Power system state estimation derives a real-time network model by extracting information from a redundant data set consisting of telemetered, predicted and static data items. This paper describes a generalized, fully developed, estimation approach that fundamentally improves the information extraction process. Its main contribution is the successful inclusion of topology and parameters in the estimation and bad data analysis processes. This is valuable both in the initial commissioning of a state estimator, and in its routine real-time and study mode application. The approach involves a variety of novel concepts and methods. It is usable in weighted least squares (WLS) and other estimation approaches.