A method for plausibility checks and data validation in power systems

摘要

A method for plausibility checks and the processing of bad data that is based on network search techniques and linear programming is proposed for power system control applications. Data validation is performed by sequential search through the network graph, during which bad marks are given to suspect measurements. Areas of the network with bad marks are further analyzed using linear programming. The procedure is efficient in detecting interacting and noninteracting bad data as well as inconsistent switch states. It also provides a set of voltages and angles. Static state estimation, if necessary, can then be performed on a reliable set of data, starting from the computed set of voltages and angles, to obtain a best fit of the measurements. The computing requirements of the method depend on the number and nature of abnormalities in the network. However, the overall CPU time requirement for validating data and data fitting is modest.