Strong Tracking Tobit Kalman Filter with Model Uncertainties
- 35 Downloads
The Tobit Kalman filter (TKF) is a good choice for applications of discrete time-varying systems with censored measurements. The traditional TKF is usually designed under the assumption of exactly knowing system state and measurement functions, which can not guarantee the good performance and is even unsuitable for system with model mismatch. To ensure the performance, this paper proposes a novel TKF algorithm in the presence of both censored measurements and model uncertainties. Our proposed algorithm, called strong tracking TKF (STTKF), adopts the orthogonal principle as an additional criterion to overcome model mismatch problem. By introducing fading factor into a priori error covariance, STTKF adaptively adjust gain matrix according to different model mismatch degree. Firstly the recursive fading factor formulation is deduced based on orthogonal principle. Then the designed STTKF process is given in a recursive manner. Finally the effectiveness of STTKF is verified by two examples of oscillator system and unmanned aerial vehicle (UAV) transmitter.
KeywordsCensored measurement fading factor Kalman filter model mismatch orthogonal principle strong tracking filter Tobit model
Unable to display preview. Download preview PDF.
- R. A. Martin and P. Wild, “Bivariate left–censored measurements in biomonitoring: a Bayesian model for the determination of biological limit values based on occupational exposure limits,” Annals of Work Exposures & Health, vol. 61, no. 5, pp. 1–5, May, 2017.Google Scholar
- J. Arthur, A. Attarian, F. Hamilton, and H. Tran, “Nonlinear Kalman filtering for censored observations,” vol. 316, pp. 155–166, September, 2017.Google Scholar
- B. Allik, C. Miller, M. J. Piovoso, and R. Zurakowski, “Estimation of saturated data using the Tobit Kalman filter,” Proc. of American Control Conference, pp. 4151–4156, June 2014.Google Scholar
- B. Allik, C. Miller, M. J Piovoso, and R. Zurakowski, “Nonlinear estimators for censored data: a comparison of the EKF, the UKF and the Tobit Kalman filter,” Proc. of American Control Conference, pp. 5146–5151, July, 2015.Google Scholar
- B. Allik, M. J. Piovoso, and R. Zurakowski, “Recursive estimation with quantized and censored measurements,” Proc. of American Control Conference, pp. 5130–5135, July, 2016.Google Scholar
- C. Miller, B. Allik, M. Piovoso, and R. Zurakowski, “Estimation of mobile vehicle range & position using the tobit Kalman filter,” Decision and Control, pp. 5001–5007, February, 2015.Google Scholar
- K. Loumponias, N. Vretos, P. Daras, and G. Tsaklidis, “Using Tobit Kalman filtering in order to improve the motion recorded by Microsoft Kinect,” Proc. of 8th International Workshop on Applied Probabilities, 2016.Google Scholar
- D. H. Zhou and P. M. Frank, “Strong tracking filtering of nonlinear time–varying stochastic systems with coloured noise: application to parameter estimation and empirical robustness analysis,” International Journal of Control, vol. 65, no. 2, pp. 295–307, September, 1996.MathSciNetCrossRefzbMATHGoogle Scholar
- X. L. Deng, W. Z. Guo, J. Y. Xie, and J. Liu, “Particle filter based on strong tracking filter,” International Conference on Machine Learning and Cybernetics, pp. 658–661, September, 2005.Google Scholar
- Z. L. Du, X. M. Li, Z. G. Zheng, and Q. Mao, “Fault prediction with combination of strong tracking square–root cubature Kalman filter and autoregressive model,” Control Theory & Applications, vol. 31, no. 8, pp. 1047–1052, August 2014.Google Scholar