State estimation by orthogonalization and splitting

Abstract State estimation is a real-time problem and must be solved online. It is shown how state-estimation methods may be adapted for implementation on the new, mostly parallel and distributed computing systems currently being developed. The method of splitting is introduced and shown to be very efficient if the subnetworks are not too small. Another method of solving state estimation, namely orthogonalization, is also reviewed. It is then shown that computing errors may go undetected if the least-squares method is used. Finally, it is shown that use of orthogonalization would increase both accuracy and speed, even without splitting, but with splitting, the proposed method is more accurate, more stable, more reliable and requires less computing time than the methods used at present.