A decomposition and coordination algorithm for reactive power optimization based on block coordinate descent

Abstract

Aiming at the inseparable augmented Lagrangian function of the decomposition-coordination reactive power optimization model, the parallel decomposition and coordination algorithm based on block coordinate descent is proposed. The algorithm can be used to distributed parallel reactive power optimization of the whole grid, which only needs to exchange the power and voltage information of boundary nodes between adjacent sub-areas to coordinate, so it can solve the problems of slow computing speed and transmission bottleneck in reactive power optimization of large-scale grid. Besides, the algorithm allows every control center to choose optimization algorithm independently. It realizes the combination of autonomous decentralized and coordinated control. The simulation shows the algorithm can significantly reduce the computing time of reactive power optimization of the whole grid, and compared with the decomposition and coordination algorithm based on auxiliary problem principle, its convergence rate is faster and its computational efficiency is higher.