Dynamic System State Estimation with a Resilience to Observation Data Anomalies

In practical scenarios, abrupt alterations in system properties can lead to data distortion and random inaccuracies in observation results. These changes often transpire due to malfunctions or failures in individual nodes or subsystems. This paper emphasizes the development of a filter that produces state estimates for control objects capable of withstanding fault actions in the measurement subsystem. To this end, we adjust the observation channel model to accommodate varying accuracy levels, including sudden, abnormal errors. Our filter synthesis leverages Kalman optimal filtering theory methods within the Bayesian framework. This synthesis comprises filtering algorithms that generate the final state vector estimate as a linear combination of model-matched pseudo-Bayesian estimates, weighted by specific coefficients. We justify the existence of these estimates and present an accuracy assessment. Our study particularly emphasizes robust estimators, which are acquired by simplifying either the structure of the optimal estimator or the calculation process of the weighted coefficients. To address the inherent uncertainty of anomalous error probabilities in the observation channel, we suggest an adaptive estimation algorithm grounded in observation outcomes. Simulations were carried out to validate the functionality of the synthesized structures. For instance, we utilized a model depicting an aircraft’s movement during an approach, using the microwave landing system’s radio-electronic equipment as an example. We performed a comparative analysis of their accuracy and the associated computational complexity based on the study results.