Distribution Network Topology Reconstruction Method Based on Lasso and Its Supplementary Criterions

In order to solve the problem of topology reconstruction in distribution network, a new data driven algorithm is proposed, which uses only the timing voltage to reconstruct the un-loopy and loopy distribution network topology without the prior knowledge. Firstly, the topology reconstruction problem is transformed into a convex optimization problem, and the Lasso regularization method is utilized to obtain a sparse correlation coefficient matrix (CCM), which represents the connectivity of the topology. Secondly, the “And” rule is employed to reduce the redundancy of CCM. And then the criterion of the voltage correlation analysis model is adopted as a supplemental criterion to reduce the error rate of CCM. Finally, the topology reconstruction of the distribution network is realized based on the accurate CCM. Simulation results show that the algorithm has high accuracy, universality and low computational complexity.