Parallel computing involving topology estimation and virtual measurement for distribution networks

This paper targets three barriers today´s 10kV distribution networks in China are facing: 1) existing real-time measurements are subject to errors; 2) measurements are insufficient; and 3) real-time calculations of large-scale distribution networks are to be realized. Covering a cross-disciplinary topic, this paper proposed the following techniques and methodologies: i) a confidence-level-based island elimination algorithm and a depth-conflict-rule-based loop identification algorithm as the components of a topology estimation (TE) module; ii) a two-stage virtual measurement (VM) approach; and iii) a parallel process of TE and VM on both the inter-task level and the intra-task level. By fully exploiting multi-level parallel potential, the considerable efficiency improvement of the parallel computing method is demonstrated in field deployment, as compared to conventional series computing.