Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target

Chuan Ma; Jianping Yuan; Dejia Che

doi:10.1007/978-3-030-27541-9_16

Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target

Chuan Ma, Jianping Yuan, Dejia Che

School of Astronautics

Northwestern Polytechnical University Xian

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

A novel characteristic frequency input network (CFIN) is investigated for inertia parameters identification of the tumbling space target from the quaternion measurements based on the back propagation neural network. The main innovation of the CFIN is it set the 15-dimensional characteristic frequency vector as the input of the neural network, which is extracted from the constant parameters of the target’s attitude quaternion. The utilization of the characteristic frequency not only reduces the required number of nodes to less than 100, but also improves the learning rate of the neural network. The CFIN is trained using 10000 samples and tested using another 2000 data. It can work well in real-time with very little computational burden and storage space once successfully trained. Moreover, effectiveness analysis illustrates that the CFIN can provide more precise estimation of the inertia parameters than the conventional estimation method, like extend Kalman filter and unscented Kalman filter in most case.

Original language	English
Title of host publication	Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings
Editors	Haibin Yu, Jinguo Liu, Lianqing Liu, Yuwang Liu, Zhaojie Ju, Dalin Zhou
Publisher	Springer Verlag
Pages	174-187
Number of pages	14
ISBN (Print)	9783030275402
DOIs	https://doi.org/10.1007/978-3-030-27541-9_16
State	Published - 2019
Event	12th International Conference on Intelligent Robotics and Applications, ICIRA 2019 - Shenyang, China Duration: 8 Aug 2019 → 11 Aug 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11744 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	12th International Conference on Intelligent Robotics and Applications, ICIRA 2019
Country/Territory	China
City	Shenyang
Period	8/08/19 → 11/08/19

Keywords

BP neural network
Characteristic frequency
Inertia identification
Tumbling space target

Access to Document

10.1007/978-3-030-27541-9_16

Cite this

Ma, C., Yuan, J., & Che, D. (2019). Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target. In H. Yu, J. Liu, L. Liu, Y. Liu, Z. Ju, & D. Zhou (Eds.), Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings (pp. 174-187). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11744 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-27541-9_16

Ma, Chuan ; Yuan, Jianping ; Che, Dejia. / Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target. Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings. editor / Haibin Yu ; Jinguo Liu ; Lianqing Liu ; Yuwang Liu ; Zhaojie Ju ; Dalin Zhou. Springer Verlag, 2019. pp. 174-187 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{f42e838f84204479b33a5c79a194bef3,

title = "Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target",

abstract = "A novel characteristic frequency input network (CFIN) is investigated for inertia parameters identification of the tumbling space target from the quaternion measurements based on the back propagation neural network. The main innovation of the CFIN is it set the 15-dimensional characteristic frequency vector as the input of the neural network, which is extracted from the constant parameters of the target{\textquoteright}s attitude quaternion. The utilization of the characteristic frequency not only reduces the required number of nodes to less than 100, but also improves the learning rate of the neural network. The CFIN is trained using 10000 samples and tested using another 2000 data. It can work well in real-time with very little computational burden and storage space once successfully trained. Moreover, effectiveness analysis illustrates that the CFIN can provide more precise estimation of the inertia parameters than the conventional estimation method, like extend Kalman filter and unscented Kalman filter in most case.",

keywords = "BP neural network, Characteristic frequency, Inertia identification, Tumbling space target",

author = "Chuan Ma and Jianping Yuan and Dejia Che",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 12th International Conference on Intelligent Robotics and Applications, ICIRA 2019 ; Conference date: 08-08-2019 Through 11-08-2019",

year = "2019",

doi = "10.1007/978-3-030-27541-9_16",

language = "英语",

isbn = "9783030275402",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "174--187",

editor = "Haibin Yu and Jinguo Liu and Lianqing Liu and Yuwang Liu and Zhaojie Ju and Dalin Zhou",

booktitle = "Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings",

}

Ma, C, Yuan, J & Che, D 2019, Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target. in H Yu, J Liu, L Liu, Y Liu, Z Ju & D Zhou (eds), Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11744 LNAI, Springer Verlag, pp. 174-187, 12th International Conference on Intelligent Robotics and Applications, ICIRA 2019, Shenyang, China, 8/08/19. https://doi.org/10.1007/978-3-030-27541-9_16

Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target. / Ma, Chuan; Yuan, Jianping; Che, Dejia.
Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings. ed. / Haibin Yu; Jinguo Liu; Lianqing Liu; Yuwang Liu; Zhaojie Ju; Dalin Zhou. Springer Verlag, 2019. p. 174-187 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11744 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target

AU - Ma, Chuan

AU - Yuan, Jianping

AU - Che, Dejia

PY - 2019

Y1 - 2019

N2 - A novel characteristic frequency input network (CFIN) is investigated for inertia parameters identification of the tumbling space target from the quaternion measurements based on the back propagation neural network. The main innovation of the CFIN is it set the 15-dimensional characteristic frequency vector as the input of the neural network, which is extracted from the constant parameters of the target’s attitude quaternion. The utilization of the characteristic frequency not only reduces the required number of nodes to less than 100, but also improves the learning rate of the neural network. The CFIN is trained using 10000 samples and tested using another 2000 data. It can work well in real-time with very little computational burden and storage space once successfully trained. Moreover, effectiveness analysis illustrates that the CFIN can provide more precise estimation of the inertia parameters than the conventional estimation method, like extend Kalman filter and unscented Kalman filter in most case.

AB - A novel characteristic frequency input network (CFIN) is investigated for inertia parameters identification of the tumbling space target from the quaternion measurements based on the back propagation neural network. The main innovation of the CFIN is it set the 15-dimensional characteristic frequency vector as the input of the neural network, which is extracted from the constant parameters of the target’s attitude quaternion. The utilization of the characteristic frequency not only reduces the required number of nodes to less than 100, but also improves the learning rate of the neural network. The CFIN is trained using 10000 samples and tested using another 2000 data. It can work well in real-time with very little computational burden and storage space once successfully trained. Moreover, effectiveness analysis illustrates that the CFIN can provide more precise estimation of the inertia parameters than the conventional estimation method, like extend Kalman filter and unscented Kalman filter in most case.

KW - BP neural network

KW - Characteristic frequency

KW - Inertia identification

KW - Tumbling space target

UR - http://www.scopus.com/inward/record.url?scp=85070816314&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-27541-9_16

DO - 10.1007/978-3-030-27541-9_16

M3 - 会议稿件

AN - SCOPUS:85070816314

SN - 9783030275402

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 174

EP - 187

BT - Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings

A2 - Yu, Haibin

A2 - Liu, Jinguo

A2 - Liu, Lianqing

A2 - Liu, Yuwang

A2 - Ju, Zhaojie

A2 - Zhou, Dalin

PB - Springer Verlag

T2 - 12th International Conference on Intelligent Robotics and Applications, ICIRA 2019

Y2 - 8 August 2019 through 11 August 2019

ER -

Ma C, Yuan J, Che D. Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target. In Yu H, Liu J, Liu L, Liu Y, Ju Z, Zhou D, editors, Intelligent Robotics and Applications - 12th International Conference, ICIRA 2019, Proceedings. Springer Verlag. 2019. p. 174-187. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-27541-9_16

Characteristic Frequency Input Neural Network for Inertia Identification of Tumbling Space Target

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this