Research on State Estimation Methods for the New Characteristics of Active Distribution Network

Abstract

With the development of the distribution network, the penetration of distributed power sources in the distribution network increases gradually. The distribution network becomes increasingly active than the original passive system, this brings some problems to the real-time monitoring of the distribution network and makes it more difficult to estimate the state of the distribution network. First, there are a bunch of zero-injection nodes in the distribution network. The algorithm that uses traditional state estimation may be extremely sensitive to the error, and even unable to converge. Based on the traditional state estimation method, the path matrix is constructed to transform the state variables, and the simplified state variables are used to solve the problem caused by the high weightings of the zero injection buses. Second, the highly random output of DG brings a lot of uncertainty to the state estimation, and meanwhile, these uncertainties will affect the state estimation accuracy of its corresponding nodes because of lacking real-time measurement. Based on the first problem, the historical output data of distributed generators are analyzed offline by k-means clustering and divided into different output modes. Second State estimation is carried out by distinguishing different output modes. This method improves the accuracy of distributed power nodes and can be used in the Chengdu Tianfu International Airport network automation system engineering and other projects.