A Recursive Least Squares-based (RLS-based) fault detection algorithm was developed, namely Parallel Two-Stage RLS-based Fault Detection (PTSRLS-FD), to safely operate against false-positive cases by improving the detection for a specific fault type, namely loss of effectiveness (LOE). The al-gorithm was compared to a simpler, time domain residual model error-based detection algorithm (TDME) on several simulation results through actuator subsystem-level simulations, system-level Monte-Carlo (MC) analyses, Hardware-in-the-Loop (HIL) tests and flight tests all of which include both healthy and faulty actuator cases. The PTSRLS-FD have higher true-positive rates and lower false-positive rates in comparison with the TDME in the numerical experiments, and provided a safer residual margin in healthy actuator cases, and finally, the algorithm turned out to be fast enough to run online with at most 1.4 ms loop time.
Development and Application of a Parallel Two-Stage RLS-based Fault Detection for Nonlinear Thrust Vector Control Actuator on a Sounding Rocket
2023-06-07
2019665 byte
Conference paper
Electronic Resource
English
Optimal Thrust Programming for Two-Stage Sounding Rocket
AIAA | 2019
|Solid rocket thrust vector control
NTRS | 1974
Two-stage sounding rocket problem
Engineering Index Backfile | 1967
|