Validation of Linearized Flight Models Using Automated System-Identification

Optimization based flight control design tools depend on automatic linearization tools, such as Simulink®’s LINMOD, to extract linear models. In order to ensure the usefulness and correctness of the generated linear model, this linearization must be accurate. So a method of independently verifying the linearized model is needed. This thesis covers the automation of a system identification tool, CIFER®, for use as a verification tool integrated with CONDUIT®, an optimization based design tool. Several test cases are built up to demonstrate the accuracy of the verification tool with respect to analytical results and matches with LINMOD. Several common nonlinearities are tested, comparing the results from CIFER and LINMOD, as well as analytical results where possible. The CIFER results show excellent agreement with analytical results. LINMOD treated most nonlinearity as a unit gain, but some nonlinearities linearized to a zero, causing the linearized model to omit that path. Although these effects are documented within Simulink, their presence may be missed by a user. The verification tool is successful in identifying these problems when present. A section is dedicated to the diagnosis of linearization errors, suggesting solutions where possible.

Name thesis:degree_name

MS in Aerospace Engineering

Discipline thesis:degree_discipline

Aerospace Engineering

Year dc:date.available

2009

Author dc:creator

• Rothman, Keith Eric

Contributors dc:contributor

• Daniel J. Biezad

• system-identifcation
• linearization
• LINMOD
• CIFER
• CONDUIT
• Navigation, Guidance, Control and Dynamics

Identifier

10.15368/theses.2009.52

OAI identifier oai:identifier

oai:digitalcommons.calpoly.edu:theses-1089