Training data to create neural networks for power system networks

Different loading profiles for active and reactive power demands at load buses are mapped to their optimal solutions (Active power generation and voltage magnitudes at generator buses), for IEEE 14 and 30-bus standard grids solved by Matpower interior point solver (MIPS). These data can be used to train neural networks to find the non-linear relationships between load demands and generation dispatches. This is useful in finding efficient algorithm that can speed up the computation time to solve ACOPF (AC Optimal Power Flow) problem.