1. The previously described modeling approach is applied to the MACE test article. The MACE program calls for three generations of hardware: the Development Model (DM), the Engineering Model (EM), and the Flight Model (FM). The purpose of the DM was to evaluate preliminary designs of the major components. It consisted of the structural bus, a reaction wheel assembly, and a single two-axis gimbal (the other gimbal was replaced by a dummy mass). The purpose of the EM, which added a second two-axis gimbal,was to evaluate the final designs of the major components and was nearly identical to the FM, which is obviously the set of hardware that will fly on STS-67 in March of 1995.
2. Figures 7 and 8 graphically depict these two tests for the FM model. From Figure 7, it is seen that only five parameters (1,2,3,8,10)are significantly aligned with the initial cost. These parameters correspond to the stiffnesses of the passive struts, active strut, and the outer two suspension rods. Not coincidentally, these five parameters contribute the most to the flexibilityof the suspended structure, and therefore as a minimum, should be retained in the set of update parameters. From Figure 8, it is seen that the orthogonalitymatrix is predominantly diagonal, with a few exceptions. Most notably, parameter pairs 1-2,5-9, 3-5, 7-11, 5-6, and 3-9 (in order of decreasing alignment) have their sensitivities significantly aligned. The pairs 1-2,and 3-5might be expected to have similar sensitivities because they correspond to different parts of the same component; i.e., passive or active strut. These pairs might be suitable for the creation of two super-parameters. The other pairs do not correspond to similar components, e.g., an active strut parameter correlated to a suspension parameter. Fortunately, most of the other aligned parameters do not affect the initial cost very much and would be discarded (test 1).
3. This work was supported by the NASA IN-STEP Program and NASA LaRC CSI Office with Mr. Gregory Stover and Dr. Jerry Newsom as contract monitors, under the contract NAS1-18690. Special thanks is given to Dr. Jonathan How, Dr. Ketao Liu, Mark Campbell, and Simon Grocott, the members of the MACE system identification and robust control teams for providing the measurement models used in the model updates and insights in model errors gained from finite element based control design.