Fault Detection and Diagnosis Using Parameter Estimation with Recursive Least Squares

The computational simplicity of Fault Detection and Diagnosis (FDD) schemes based on parameter estimation, as well as their suitability for integrated design with Fault Tolerant Control (FTC), make them attractive from the point of view of their implementation within real-time practical applications. This paper focuses on the advantages that these methods can offer as opposed to state estimation and parity equations based system diagnosis, while also highlighting the disadvantages of the parameter estimation approach in view of identifying the problems that could benefit from a simultaneous state and parameter estimation based scheme. In order to achieve this objective, a system diagnosis procedure is proposed, which makes use of the Recursive Least Squares (RLS) method with exponential weighting and constant forgetting factor in view of detecting the occurrence of a fault within the system, as well as providing the post-fault system model. An application example from the aircraft control field is also presented in order to illustrate the efficiency of the proposed approach.