Development of a Static State Estimator Part II: Bad Data Replacement and Generation of Pseudomeasurements

This paper forms Part II of a two-part series which describes the construction and the overall software structure of a static state estimator. This part outlines the methods used for bad data replacement and the generation of pseudomeasurements. An iterative method is developed for the replacement of bad data which is compared to a previously published technique and shown to be superior in practical situations. The previously published technique required an accurate measure of the standard deviation of the error covariance matrix, which is generally not known in practice. The method proposed here gets over this problem. It relies on the closeness of the initial estimate of the state variables to the true solution as a result of the excellent bad data suppression property of the variable quadraticflat (developed in Part I of the series). Testing of a system with up to forty bad data points has been successfully carried out. The second half of the paper deals with the generation of pseudomeasurements to replace lost measurements. Two algorithms based on load forecasting techniques have been tested. The first one assumes that between successive sets of measurements to be processed, the incremental load demand is negligible. The second one includes the variation of the incremental load demand.