Intelligent estimation strategies applied to a flight surface actuator

The Kalman filter (KF) has drastically changed and formed the field of state and parameter estimation theory and has impacted a number of applications: spacecraft, GPS, fault detection and diagnosis, stock market analysis, cell phones, autonomous vehicles, to name only a few. A statistically optimal solution for known linear systems is provided by the KF, in the presence of Gaussian white noise. However, the optimality of the KF affects numerical stability and robustness. A number of linear and nonlinear forms of the KF have been introduced to overcome numerical, stability, and nonlinearity issues. In recent years, intelligent or cognitive-based KFs have been proposed. Intelligent filters generally include adaptive gains and feedback for improved estimation accuracy and robustness. These types of filters are typically more robustness to modeling uncertainties and disturbances. This paper provides a comparison of two popular KF methods: fuzzy-based and machine learning-based. These strategies are applied on a flight surface system and the estimation results are compared and discussed. Future trends in intelligent estimation theory are also considered.