (SDNR), which optimizes (radial) topology of a under uncertain We propose deep learning method solve computationally challenging problem. Specifically, build convolutional neural model predict relevant index from SDNR decisions. Then integrate prediction into successive branch reduction algorithms reconfigure radial with optimized performance in terms loss enhancement. Numerical results two IEEE models verify significance computational efficiency proposed method.