MULTIVARIABLE FLEX MODEL REDUCTION

摘要

To analyze large dimensional flex models such as the International Space Station (ISS) requires substantial computational effort. The popular Balanced Truncation (BT) method has been frequently used to reduce computational load. However, the BT method often requires extensive computational effort and sometimes lacks numerical robustness. In this paper, an innovative model reduction technique, which converts a flex model into a "tri-diagonal modal form" and then truncates the resulting model based on a multivariable Modal Gain Factor approach, is presented. The proposed technique takes less computational effort and is numerically stable for ISS flex models. Comparisons with the BT method will be presented for various ISS assembly stage flex models.