A Comparison between Kalman Filtering Approaches in Aircraft Flight Signal Estimation

At present, the requirements for the accuracy of aircraft on-board measurement systems are constantly increasing, while sensors contain various errors in signal measurement, primarily random. Noisy signals from onboard measurements can be smoothed or filtered out in a variety of ways. One of the most popular approaches is Kalman filtering, the effectiveness of which has been proven by many studies. This paper presents a comparative analysis of the extended Kalman filter (EKF) and unscented Kalman filter (UKF), used to estimate the pitch angle of an aircraft using bench modeling data. During the simulation, the normal measurement noises are also generated. According to the results obtained in this paper, it can be noted that UKF performs better when a priori knowledge about the process noise is certain. However, the efficiency of UKF in estimating the signal deteriorates when a priori knowledge about the process becomes uncertain while the performance of EKF remains stable. This is due to the fact that UKF uses more sophisticated assumptions and therefore is more sensitive to these assumptions violation. The obtained results also show that various variants of Kalman filtering remain relevant in comparison with the smoothing methods that have spread in recent years, based on the ideas of optimal control and evolutionary algorithms for numerical optimization.