Robust Air Data Reconstruction: On the Use of Robust Cost Functions for Flight Path Reconstruction Applications

摘要

The flight path reconstruction (FPR) problem is formulated as a robust estimation problem to address the practical limitations of modern air data measurement techniques. The FPR problem is first formulated as maximum a posteriori (MAP) estimation problem and the philosophy of robust cost functions (RCFs) is adopted for the development of optimization weight deflation schemes. Using the iteratively reweighted least squares algorithm for solving the formulated robust estimation problem, the popular redescending Geman-McClure (GM), Cauchy, and threshold RCFs are applied to a simulated dataset and compared. Experimental results demonstrate the effectiveness of the GM RCF at mitigating a wide spectrum of deterministic air data measurement errors indicative of those encountered in practice. Further estimation accuracy is demonstrated when trajectory estimates obtained using the GM RCF are used as the initial trajectory estimates for robust estimation using the threshold RCF. Overall, the investigation is successful in establishing the feasibility of the use of RCFs and FPR to mitigate practical air data instrumentation limitations routinely encountered in aircraft system identification and control activities.