| The Gravity Probe B Relativity Mission (GP-B) is a space experiment intended to check previously untested aspects of Einstein's General Theory of Relativity. It will measure, over the course of a one-year experiment, the directional change of an Earth-orbiting gyroscope's spin axis relative to inertial space as defined by the fixed stars. According to general relativity, the spin axis undergoes two orthogonal precessions of magnitude 6.6 and 0.04 arcseconds per year for a 650-km polar orbit. These tiny drift angles will be measured to sub-milliarcsecond accuracy, a requirement which puts extreme constraints on the gyroscope and the readout system. Given these requirements, it is crucial that the gyroscope, readout system and data reduction scheme be verified to the fullest extent possible on the ground before the mission is flown.; This dissertation is a summary of recent work on the GP-B Niobium Bird project, which is intended to provide precisely this end-to-end verification. Analogous to the "Iron Birds" of the aircraft industry, the Niobium Bird (the name refers to the extensive use of niobium in the experiment) is a hardware-in-the-loop simulation, integrating computer simulation of the science signal with prototypical readout hardware. Contributions have been made in several areas of the Niobium Bird. Improvements in the signal simulation include modeling of extraneous signals due to temperature variations and magnetic flux trapped on the surface of the gyroscopes. Upgrades to the experiment include seamless high-speed data injection and acquisition, monitoring of environmental variables such as temperature and vibrations, and characterization of system properties, including linearity and stability. An optimal, recursive nonlinear least-squares estimation algorithm has been developed for performing the data reduction. This algorithm has been generalized for application to other nonlinear problems. Examples are given in which significant improvement was found over other common nonlinear estimation algorithms, such as the extended Kalman filter and the iterated extended Kalman filter. The new algorithm has been implemented in simulation to demonstrate 0.2 milliarcsecond accuracy of the relativity estimates after one year. One-day experimental runs have helped confirm these simulations. |
