Please use this identifier to cite or link to this item: http://localhost/handle/Hannan/227906
Title: Maximum Correntropy Kalman Filter With State Constraints
Authors: Xi Liu;Badong Chen;Haiquan Zhao;Jing Qin;Jiuwen Cao
Year: 2017
Publisher: IEEE
Abstract: For linear systems, the original Kalman filter under the minimum mean square error (MMSE) criterion is an optimal filter under a Gaussian assumption. However, when the signals follow non-Gaussian distributions, the performance of this filter deteriorates significantly. An efficient way to solve this problem is to use the maximum correntropy criterion (MCC) instead of the MMSE criterion to develop the filters. In a recent work, the maximum correntropy Kalman filter (MCKF) was derived. The MCKF performs very well in filtering heavy-tailed non-Gaussian noise, and its performance can be further improved when some prior information about the system is available (e.g., the system states satisfy some equality constraints). In this paper, to address the problem of state estimation under equality constraints, we develop a new filter, called the MCKF with state constraints, which combines the advantages of the MCC and constrained estimation technology. The performance of the new algorithm is confirmed with two illustrative examples.
Description: 
URI: http://localhost/handle/Hannan/227906
volume: 5
More Information: 25846,
25853
Appears in Collections:2017

Files in This Item:
File SizeFormat 
8094856.pdf3.52 MBAdobe PDF
Title: Maximum Correntropy Kalman Filter With State Constraints
Authors: Xi Liu;Badong Chen;Haiquan Zhao;Jing Qin;Jiuwen Cao
Year: 2017
Publisher: IEEE
Abstract: For linear systems, the original Kalman filter under the minimum mean square error (MMSE) criterion is an optimal filter under a Gaussian assumption. However, when the signals follow non-Gaussian distributions, the performance of this filter deteriorates significantly. An efficient way to solve this problem is to use the maximum correntropy criterion (MCC) instead of the MMSE criterion to develop the filters. In a recent work, the maximum correntropy Kalman filter (MCKF) was derived. The MCKF performs very well in filtering heavy-tailed non-Gaussian noise, and its performance can be further improved when some prior information about the system is available (e.g., the system states satisfy some equality constraints). In this paper, to address the problem of state estimation under equality constraints, we develop a new filter, called the MCKF with state constraints, which combines the advantages of the MCC and constrained estimation technology. The performance of the new algorithm is confirmed with two illustrative examples.
Description: 
URI: http://localhost/handle/Hannan/227906
volume: 5
More Information: 25846,
25853
Appears in Collections:2017

Files in This Item:
File SizeFormat 
8094856.pdf3.52 MBAdobe PDF
Title: Maximum Correntropy Kalman Filter With State Constraints
Authors: Xi Liu;Badong Chen;Haiquan Zhao;Jing Qin;Jiuwen Cao
Year: 2017
Publisher: IEEE
Abstract: For linear systems, the original Kalman filter under the minimum mean square error (MMSE) criterion is an optimal filter under a Gaussian assumption. However, when the signals follow non-Gaussian distributions, the performance of this filter deteriorates significantly. An efficient way to solve this problem is to use the maximum correntropy criterion (MCC) instead of the MMSE criterion to develop the filters. In a recent work, the maximum correntropy Kalman filter (MCKF) was derived. The MCKF performs very well in filtering heavy-tailed non-Gaussian noise, and its performance can be further improved when some prior information about the system is available (e.g., the system states satisfy some equality constraints). In this paper, to address the problem of state estimation under equality constraints, we develop a new filter, called the MCKF with state constraints, which combines the advantages of the MCC and constrained estimation technology. The performance of the new algorithm is confirmed with two illustrative examples.
Description: 
URI: http://localhost/handle/Hannan/227906
volume: 5
More Information: 25846,
25853
Appears in Collections:2017

Files in This Item:
File SizeFormat 
8094856.pdf3.52 MBAdobe PDF